diff --git "a/checkpoint-40000/trainer_state.json" "b/checkpoint-40000/trainer_state.json"
--- "a/checkpoint-40000/trainer_state.json"
+++ "b/checkpoint-40000/trainer_state.json"
@@ -1,7 +1,7 @@
 {
-  "best_metric": 3.4361512660980225,
-  "best_model_checkpoint": "/scratch/cl5625/exceptions/models/100M_8397/checkpoint-40000",
-  "epoch": 4.305241631686578,
+  "best_metric": 3.4450483322143555,
+  "best_model_checkpoint": "/scratch/cl5625/exceptions/models/100M_high_100_8397/checkpoint-40000",
+  "epoch": 4.3126684636118595,
   "eval_steps": 1000,
   "global_step": 40000,
   "is_hyper_param_search": false,
@@ -9,5968 +9,5968 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.005381552039608223,
-      "grad_norm": 1.4348148107528687,
+      "epoch": 0.005390835579514825,
+      "grad_norm": 1.4838696718215942,
       "learning_rate": 0.0003,
-      "loss": 8.6345,
+      "loss": 8.6352,
       "step": 50
     },
     {
-      "epoch": 0.010763104079216447,
-      "grad_norm": 2.487840175628662,
+      "epoch": 0.01078167115902965,
+      "grad_norm": 3.1202118396759033,
       "learning_rate": 0.0006,
-      "loss": 6.8736,
+      "loss": 6.8964,
       "step": 100
     },
     {
-      "epoch": 0.01614465611882467,
-      "grad_norm": 1.9002207517623901,
-      "learning_rate": 0.0005996767589699385,
-      "loss": 6.456,
+      "epoch": 0.016172506738544475,
+      "grad_norm": 1.9729210138320923,
+      "learning_rate": 0.0005996762007555315,
+      "loss": 6.474,
       "step": 150
     },
     {
-      "epoch": 0.021526208158432893,
-      "grad_norm": 1.5307645797729492,
-      "learning_rate": 0.0005993535179398771,
-      "loss": 6.2263,
+      "epoch": 0.0215633423180593,
+      "grad_norm": 1.4909285306930542,
+      "learning_rate": 0.000599352401511063,
+      "loss": 6.2347,
       "step": 200
     },
     {
-      "epoch": 0.026907760198041114,
-      "grad_norm": 2.2243926525115967,
-      "learning_rate": 0.0005990302769098158,
-      "loss": 6.085,
+      "epoch": 0.026954177897574125,
+      "grad_norm": 1.0986427068710327,
+      "learning_rate": 0.0005990286022665946,
+      "loss": 6.0861,
       "step": 250
     },
     {
-      "epoch": 0.03228931223764934,
-      "grad_norm": 1.4903671741485596,
-      "learning_rate": 0.0005987070358797543,
-      "loss": 5.9931,
+      "epoch": 0.03234501347708895,
+      "grad_norm": 1.4141861200332642,
+      "learning_rate": 0.0005987048030221263,
+      "loss": 5.9872,
       "step": 300
     },
     {
-      "epoch": 0.03767086427725756,
-      "grad_norm": 1.9050363302230835,
-      "learning_rate": 0.0005983837948496929,
-      "loss": 5.8611,
+      "epoch": 0.03773584905660377,
+      "grad_norm": 1.771342158317566,
+      "learning_rate": 0.0005983810037776578,
+      "loss": 5.8799,
       "step": 350
     },
     {
-      "epoch": 0.04305241631686579,
-      "grad_norm": 1.8527716398239136,
-      "learning_rate": 0.0005980605538196314,
-      "loss": 5.8097,
+      "epoch": 0.0431266846361186,
+      "grad_norm": 1.0568146705627441,
+      "learning_rate": 0.0005980572045331894,
+      "loss": 5.8406,
       "step": 400
     },
     {
-      "epoch": 0.048433968356474004,
-      "grad_norm": 1.9086118936538696,
-      "learning_rate": 0.0005977373127895701,
-      "loss": 5.7233,
+      "epoch": 0.04851752021563342,
+      "grad_norm": 1.4827862977981567,
+      "learning_rate": 0.0005977334052887209,
+      "loss": 5.7664,
       "step": 450
     },
     {
-      "epoch": 0.05381552039608223,
-      "grad_norm": 1.4407856464385986,
-      "learning_rate": 0.0005974140717595086,
-      "loss": 5.6468,
+      "epoch": 0.05390835579514825,
+      "grad_norm": 1.1048723459243774,
+      "learning_rate": 0.0005974096060442526,
+      "loss": 5.6768,
       "step": 500
     },
     {
-      "epoch": 0.05919707243569045,
-      "grad_norm": 1.4049276113510132,
-      "learning_rate": 0.0005970908307294472,
-      "loss": 5.5679,
+      "epoch": 0.05929919137466307,
+      "grad_norm": 1.177209734916687,
+      "learning_rate": 0.0005970858067997841,
+      "loss": 5.591,
       "step": 550
     },
     {
-      "epoch": 0.06457862447529868,
-      "grad_norm": 1.3480347394943237,
-      "learning_rate": 0.0005967675896993858,
-      "loss": 5.4894,
+      "epoch": 0.0646900269541779,
+      "grad_norm": 1.0635206699371338,
+      "learning_rate": 0.0005967620075553157,
+      "loss": 5.5163,
       "step": 600
     },
     {
-      "epoch": 0.0699601765149069,
-      "grad_norm": 2.005707263946533,
-      "learning_rate": 0.0005964443486693243,
-      "loss": 5.4059,
+      "epoch": 0.07008086253369272,
+      "grad_norm": 1.1746788024902344,
+      "learning_rate": 0.0005964382083108472,
+      "loss": 5.4473,
       "step": 650
     },
     {
-      "epoch": 0.07534172855451512,
-      "grad_norm": 1.2022747993469238,
-      "learning_rate": 0.000596121107639263,
-      "loss": 5.3226,
+      "epoch": 0.07547169811320754,
+      "grad_norm": 1.4562958478927612,
+      "learning_rate": 0.0005961144090663788,
+      "loss": 5.3979,
       "step": 700
     },
     {
-      "epoch": 0.08072328059412334,
-      "grad_norm": 1.1703269481658936,
-      "learning_rate": 0.0005957978666092015,
-      "loss": 5.2998,
+      "epoch": 0.08086253369272237,
+      "grad_norm": 1.1735752820968628,
+      "learning_rate": 0.0005957906098219104,
+      "loss": 5.3239,
       "step": 750
     },
     {
-      "epoch": 0.08610483263373157,
-      "grad_norm": 1.2556697130203247,
-      "learning_rate": 0.0005954746255791401,
-      "loss": 5.2564,
+      "epoch": 0.0862533692722372,
+      "grad_norm": 1.1097311973571777,
+      "learning_rate": 0.0005954668105774419,
+      "loss": 5.2782,
       "step": 800
     },
     {
-      "epoch": 0.09148638467333979,
-      "grad_norm": 1.3023720979690552,
-      "learning_rate": 0.0005951513845490787,
-      "loss": 5.196,
+      "epoch": 0.09164420485175202,
+      "grad_norm": 1.6176457405090332,
+      "learning_rate": 0.0005951430113329735,
+      "loss": 5.2377,
       "step": 850
     },
     {
-      "epoch": 0.09686793671294801,
-      "grad_norm": 1.180203914642334,
-      "learning_rate": 0.0005948281435190174,
-      "loss": 5.1727,
+      "epoch": 0.09703504043126684,
+      "grad_norm": 1.1616325378417969,
+      "learning_rate": 0.0005948192120885051,
+      "loss": 5.1806,
       "step": 900
     },
     {
-      "epoch": 0.10224948875255624,
-      "grad_norm": 1.2121342420578003,
-      "learning_rate": 0.0005945049024889559,
-      "loss": 5.1051,
+      "epoch": 0.10242587601078167,
+      "grad_norm": 1.2763676643371582,
+      "learning_rate": 0.0005944954128440366,
+      "loss": 5.1497,
       "step": 950
     },
     {
-      "epoch": 0.10763104079216446,
-      "grad_norm": 1.023581624031067,
-      "learning_rate": 0.0005941816614588944,
-      "loss": 5.0744,
+      "epoch": 0.1078167115902965,
+      "grad_norm": 1.4034849405288696,
+      "learning_rate": 0.0005941716135995682,
+      "loss": 5.1096,
       "step": 1000
     },
     {
-      "epoch": 0.10763104079216446,
-      "eval_accuracy": 0.22928562573361072,
-      "eval_loss": 4.99822998046875,
-      "eval_runtime": 186.5326,
-      "eval_samples_per_second": 96.557,
-      "eval_steps_per_second": 6.036,
+      "epoch": 0.1078167115902965,
+      "eval_accuracy": 0.22586664531275244,
+      "eval_loss": 5.038193702697754,
+      "eval_runtime": 185.2686,
+      "eval_samples_per_second": 97.216,
+      "eval_steps_per_second": 6.078,
       "step": 1000
     },
     {
-      "epoch": 0.11301259283177269,
-      "grad_norm": 1.0498579740524292,
-      "learning_rate": 0.000593858420428833,
-      "loss": 5.0405,
+      "epoch": 0.11320754716981132,
+      "grad_norm": 1.0121814012527466,
+      "learning_rate": 0.0005938478143550997,
+      "loss": 5.0558,
       "step": 1050
     },
     {
-      "epoch": 0.1183941448713809,
-      "grad_norm": 0.9243935346603394,
-      "learning_rate": 0.0005935351793987716,
-      "loss": 4.9895,
+      "epoch": 0.11859838274932614,
+      "grad_norm": 1.0034334659576416,
+      "learning_rate": 0.0005935240151106314,
+      "loss": 5.0331,
       "step": 1100
     },
     {
-      "epoch": 0.12377569691098914,
-      "grad_norm": 1.389527440071106,
-      "learning_rate": 0.0005932119383687103,
-      "loss": 4.9848,
+      "epoch": 0.12398921832884097,
+      "grad_norm": 0.8851997256278992,
+      "learning_rate": 0.0005932002158661629,
+      "loss": 5.0028,
       "step": 1150
     },
     {
-      "epoch": 0.12915724895059735,
-      "grad_norm": 1.0520302057266235,
-      "learning_rate": 0.0005928886973386488,
-      "loss": 4.9556,
+      "epoch": 0.1293800539083558,
+      "grad_norm": 1.2870864868164062,
+      "learning_rate": 0.0005928764166216945,
+      "loss": 4.9641,
       "step": 1200
     },
     {
-      "epoch": 0.13453880099020557,
-      "grad_norm": 1.021660566329956,
-      "learning_rate": 0.0005925654563085874,
-      "loss": 4.942,
+      "epoch": 0.1347708894878706,
+      "grad_norm": 0.8656995892524719,
+      "learning_rate": 0.000592552617377226,
+      "loss": 4.9231,
       "step": 1250
     },
     {
-      "epoch": 0.1399203530298138,
-      "grad_norm": 1.0506354570388794,
-      "learning_rate": 0.000592242215278526,
-      "loss": 4.8751,
+      "epoch": 0.14016172506738545,
+      "grad_norm": 0.834179699420929,
+      "learning_rate": 0.0005922288181327577,
+      "loss": 4.893,
       "step": 1300
     },
     {
-      "epoch": 0.14530190506942203,
-      "grad_norm": 0.885286271572113,
-      "learning_rate": 0.0005919189742484645,
-      "loss": 4.839,
+      "epoch": 0.14555256064690028,
+      "grad_norm": 0.8964150547981262,
+      "learning_rate": 0.0005919050188882893,
+      "loss": 4.8803,
       "step": 1350
     },
     {
-      "epoch": 0.15068345710903025,
-      "grad_norm": 1.0981992483139038,
-      "learning_rate": 0.0005915957332184032,
-      "loss": 4.8421,
+      "epoch": 0.1509433962264151,
+      "grad_norm": 0.8977037072181702,
+      "learning_rate": 0.0005915812196438207,
+      "loss": 4.8643,
       "step": 1400
     },
     {
-      "epoch": 0.15606500914863847,
-      "grad_norm": 0.9436761736869812,
-      "learning_rate": 0.0005912724921883417,
-      "loss": 4.8018,
+      "epoch": 0.15633423180592992,
+      "grad_norm": 0.8278561234474182,
+      "learning_rate": 0.0005912574203993524,
+      "loss": 4.8143,
       "step": 1450
     },
     {
-      "epoch": 0.16144656118824668,
-      "grad_norm": 0.9398601055145264,
-      "learning_rate": 0.0005909492511582803,
-      "loss": 4.8121,
+      "epoch": 0.16172506738544473,
+      "grad_norm": 0.8696197271347046,
+      "learning_rate": 0.0005909336211548839,
+      "loss": 4.8275,
       "step": 1500
     },
     {
-      "epoch": 0.1668281132278549,
-      "grad_norm": 1.1647937297821045,
-      "learning_rate": 0.0005906260101282189,
-      "loss": 4.7572,
+      "epoch": 0.16711590296495957,
+      "grad_norm": 1.168330430984497,
+      "learning_rate": 0.0005906098219104155,
+      "loss": 4.7746,
       "step": 1550
     },
     {
-      "epoch": 0.17220966526746315,
-      "grad_norm": 1.0045080184936523,
-      "learning_rate": 0.0005903027690981575,
-      "loss": 4.7394,
+      "epoch": 0.1725067385444744,
+      "grad_norm": 0.8590855002403259,
+      "learning_rate": 0.000590286022665947,
+      "loss": 4.7758,
       "step": 1600
     },
     {
-      "epoch": 0.17759121730707136,
-      "grad_norm": 1.175984263420105,
-      "learning_rate": 0.000589979528068096,
-      "loss": 4.6874,
+      "epoch": 0.1778975741239892,
+      "grad_norm": 0.8608685731887817,
+      "learning_rate": 0.0005899622234214787,
+      "loss": 4.7332,
       "step": 1650
     },
     {
-      "epoch": 0.18297276934667958,
-      "grad_norm": 1.1310313940048218,
-      "learning_rate": 0.0005896562870380347,
-      "loss": 4.7045,
+      "epoch": 0.18328840970350405,
+      "grad_norm": 0.9251887202262878,
+      "learning_rate": 0.0005896384241770102,
+      "loss": 4.7203,
       "step": 1700
     },
     {
-      "epoch": 0.1883543213862878,
-      "grad_norm": 1.0819603204727173,
-      "learning_rate": 0.0005893330460079732,
-      "loss": 4.6951,
+      "epoch": 0.18867924528301888,
+      "grad_norm": 1.0499516725540161,
+      "learning_rate": 0.0005893146249325418,
+      "loss": 4.6829,
       "step": 1750
     },
     {
-      "epoch": 0.19373587342589602,
-      "grad_norm": 0.8239721059799194,
-      "learning_rate": 0.0005890098049779118,
-      "loss": 4.6495,
+      "epoch": 0.1940700808625337,
+      "grad_norm": 1.194277048110962,
+      "learning_rate": 0.0005889908256880733,
+      "loss": 4.693,
       "step": 1800
     },
     {
-      "epoch": 0.19911742546550426,
-      "grad_norm": 0.9362748265266418,
-      "learning_rate": 0.0005886865639478504,
-      "loss": 4.6468,
+      "epoch": 0.19946091644204852,
+      "grad_norm": 1.0574666261672974,
+      "learning_rate": 0.0005886670264436049,
+      "loss": 4.667,
       "step": 1850
     },
     {
-      "epoch": 0.20449897750511248,
-      "grad_norm": 0.7697787284851074,
-      "learning_rate": 0.0005883633229177889,
-      "loss": 4.6028,
+      "epoch": 0.20485175202156333,
+      "grad_norm": 0.9819419980049133,
+      "learning_rate": 0.0005883432271991365,
+      "loss": 4.6478,
       "step": 1900
     },
     {
-      "epoch": 0.2098805295447207,
-      "grad_norm": 0.9712828993797302,
-      "learning_rate": 0.0005880400818877276,
-      "loss": 4.5853,
+      "epoch": 0.21024258760107817,
+      "grad_norm": 0.9217617511749268,
+      "learning_rate": 0.0005880194279546681,
+      "loss": 4.6208,
       "step": 1950
     },
     {
-      "epoch": 0.2152620815843289,
-      "grad_norm": 0.9445996880531311,
-      "learning_rate": 0.0005877168408576662,
-      "loss": 4.5561,
+      "epoch": 0.215633423180593,
+      "grad_norm": 0.9863111972808838,
+      "learning_rate": 0.0005876956287101996,
+      "loss": 4.604,
       "step": 2000
     },
     {
-      "epoch": 0.2152620815843289,
-      "eval_accuracy": 0.2722129692215705,
-      "eval_loss": 4.49458122253418,
-      "eval_runtime": 186.7016,
-      "eval_samples_per_second": 96.469,
-      "eval_steps_per_second": 6.031,
+      "epoch": 0.215633423180593,
+      "eval_accuracy": 0.2681015439466706,
+      "eval_loss": 4.527613639831543,
+      "eval_runtime": 184.4764,
+      "eval_samples_per_second": 97.633,
+      "eval_steps_per_second": 6.104,
       "step": 2000
     },
     {
-      "epoch": 0.22064363362393713,
-      "grad_norm": 0.9033973217010498,
-      "learning_rate": 0.0005873935998276048,
-      "loss": 4.5482,
+      "epoch": 0.2210242587601078,
+      "grad_norm": 1.0152137279510498,
+      "learning_rate": 0.0005873718294657312,
+      "loss": 4.5828,
       "step": 2050
     },
     {
-      "epoch": 0.22602518566354537,
-      "grad_norm": 0.8677470684051514,
-      "learning_rate": 0.0005870703587975433,
-      "loss": 4.5466,
+      "epoch": 0.22641509433962265,
+      "grad_norm": 0.7376741766929626,
+      "learning_rate": 0.0005870480302212628,
+      "loss": 4.5676,
       "step": 2100
     },
     {
-      "epoch": 0.2314067377031536,
-      "grad_norm": 0.9266975522041321,
-      "learning_rate": 0.0005867471177674818,
-      "loss": 4.5252,
+      "epoch": 0.23180592991913745,
+      "grad_norm": 1.0539066791534424,
+      "learning_rate": 0.0005867242309767943,
+      "loss": 4.5439,
       "step": 2150
     },
     {
-      "epoch": 0.2367882897427618,
-      "grad_norm": 0.7946595549583435,
-      "learning_rate": 0.0005864238767374205,
-      "loss": 4.503,
+      "epoch": 0.2371967654986523,
+      "grad_norm": 0.930075466632843,
+      "learning_rate": 0.0005864004317323259,
+      "loss": 4.5241,
       "step": 2200
     },
     {
-      "epoch": 0.24216984178237003,
-      "grad_norm": 0.9363351464271545,
-      "learning_rate": 0.0005861006357073591,
-      "loss": 4.4913,
+      "epoch": 0.24258760107816713,
+      "grad_norm": 0.8790664672851562,
+      "learning_rate": 0.0005860766324878575,
+      "loss": 4.5101,
       "step": 2250
     },
     {
-      "epoch": 0.24755139382197827,
-      "grad_norm": 0.8072689175605774,
-      "learning_rate": 0.0005857773946772977,
-      "loss": 4.4651,
+      "epoch": 0.24797843665768193,
+      "grad_norm": 0.7741200923919678,
+      "learning_rate": 0.000585752833243389,
+      "loss": 4.521,
       "step": 2300
     },
     {
-      "epoch": 0.2529329458615865,
-      "grad_norm": 0.8488495349884033,
-      "learning_rate": 0.0005854541536472362,
-      "loss": 4.4535,
+      "epoch": 0.25336927223719674,
+      "grad_norm": 0.7596673369407654,
+      "learning_rate": 0.0005854290339989206,
+      "loss": 4.4722,
       "step": 2350
     },
     {
-      "epoch": 0.2583144979011947,
-      "grad_norm": 0.8391003608703613,
-      "learning_rate": 0.0005851309126171749,
-      "loss": 4.4476,
+      "epoch": 0.2587601078167116,
+      "grad_norm": 0.8529781699180603,
+      "learning_rate": 0.0005851052347544521,
+      "loss": 4.4743,
       "step": 2400
     },
     {
-      "epoch": 0.2636960499408029,
-      "grad_norm": 1.0908865928649902,
-      "learning_rate": 0.0005848076715871134,
-      "loss": 4.4171,
+      "epoch": 0.2641509433962264,
+      "grad_norm": 1.082383155822754,
+      "learning_rate": 0.0005847814355099838,
+      "loss": 4.4542,
       "step": 2450
     },
     {
-      "epoch": 0.26907760198041114,
-      "grad_norm": 1.001535415649414,
-      "learning_rate": 0.000584484430557052,
-      "loss": 4.4273,
+      "epoch": 0.2695417789757412,
+      "grad_norm": 0.8960773944854736,
+      "learning_rate": 0.0005844576362655154,
+      "loss": 4.4376,
       "step": 2500
     },
     {
-      "epoch": 0.27445915402001936,
-      "grad_norm": 0.7375698685646057,
-      "learning_rate": 0.0005841611895269906,
-      "loss": 4.4006,
+      "epoch": 0.2749326145552561,
+      "grad_norm": 0.7989738583564758,
+      "learning_rate": 0.0005841338370210469,
+      "loss": 4.4241,
       "step": 2550
     },
     {
-      "epoch": 0.2798407060596276,
-      "grad_norm": 0.759346604347229,
-      "learning_rate": 0.0005838379484969291,
-      "loss": 4.3815,
+      "epoch": 0.2803234501347709,
+      "grad_norm": 0.8554348945617676,
+      "learning_rate": 0.0005838100377765785,
+      "loss": 4.391,
       "step": 2600
     },
     {
-      "epoch": 0.2852222580992358,
-      "grad_norm": 0.8485270738601685,
-      "learning_rate": 0.0005835147074668678,
-      "loss": 4.369,
+      "epoch": 0.2857142857142857,
+      "grad_norm": 0.7921100854873657,
+      "learning_rate": 0.0005834862385321101,
+      "loss": 4.4044,
       "step": 2650
     },
     {
-      "epoch": 0.29060381013884407,
-      "grad_norm": 0.7628142833709717,
-      "learning_rate": 0.0005831914664368063,
-      "loss": 4.3509,
+      "epoch": 0.29110512129380056,
+      "grad_norm": 1.088876724243164,
+      "learning_rate": 0.0005831624392876417,
+      "loss": 4.4093,
       "step": 2700
     },
     {
-      "epoch": 0.2959853621784523,
-      "grad_norm": 0.849441647529602,
-      "learning_rate": 0.0005828682254067449,
-      "loss": 4.3661,
+      "epoch": 0.29649595687331537,
+      "grad_norm": 0.7506400942802429,
+      "learning_rate": 0.0005828386400431731,
+      "loss": 4.3686,
       "step": 2750
     },
     {
-      "epoch": 0.3013669142180605,
-      "grad_norm": 0.8873631954193115,
-      "learning_rate": 0.0005825449843766835,
-      "loss": 4.3404,
+      "epoch": 0.3018867924528302,
+      "grad_norm": 0.8032007217407227,
+      "learning_rate": 0.0005825148407987048,
+      "loss": 4.3501,
       "step": 2800
     },
     {
-      "epoch": 0.3067484662576687,
-      "grad_norm": 0.8451576232910156,
-      "learning_rate": 0.0005822217433466221,
-      "loss": 4.3318,
+      "epoch": 0.30727762803234504,
+      "grad_norm": 0.766340970993042,
+      "learning_rate": 0.0005821910415542363,
+      "loss": 4.3456,
       "step": 2850
     },
     {
-      "epoch": 0.31213001829727693,
-      "grad_norm": 0.8104711771011353,
-      "learning_rate": 0.0005818985023165607,
-      "loss": 4.3286,
+      "epoch": 0.31266846361185985,
+      "grad_norm": 0.8743217587471008,
+      "learning_rate": 0.0005818672423097679,
+      "loss": 4.3387,
       "step": 2900
     },
     {
-      "epoch": 0.31751157033688515,
-      "grad_norm": 0.7698219418525696,
-      "learning_rate": 0.0005815752612864992,
-      "loss": 4.3099,
+      "epoch": 0.31805929919137466,
+      "grad_norm": 0.9630153775215149,
+      "learning_rate": 0.0005815434430652994,
+      "loss": 4.328,
       "step": 2950
     },
     {
-      "epoch": 0.32289312237649337,
-      "grad_norm": 0.9506596922874451,
-      "learning_rate": 0.0005812520202564378,
-      "loss": 4.3234,
+      "epoch": 0.32345013477088946,
+      "grad_norm": 0.8393748998641968,
+      "learning_rate": 0.0005812196438208311,
+      "loss": 4.3278,
       "step": 3000
     },
     {
-      "epoch": 0.32289312237649337,
-      "eval_accuracy": 0.29898786575414177,
-      "eval_loss": 4.232949256896973,
-      "eval_runtime": 186.4523,
-      "eval_samples_per_second": 96.598,
-      "eval_steps_per_second": 6.039,
+      "epoch": 0.32345013477088946,
+      "eval_accuracy": 0.2963310853413021,
+      "eval_loss": 4.263187885284424,
+      "eval_runtime": 184.4828,
+      "eval_samples_per_second": 97.63,
+      "eval_steps_per_second": 6.104,
       "step": 3000
     },
     {
-      "epoch": 0.3282746744161016,
-      "grad_norm": 0.8630695939064026,
-      "learning_rate": 0.0005809287792263764,
-      "loss": 4.2979,
+      "epoch": 0.3288409703504043,
+      "grad_norm": 0.6937326788902283,
+      "learning_rate": 0.0005808958445763626,
+      "loss": 4.3134,
       "step": 3050
     },
     {
-      "epoch": 0.3336562264557098,
-      "grad_norm": 0.7666266560554504,
-      "learning_rate": 0.0005806055381963151,
-      "loss": 4.2939,
+      "epoch": 0.33423180592991913,
+      "grad_norm": 0.7280919551849365,
+      "learning_rate": 0.0005805720453318942,
+      "loss": 4.3174,
       "step": 3100
     },
     {
-      "epoch": 0.3390377784953181,
-      "grad_norm": 0.6829612851142883,
-      "learning_rate": 0.0005802822971662536,
-      "loss": 4.2741,
+      "epoch": 0.33962264150943394,
+      "grad_norm": 0.6627631187438965,
+      "learning_rate": 0.0005802482460874257,
+      "loss": 4.2816,
       "step": 3150
     },
     {
-      "epoch": 0.3444193305349263,
-      "grad_norm": 0.7251335978507996,
-      "learning_rate": 0.0005799590561361922,
-      "loss": 4.2651,
+      "epoch": 0.3450134770889488,
+      "grad_norm": 0.6677783727645874,
+      "learning_rate": 0.0005799244468429573,
+      "loss": 4.2732,
       "step": 3200
     },
     {
-      "epoch": 0.3498008825745345,
-      "grad_norm": 0.7338356971740723,
-      "learning_rate": 0.0005796358151061307,
-      "loss": 4.2541,
+      "epoch": 0.3504043126684636,
+      "grad_norm": 0.9010113477706909,
+      "learning_rate": 0.0005796006475984889,
+      "loss": 4.2735,
       "step": 3250
     },
     {
-      "epoch": 0.35518243461414273,
-      "grad_norm": 0.7907029986381531,
-      "learning_rate": 0.0005793125740760694,
-      "loss": 4.2562,
+      "epoch": 0.3557951482479784,
+      "grad_norm": 0.9248265624046326,
+      "learning_rate": 0.0005792768483540205,
+      "loss": 4.257,
       "step": 3300
     },
     {
-      "epoch": 0.36056398665375095,
-      "grad_norm": 0.6819180250167847,
-      "learning_rate": 0.0005789893330460079,
-      "loss": 4.2412,
+      "epoch": 0.3611859838274933,
+      "grad_norm": 0.7695420384407043,
+      "learning_rate": 0.000578953049109552,
+      "loss": 4.272,
       "step": 3350
     },
     {
-      "epoch": 0.36594553869335916,
-      "grad_norm": 0.7512629628181458,
-      "learning_rate": 0.0005786660920159465,
-      "loss": 4.2439,
+      "epoch": 0.3665768194070081,
+      "grad_norm": 0.6875022053718567,
+      "learning_rate": 0.0005786292498650836,
+      "loss": 4.2455,
       "step": 3400
     },
     {
-      "epoch": 0.3713270907329674,
-      "grad_norm": 0.6798344254493713,
-      "learning_rate": 0.0005783428509858851,
-      "loss": 4.2309,
+      "epoch": 0.3719676549865229,
+      "grad_norm": 0.624302089214325,
+      "learning_rate": 0.0005783054506206152,
+      "loss": 4.221,
       "step": 3450
     },
     {
-      "epoch": 0.3767086427725756,
-      "grad_norm": 0.7748376727104187,
-      "learning_rate": 0.0005780196099558237,
-      "loss": 4.2001,
+      "epoch": 0.37735849056603776,
+      "grad_norm": 0.7334246635437012,
+      "learning_rate": 0.0005779816513761467,
+      "loss": 4.2328,
       "step": 3500
     },
     {
-      "epoch": 0.3820901948121838,
-      "grad_norm": 0.7409430146217346,
-      "learning_rate": 0.0005776963689257623,
-      "loss": 4.2235,
+      "epoch": 0.38274932614555257,
+      "grad_norm": 0.7634935975074768,
+      "learning_rate": 0.0005776578521316782,
+      "loss": 4.2302,
       "step": 3550
     },
     {
-      "epoch": 0.38747174685179203,
-      "grad_norm": 0.9617530703544617,
-      "learning_rate": 0.0005773731278957008,
-      "loss": 4.2052,
+      "epoch": 0.3881401617250674,
+      "grad_norm": 0.6008784770965576,
+      "learning_rate": 0.0005773340528872099,
+      "loss": 4.2249,
       "step": 3600
     },
     {
-      "epoch": 0.3928532988914003,
-      "grad_norm": 0.655745267868042,
-      "learning_rate": 0.0005770498868656394,
-      "loss": 4.1884,
+      "epoch": 0.3935309973045822,
+      "grad_norm": 0.8262253403663635,
+      "learning_rate": 0.0005770102536427414,
+      "loss": 4.201,
       "step": 3650
     },
     {
-      "epoch": 0.3982348509310085,
-      "grad_norm": 0.6762064695358276,
-      "learning_rate": 0.000576726645835578,
-      "loss": 4.1592,
+      "epoch": 0.39892183288409705,
+      "grad_norm": 0.7404162883758545,
+      "learning_rate": 0.000576686454398273,
+      "loss": 4.2058,
       "step": 3700
     },
     {
-      "epoch": 0.40361640297061674,
-      "grad_norm": 0.6524355411529541,
-      "learning_rate": 0.0005764034048055167,
-      "loss": 4.179,
+      "epoch": 0.40431266846361186,
+      "grad_norm": 0.6769477128982544,
+      "learning_rate": 0.0005763626551538045,
+      "loss": 4.2074,
       "step": 3750
     },
     {
-      "epoch": 0.40899795501022496,
-      "grad_norm": 0.7709828615188599,
-      "learning_rate": 0.0005760801637754552,
-      "loss": 4.1615,
+      "epoch": 0.40970350404312667,
+      "grad_norm": 0.748205304145813,
+      "learning_rate": 0.0005760388559093362,
+      "loss": 4.1993,
       "step": 3800
     },
     {
-      "epoch": 0.4143795070498332,
-      "grad_norm": 0.6991027593612671,
-      "learning_rate": 0.0005757569227453937,
-      "loss": 4.1784,
+      "epoch": 0.41509433962264153,
+      "grad_norm": 0.6619985103607178,
+      "learning_rate": 0.0005757150566648678,
+      "loss": 4.1803,
       "step": 3850
     },
     {
-      "epoch": 0.4197610590894414,
-      "grad_norm": 0.7344850301742554,
-      "learning_rate": 0.0005754336817153324,
-      "loss": 4.1851,
+      "epoch": 0.42048517520215634,
+      "grad_norm": 0.796105146408081,
+      "learning_rate": 0.0005753912574203993,
+      "loss": 4.1784,
       "step": 3900
     },
     {
-      "epoch": 0.4251426111290496,
-      "grad_norm": 0.628821849822998,
-      "learning_rate": 0.0005751104406852709,
-      "loss": 4.1582,
+      "epoch": 0.42587601078167114,
+      "grad_norm": 0.7692325115203857,
+      "learning_rate": 0.0005750674581759309,
+      "loss": 4.1716,
       "step": 3950
     },
     {
-      "epoch": 0.4305241631686578,
-      "grad_norm": 0.7425068616867065,
-      "learning_rate": 0.0005747871996552096,
-      "loss": 4.1511,
+      "epoch": 0.431266846361186,
+      "grad_norm": 0.6746971011161804,
+      "learning_rate": 0.0005747436589314624,
+      "loss": 4.176,
       "step": 4000
     },
     {
-      "epoch": 0.4305241631686578,
-      "eval_accuracy": 0.31297170972164107,
-      "eval_loss": 4.083059310913086,
-      "eval_runtime": 186.6296,
-      "eval_samples_per_second": 96.507,
-      "eval_steps_per_second": 6.033,
+      "epoch": 0.431266846361186,
+      "eval_accuracy": 0.31160507369871054,
+      "eval_loss": 4.0974860191345215,
+      "eval_runtime": 184.4687,
+      "eval_samples_per_second": 97.637,
+      "eval_steps_per_second": 6.104,
       "step": 4000
     },
     {
-      "epoch": 0.43590571520826604,
-      "grad_norm": 0.6493635773658752,
-      "learning_rate": 0.0005744639586251481,
-      "loss": 4.143,
+      "epoch": 0.4366576819407008,
+      "grad_norm": 0.5227634906768799,
+      "learning_rate": 0.0005744198596869941,
+      "loss": 4.1619,
       "step": 4050
     },
     {
-      "epoch": 0.44128726724787426,
-      "grad_norm": 0.6777784824371338,
-      "learning_rate": 0.0005741407175950867,
-      "loss": 4.1415,
+      "epoch": 0.4420485175202156,
+      "grad_norm": 0.6957909464836121,
+      "learning_rate": 0.0005740960604425255,
+      "loss": 4.162,
       "step": 4100
     },
     {
-      "epoch": 0.44666881928748253,
-      "grad_norm": 0.6996566653251648,
-      "learning_rate": 0.0005738174765650253,
-      "loss": 4.1455,
+      "epoch": 0.4474393530997305,
+      "grad_norm": 0.6677095890045166,
+      "learning_rate": 0.0005737722611980572,
+      "loss": 4.1641,
       "step": 4150
     },
     {
-      "epoch": 0.45205037132709075,
-      "grad_norm": 0.612664520740509,
-      "learning_rate": 0.0005734942355349638,
-      "loss": 4.1115,
+      "epoch": 0.4528301886792453,
+      "grad_norm": 0.6342118382453918,
+      "learning_rate": 0.0005734484619535887,
+      "loss": 4.1476,
       "step": 4200
     },
     {
-      "epoch": 0.45743192336669897,
-      "grad_norm": 0.6381308436393738,
-      "learning_rate": 0.0005731709945049025,
-      "loss": 4.1224,
+      "epoch": 0.4582210242587601,
+      "grad_norm": 0.6470354199409485,
+      "learning_rate": 0.0005731246627091203,
+      "loss": 4.1337,
       "step": 4250
     },
     {
-      "epoch": 0.4628134754063072,
-      "grad_norm": 0.6527543067932129,
-      "learning_rate": 0.000572847753474841,
-      "loss": 4.1183,
+      "epoch": 0.4636118598382749,
+      "grad_norm": 0.9002943634986877,
+      "learning_rate": 0.0005728008634646518,
+      "loss": 4.1294,
       "step": 4300
     },
     {
-      "epoch": 0.4681950274459154,
-      "grad_norm": 0.6054431200027466,
-      "learning_rate": 0.0005725245124447796,
-      "loss": 4.1338,
+      "epoch": 0.46900269541778977,
+      "grad_norm": 0.5656282305717468,
+      "learning_rate": 0.0005724770642201835,
+      "loss": 4.127,
       "step": 4350
     },
     {
-      "epoch": 0.4735765794855236,
-      "grad_norm": 0.8053311705589294,
-      "learning_rate": 0.0005722012714147182,
-      "loss": 4.1137,
+      "epoch": 0.4743935309973046,
+      "grad_norm": 0.6128329038619995,
+      "learning_rate": 0.000572153264975715,
+      "loss": 4.1266,
       "step": 4400
     },
     {
-      "epoch": 0.47895813152513184,
-      "grad_norm": 0.5973644852638245,
-      "learning_rate": 0.0005718780303846568,
-      "loss": 4.103,
+      "epoch": 0.4797843665768194,
+      "grad_norm": 0.82874995470047,
+      "learning_rate": 0.0005718294657312466,
+      "loss": 4.1118,
       "step": 4450
     },
     {
-      "epoch": 0.48433968356474005,
-      "grad_norm": 0.6401163339614868,
-      "learning_rate": 0.0005715547893545953,
-      "loss": 4.0949,
+      "epoch": 0.48517520215633425,
+      "grad_norm": 0.7268750071525574,
+      "learning_rate": 0.0005715056664867781,
+      "loss": 4.1196,
       "step": 4500
     },
     {
-      "epoch": 0.48972123560434827,
-      "grad_norm": 0.6806808710098267,
-      "learning_rate": 0.000571231548324534,
-      "loss": 4.1037,
+      "epoch": 0.49056603773584906,
+      "grad_norm": 0.6136605739593506,
+      "learning_rate": 0.0005711818672423097,
+      "loss": 4.1113,
       "step": 4550
     },
     {
-      "epoch": 0.49510278764395654,
-      "grad_norm": 0.6933580040931702,
-      "learning_rate": 0.0005709083072944725,
-      "loss": 4.0763,
+      "epoch": 0.49595687331536387,
+      "grad_norm": 0.6665817499160767,
+      "learning_rate": 0.0005708580679978413,
+      "loss": 4.0952,
       "step": 4600
     },
     {
-      "epoch": 0.5004843396835648,
-      "grad_norm": 0.6767447590827942,
-      "learning_rate": 0.0005705850662644111,
-      "loss": 4.0822,
+      "epoch": 0.5013477088948787,
+      "grad_norm": 0.6104462146759033,
+      "learning_rate": 0.0005705342687533729,
+      "loss": 4.0902,
       "step": 4650
     },
     {
-      "epoch": 0.505865891723173,
-      "grad_norm": 0.6995888352394104,
-      "learning_rate": 0.0005702618252343497,
-      "loss": 4.0698,
+      "epoch": 0.5067385444743935,
+      "grad_norm": 0.6166984438896179,
+      "learning_rate": 0.0005702104695089044,
+      "loss": 4.1024,
       "step": 4700
     },
     {
-      "epoch": 0.5112474437627812,
-      "grad_norm": 0.6572487354278564,
-      "learning_rate": 0.0005699385842042882,
-      "loss": 4.0825,
+      "epoch": 0.5121293800539084,
+      "grad_norm": 0.5905686616897583,
+      "learning_rate": 0.000569886670264436,
+      "loss": 4.0918,
       "step": 4750
     },
     {
-      "epoch": 0.5166289958023894,
-      "grad_norm": 0.6858201622962952,
-      "learning_rate": 0.0005696153431742269,
-      "loss": 4.0665,
+      "epoch": 0.5175202156334232,
+      "grad_norm": 0.7475683689117432,
+      "learning_rate": 0.0005695628710199675,
+      "loss": 4.0868,
       "step": 4800
     },
     {
-      "epoch": 0.5220105478419976,
-      "grad_norm": 0.7944272756576538,
-      "learning_rate": 0.0005692921021441655,
-      "loss": 4.0662,
+      "epoch": 0.522911051212938,
+      "grad_norm": 0.6524274349212646,
+      "learning_rate": 0.0005692390717754991,
+      "loss": 4.0771,
       "step": 4850
     },
     {
-      "epoch": 0.5273920998816058,
-      "grad_norm": 0.67097407579422,
-      "learning_rate": 0.0005689688611141041,
-      "loss": 4.0623,
+      "epoch": 0.5283018867924528,
+      "grad_norm": 0.6513181924819946,
+      "learning_rate": 0.0005689152725310306,
+      "loss": 4.0601,
       "step": 4900
     },
     {
-      "epoch": 0.5327736519212141,
-      "grad_norm": 0.6387117505073547,
-      "learning_rate": 0.0005686456200840426,
-      "loss": 4.0638,
+      "epoch": 0.5336927223719676,
+      "grad_norm": 0.6451146602630615,
+      "learning_rate": 0.0005685914732865623,
+      "loss": 4.0743,
       "step": 4950
     },
     {
-      "epoch": 0.5381552039608223,
-      "grad_norm": 0.6278985738754272,
-      "learning_rate": 0.0005683223790539811,
-      "loss": 4.0601,
+      "epoch": 0.5390835579514824,
+      "grad_norm": 0.5819371938705444,
+      "learning_rate": 0.0005682676740420939,
+      "loss": 4.0578,
       "step": 5000
     },
     {
-      "epoch": 0.5381552039608223,
-      "eval_accuracy": 0.3228039268457491,
-      "eval_loss": 3.9849302768707275,
-      "eval_runtime": 186.7065,
-      "eval_samples_per_second": 96.467,
-      "eval_steps_per_second": 6.031,
+      "epoch": 0.5390835579514824,
+      "eval_accuracy": 0.320412694090728,
+      "eval_loss": 4.000001430511475,
+      "eval_runtime": 184.3009,
+      "eval_samples_per_second": 97.726,
+      "eval_steps_per_second": 6.11,
       "step": 5000
     },
     {
-      "epoch": 0.5435367560004305,
-      "grad_norm": 0.703794002532959,
-      "learning_rate": 0.0005679991380239198,
-      "loss": 4.0259,
+      "epoch": 0.5444743935309974,
+      "grad_norm": 0.7117781043052673,
+      "learning_rate": 0.0005679438747976254,
+      "loss": 4.0674,
       "step": 5050
     },
     {
-      "epoch": 0.5489183080400387,
-      "grad_norm": 0.6369916200637817,
-      "learning_rate": 0.0005676758969938584,
-      "loss": 4.0381,
+      "epoch": 0.5498652291105122,
+      "grad_norm": 0.6730750799179077,
+      "learning_rate": 0.000567620075553157,
+      "loss": 4.059,
       "step": 5100
     },
     {
-      "epoch": 0.5542998600796469,
-      "grad_norm": 0.7323727011680603,
-      "learning_rate": 0.000567352655963797,
-      "loss": 4.0341,
+      "epoch": 0.555256064690027,
+      "grad_norm": 0.7025057673454285,
+      "learning_rate": 0.0005672962763086886,
+      "loss": 4.0563,
       "step": 5150
     },
     {
-      "epoch": 0.5596814121192552,
-      "grad_norm": 0.6766077280044556,
-      "learning_rate": 0.0005670294149337355,
-      "loss": 4.0433,
+      "epoch": 0.5606469002695418,
+      "grad_norm": 0.6750470995903015,
+      "learning_rate": 0.0005669724770642202,
+      "loss": 4.0779,
       "step": 5200
     },
     {
-      "epoch": 0.5650629641588634,
-      "grad_norm": 0.5812702178955078,
-      "learning_rate": 0.0005667061739036742,
-      "loss": 4.029,
+      "epoch": 0.5660377358490566,
+      "grad_norm": 0.620905339717865,
+      "learning_rate": 0.0005666486778197517,
+      "loss": 4.0533,
       "step": 5250
     },
     {
-      "epoch": 0.5704445161984716,
-      "grad_norm": 0.5775976181030273,
-      "learning_rate": 0.0005663829328736127,
-      "loss": 4.0243,
+      "epoch": 0.5714285714285714,
+      "grad_norm": 0.7228374481201172,
+      "learning_rate": 0.0005663248785752833,
+      "loss": 4.0429,
       "step": 5300
     },
     {
-      "epoch": 0.5758260682380799,
-      "grad_norm": 0.5382300019264221,
-      "learning_rate": 0.0005660596918435512,
-      "loss": 4.0105,
+      "epoch": 0.5768194070080862,
+      "grad_norm": 0.6635624766349792,
+      "learning_rate": 0.0005660010793308148,
+      "loss": 4.0345,
       "step": 5350
     },
     {
-      "epoch": 0.5812076202776881,
-      "grad_norm": 0.5902352929115295,
-      "learning_rate": 0.0005657364508134899,
-      "loss": 3.9967,
+      "epoch": 0.5822102425876011,
+      "grad_norm": 0.5901010036468506,
+      "learning_rate": 0.0005656772800863465,
+      "loss": 4.0504,
       "step": 5400
     },
     {
-      "epoch": 0.5865891723172963,
-      "grad_norm": 0.6404397487640381,
-      "learning_rate": 0.0005654132097834284,
-      "loss": 4.0134,
+      "epoch": 0.5876010781671159,
+      "grad_norm": 0.5243663787841797,
+      "learning_rate": 0.0005653534808418779,
+      "loss": 4.0443,
       "step": 5450
     },
     {
-      "epoch": 0.5919707243569046,
-      "grad_norm": 0.6157573461532593,
-      "learning_rate": 0.0005650899687533671,
-      "loss": 4.0014,
+      "epoch": 0.5929919137466307,
+      "grad_norm": 0.7339888215065002,
+      "learning_rate": 0.0005650296815974096,
+      "loss": 4.0266,
       "step": 5500
     },
     {
-      "epoch": 0.5973522763965128,
-      "grad_norm": 0.6050103306770325,
-      "learning_rate": 0.0005647667277233056,
-      "loss": 4.0225,
+      "epoch": 0.5983827493261455,
+      "grad_norm": 0.6756436824798584,
+      "learning_rate": 0.0005647058823529411,
+      "loss": 4.0169,
       "step": 5550
     },
     {
-      "epoch": 0.602733828436121,
-      "grad_norm": 0.6549403667449951,
-      "learning_rate": 0.0005644434866932442,
-      "loss": 3.9842,
+      "epoch": 0.6037735849056604,
+      "grad_norm": 0.7483716607093811,
+      "learning_rate": 0.0005643820831084727,
+      "loss": 4.035,
       "step": 5600
     },
     {
-      "epoch": 0.6081153804757292,
-      "grad_norm": 0.7169498205184937,
-      "learning_rate": 0.0005641202456631828,
-      "loss": 4.0126,
+      "epoch": 0.6091644204851752,
+      "grad_norm": 0.6512123942375183,
+      "learning_rate": 0.0005640582838640042,
+      "loss": 4.0145,
       "step": 5650
     },
     {
-      "epoch": 0.6134969325153374,
-      "grad_norm": 0.5801643133163452,
-      "learning_rate": 0.0005637970046331214,
-      "loss": 3.9927,
+      "epoch": 0.6145552560646901,
+      "grad_norm": 0.6199039816856384,
+      "learning_rate": 0.0005637344846195358,
+      "loss": 4.0128,
       "step": 5700
     },
     {
-      "epoch": 0.6188784845549457,
-      "grad_norm": 0.6069376468658447,
-      "learning_rate": 0.00056347376360306,
-      "loss": 3.9963,
+      "epoch": 0.6199460916442049,
+      "grad_norm": 0.6374390721321106,
+      "learning_rate": 0.0005634106853750674,
+      "loss": 3.9857,
       "step": 5750
     },
     {
-      "epoch": 0.6242600365945539,
-      "grad_norm": 0.8185081481933594,
-      "learning_rate": 0.0005631505225729985,
-      "loss": 3.9933,
+      "epoch": 0.6253369272237197,
+      "grad_norm": 0.5695760250091553,
+      "learning_rate": 0.000563086886130599,
+      "loss": 3.9947,
       "step": 5800
     },
     {
-      "epoch": 0.6296415886341621,
-      "grad_norm": 0.5788935422897339,
-      "learning_rate": 0.0005628272815429371,
-      "loss": 3.9856,
+      "epoch": 0.6307277628032345,
+      "grad_norm": 0.7779887914657593,
+      "learning_rate": 0.0005627630868861305,
+      "loss": 4.0067,
       "step": 5850
     },
     {
-      "epoch": 0.6350231406737703,
-      "grad_norm": 0.7359537482261658,
-      "learning_rate": 0.0005625040405128757,
-      "loss": 3.9817,
+      "epoch": 0.6361185983827493,
+      "grad_norm": 0.676850438117981,
+      "learning_rate": 0.0005624392876416621,
+      "loss": 4.0051,
       "step": 5900
     },
     {
-      "epoch": 0.6404046927133785,
-      "grad_norm": 0.5951809287071228,
-      "learning_rate": 0.0005621807994828143,
-      "loss": 3.9946,
+      "epoch": 0.6415094339622641,
+      "grad_norm": 0.6918540000915527,
+      "learning_rate": 0.0005621154883971937,
+      "loss": 3.9884,
       "step": 5950
     },
     {
-      "epoch": 0.6457862447529867,
-      "grad_norm": 0.720151424407959,
-      "learning_rate": 0.0005618575584527529,
-      "loss": 3.9803,
+      "epoch": 0.6469002695417789,
+      "grad_norm": 0.6243817210197449,
+      "learning_rate": 0.0005617916891527253,
+      "loss": 4.0039,
       "step": 6000
     },
     {
-      "epoch": 0.6457862447529867,
-      "eval_accuracy": 0.3290248479375672,
-      "eval_loss": 3.9098801612854004,
-      "eval_runtime": 186.5341,
-      "eval_samples_per_second": 96.556,
-      "eval_steps_per_second": 6.036,
+      "epoch": 0.6469002695417789,
+      "eval_accuracy": 0.3272036082320208,
+      "eval_loss": 3.9277641773223877,
+      "eval_runtime": 184.5228,
+      "eval_samples_per_second": 97.609,
+      "eval_steps_per_second": 6.102,
       "step": 6000
     },
     {
-      "epoch": 0.651167796792595,
-      "grad_norm": 0.622701108455658,
-      "learning_rate": 0.0005615343174226915,
-      "loss": 3.9688,
+      "epoch": 0.6522911051212938,
+      "grad_norm": 0.6072673797607422,
+      "learning_rate": 0.0005614678899082568,
+      "loss": 3.9743,
       "step": 6050
     },
     {
-      "epoch": 0.6565493488322032,
-      "grad_norm": 0.6100315451622009,
-      "learning_rate": 0.00056121107639263,
-      "loss": 3.9644,
+      "epoch": 0.6576819407008087,
+      "grad_norm": 0.6629570722579956,
+      "learning_rate": 0.0005611440906637884,
+      "loss": 4.0071,
       "step": 6100
     },
     {
-      "epoch": 0.6619309008718114,
-      "grad_norm": 0.6101928949356079,
-      "learning_rate": 0.0005608878353625687,
-      "loss": 3.9617,
+      "epoch": 0.6630727762803235,
+      "grad_norm": 0.8738489151000977,
+      "learning_rate": 0.00056082029141932,
+      "loss": 3.99,
       "step": 6150
     },
     {
-      "epoch": 0.6673124529114196,
-      "grad_norm": 0.5649211406707764,
-      "learning_rate": 0.0005605645943325072,
-      "loss": 3.9615,
+      "epoch": 0.6684636118598383,
+      "grad_norm": 0.575965166091919,
+      "learning_rate": 0.0005604964921748515,
+      "loss": 3.9612,
       "step": 6200
     },
     {
-      "epoch": 0.6726940049510278,
-      "grad_norm": 0.5823198556900024,
-      "learning_rate": 0.0005602413533024458,
-      "loss": 3.9486,
+      "epoch": 0.6738544474393531,
+      "grad_norm": 0.6118637323379517,
+      "learning_rate": 0.000560172692930383,
+      "loss": 3.9798,
       "step": 6250
     },
     {
-      "epoch": 0.6780755569906362,
-      "grad_norm": 0.583858072757721,
-      "learning_rate": 0.0005599181122723844,
-      "loss": 3.9608,
+      "epoch": 0.6792452830188679,
+      "grad_norm": 0.6780696511268616,
+      "learning_rate": 0.0005598488936859147,
+      "loss": 3.9855,
       "step": 6300
     },
     {
-      "epoch": 0.6834571090302444,
-      "grad_norm": 0.5719197988510132,
-      "learning_rate": 0.000559594871242323,
-      "loss": 3.9735,
+      "epoch": 0.6846361185983828,
+      "grad_norm": 0.5279508233070374,
+      "learning_rate": 0.0005595250944414463,
+      "loss": 3.9704,
       "step": 6350
     },
     {
-      "epoch": 0.6888386610698526,
-      "grad_norm": 0.5020032525062561,
-      "learning_rate": 0.0005592716302122616,
-      "loss": 3.9507,
+      "epoch": 0.6900269541778976,
+      "grad_norm": 0.6573193669319153,
+      "learning_rate": 0.0005592012951969778,
+      "loss": 3.9787,
       "step": 6400
     },
     {
-      "epoch": 0.6942202131094608,
-      "grad_norm": 0.6220969557762146,
-      "learning_rate": 0.0005589483891822001,
-      "loss": 3.9613,
+      "epoch": 0.6954177897574124,
+      "grad_norm": 0.7975975871086121,
+      "learning_rate": 0.0005588839719373988,
+      "loss": 3.9782,
       "step": 6450
     },
     {
-      "epoch": 0.699601765149069,
-      "grad_norm": 0.5914574265480042,
-      "learning_rate": 0.0005586251481521387,
-      "loss": 3.9499,
+      "epoch": 0.7008086253369272,
+      "grad_norm": 0.670693039894104,
+      "learning_rate": 0.0005585601726929303,
+      "loss": 3.9711,
       "step": 6500
     },
     {
-      "epoch": 0.7049833171886772,
-      "grad_norm": 0.5620157718658447,
-      "learning_rate": 0.0005583019071220773,
-      "loss": 3.9545,
+      "epoch": 0.706199460916442,
+      "grad_norm": 0.5949117541313171,
+      "learning_rate": 0.0005582363734484619,
+      "loss": 3.9672,
       "step": 6550
     },
     {
-      "epoch": 0.7103648692282855,
-      "grad_norm": 0.5809386372566223,
-      "learning_rate": 0.000557978666092016,
-      "loss": 3.9384,
+      "epoch": 0.7115902964959568,
+      "grad_norm": 0.6047245264053345,
+      "learning_rate": 0.0005579125742039935,
+      "loss": 3.9684,
       "step": 6600
     },
     {
-      "epoch": 0.7157464212678937,
-      "grad_norm": 0.6262916326522827,
-      "learning_rate": 0.0005576554250619545,
-      "loss": 3.9595,
+      "epoch": 0.7169811320754716,
+      "grad_norm": 0.6255912184715271,
+      "learning_rate": 0.0005575952509444145,
+      "loss": 3.981,
       "step": 6650
     },
     {
-      "epoch": 0.7211279733075019,
-      "grad_norm": 0.5774989128112793,
-      "learning_rate": 0.000557332184031893,
-      "loss": 3.9212,
+      "epoch": 0.7223719676549866,
+      "grad_norm": 0.593988835811615,
+      "learning_rate": 0.000557271451699946,
+      "loss": 3.9416,
       "step": 6700
     },
     {
-      "epoch": 0.7265095253471101,
-      "grad_norm": 0.6159483790397644,
-      "learning_rate": 0.0005570089430018317,
-      "loss": 3.9184,
+      "epoch": 0.7277628032345014,
+      "grad_norm": 0.628450870513916,
+      "learning_rate": 0.0005569476524554775,
+      "loss": 3.9688,
       "step": 6750
     },
     {
-      "epoch": 0.7318910773867183,
-      "grad_norm": 0.6034411787986755,
-      "learning_rate": 0.0005566857019717702,
-      "loss": 3.934,
+      "epoch": 0.7331536388140162,
+      "grad_norm": 0.6474142074584961,
+      "learning_rate": 0.0005566238532110091,
+      "loss": 3.9424,
       "step": 6800
     },
     {
-      "epoch": 0.7372726294263265,
-      "grad_norm": 0.6545365452766418,
-      "learning_rate": 0.0005563624609417089,
-      "loss": 3.9149,
+      "epoch": 0.738544474393531,
+      "grad_norm": 0.5570947527885437,
+      "learning_rate": 0.0005563000539665407,
+      "loss": 3.9429,
       "step": 6850
     },
     {
-      "epoch": 0.7426541814659348,
-      "grad_norm": 0.6036293506622314,
-      "learning_rate": 0.0005560392199116474,
-      "loss": 3.9266,
+      "epoch": 0.7439353099730458,
+      "grad_norm": 0.5360486507415771,
+      "learning_rate": 0.0005559762547220722,
+      "loss": 3.9446,
       "step": 6900
     },
     {
-      "epoch": 0.748035733505543,
-      "grad_norm": 0.5515508055686951,
-      "learning_rate": 0.000555715978881586,
-      "loss": 3.9142,
+      "epoch": 0.7493261455525606,
+      "grad_norm": 0.6167177557945251,
+      "learning_rate": 0.0005556524554776038,
+      "loss": 3.9469,
       "step": 6950
     },
     {
-      "epoch": 0.7534172855451512,
-      "grad_norm": 0.5949258804321289,
-      "learning_rate": 0.0005553927378515246,
-      "loss": 3.9214,
+      "epoch": 0.7547169811320755,
+      "grad_norm": 0.6586635112762451,
+      "learning_rate": 0.0005553286562331354,
+      "loss": 3.9328,
       "step": 7000
     },
     {
-      "epoch": 0.7534172855451512,
-      "eval_accuracy": 0.33420052824860996,
-      "eval_loss": 3.85199236869812,
-      "eval_runtime": 186.3779,
-      "eval_samples_per_second": 96.637,
-      "eval_steps_per_second": 6.041,
+      "epoch": 0.7547169811320755,
+      "eval_accuracy": 0.3323766808737561,
+      "eval_loss": 3.8738908767700195,
+      "eval_runtime": 184.1434,
+      "eval_samples_per_second": 97.81,
+      "eval_steps_per_second": 6.115,
       "step": 7000
     },
     {
-      "epoch": 0.7587988375847594,
-      "grad_norm": 0.5928055644035339,
-      "learning_rate": 0.0005550694968214631,
-      "loss": 3.9401,
+      "epoch": 0.7601078167115903,
+      "grad_norm": 0.7064656615257263,
+      "learning_rate": 0.000555004856988667,
+      "loss": 3.9442,
       "step": 7050
     },
     {
-      "epoch": 0.7641803896243676,
-      "grad_norm": 0.5792833566665649,
-      "learning_rate": 0.0005547462557914018,
-      "loss": 3.9288,
+      "epoch": 0.7654986522911051,
+      "grad_norm": 0.8729749917984009,
+      "learning_rate": 0.0005546810577441985,
+      "loss": 3.9179,
       "step": 7100
     },
     {
-      "epoch": 0.7695619416639758,
-      "grad_norm": 0.6082324981689453,
-      "learning_rate": 0.0005544230147613403,
-      "loss": 3.926,
+      "epoch": 0.77088948787062,
+      "grad_norm": 0.6396660208702087,
+      "learning_rate": 0.0005543572584997301,
+      "loss": 3.9424,
       "step": 7150
     },
     {
-      "epoch": 0.7749434937035841,
-      "grad_norm": 0.5671685338020325,
-      "learning_rate": 0.0005540997737312789,
-      "loss": 3.8848,
+      "epoch": 0.7762803234501348,
+      "grad_norm": 0.6164191365242004,
+      "learning_rate": 0.0005540334592552616,
+      "loss": 3.9143,
       "step": 7200
     },
     {
-      "epoch": 0.7803250457431924,
-      "grad_norm": 0.5366504192352295,
-      "learning_rate": 0.0005537829975218188,
-      "loss": 3.9265,
+      "epoch": 0.7816711590296496,
+      "grad_norm": 0.6043381690979004,
+      "learning_rate": 0.0005537096600107933,
+      "loss": 3.9084,
       "step": 7250
     },
     {
-      "epoch": 0.7857065977828006,
-      "grad_norm": 0.600283145904541,
-      "learning_rate": 0.0005534597564917573,
-      "loss": 3.9159,
+      "epoch": 0.7870619946091644,
+      "grad_norm": 0.5565491914749146,
+      "learning_rate": 0.0005533858607663249,
+      "loss": 3.9203,
       "step": 7300
     },
     {
-      "epoch": 0.7910881498224088,
-      "grad_norm": 0.6155536770820618,
-      "learning_rate": 0.0005531365154616959,
-      "loss": 3.905,
+      "epoch": 0.7924528301886793,
+      "grad_norm": 0.619292676448822,
+      "learning_rate": 0.0005530620615218564,
+      "loss": 3.9231,
       "step": 7350
     },
     {
-      "epoch": 0.796469701862017,
-      "grad_norm": 0.5430896282196045,
-      "learning_rate": 0.0005528132744316344,
-      "loss": 3.8915,
+      "epoch": 0.7978436657681941,
+      "grad_norm": 0.5440534949302673,
+      "learning_rate": 0.000552738262277388,
+      "loss": 3.9186,
       "step": 7400
     },
     {
-      "epoch": 0.8018512539016253,
-      "grad_norm": 0.6370888352394104,
-      "learning_rate": 0.0005524900334015731,
-      "loss": 3.8976,
+      "epoch": 0.8032345013477089,
+      "grad_norm": 0.5794094204902649,
+      "learning_rate": 0.0005524144630329196,
+      "loss": 3.8956,
       "step": 7450
     },
     {
-      "epoch": 0.8072328059412335,
-      "grad_norm": 0.5839101672172546,
-      "learning_rate": 0.0005521667923715117,
-      "loss": 3.8896,
+      "epoch": 0.8086253369272237,
+      "grad_norm": 0.5809280276298523,
+      "learning_rate": 0.000552090663788451,
+      "loss": 3.9112,
       "step": 7500
     },
     {
-      "epoch": 0.8126143579808417,
-      "grad_norm": 0.6330485343933105,
-      "learning_rate": 0.0005518435513414502,
-      "loss": 3.8902,
+      "epoch": 0.8140161725067385,
+      "grad_norm": 0.6247937679290771,
+      "learning_rate": 0.0005517668645439827,
+      "loss": 3.9061,
       "step": 7550
     },
     {
-      "epoch": 0.8179959100204499,
-      "grad_norm": 0.6191490292549133,
-      "learning_rate": 0.0005515203103113888,
-      "loss": 3.8757,
+      "epoch": 0.8194070080862533,
+      "grad_norm": 0.5626509785652161,
+      "learning_rate": 0.0005514430652995143,
+      "loss": 3.9007,
       "step": 7600
     },
     {
-      "epoch": 0.8233774620600581,
-      "grad_norm": 0.5859811305999756,
-      "learning_rate": 0.0005511970692813274,
-      "loss": 3.871,
+      "epoch": 0.8247978436657682,
+      "grad_norm": 0.5623478889465332,
+      "learning_rate": 0.0005511192660550458,
+      "loss": 3.905,
       "step": 7650
     },
     {
-      "epoch": 0.8287590140996663,
-      "grad_norm": 0.5827727317810059,
-      "learning_rate": 0.000550873828251266,
-      "loss": 3.8757,
+      "epoch": 0.8301886792452831,
+      "grad_norm": 0.6041727662086487,
+      "learning_rate": 0.0005507954668105774,
+      "loss": 3.8991,
       "step": 7700
     },
     {
-      "epoch": 0.8341405661392746,
-      "grad_norm": 0.5251996517181396,
-      "learning_rate": 0.0005505505872212045,
-      "loss": 3.8746,
+      "epoch": 0.8355795148247979,
+      "grad_norm": 0.5649747848510742,
+      "learning_rate": 0.0005504716675661089,
+      "loss": 3.9012,
       "step": 7750
     },
     {
-      "epoch": 0.8395221181788828,
-      "grad_norm": 0.5536118745803833,
-      "learning_rate": 0.0005502273461911432,
-      "loss": 3.8749,
+      "epoch": 0.8409703504043127,
+      "grad_norm": 0.7091898322105408,
+      "learning_rate": 0.0005501478683216406,
+      "loss": 3.8844,
       "step": 7800
     },
     {
-      "epoch": 0.844903670218491,
-      "grad_norm": 0.5450619459152222,
-      "learning_rate": 0.0005499041051610817,
-      "loss": 3.8721,
+      "epoch": 0.8463611859838275,
+      "grad_norm": 0.5706016421318054,
+      "learning_rate": 0.0005498240690771721,
+      "loss": 3.8897,
       "step": 7850
     },
     {
-      "epoch": 0.8502852222580992,
-      "grad_norm": 0.6152466535568237,
-      "learning_rate": 0.0005495808641310204,
-      "loss": 3.8611,
+      "epoch": 0.8517520215633423,
+      "grad_norm": 0.5604148507118225,
+      "learning_rate": 0.0005495002698327037,
+      "loss": 3.8844,
       "step": 7900
     },
     {
-      "epoch": 0.8556667742977074,
-      "grad_norm": 0.5673686861991882,
-      "learning_rate": 0.0005492576231009589,
-      "loss": 3.8843,
+      "epoch": 0.8571428571428571,
+      "grad_norm": 0.6884514689445496,
+      "learning_rate": 0.0005491764705882352,
+      "loss": 3.9167,
       "step": 7950
     },
     {
-      "epoch": 0.8610483263373157,
-      "grad_norm": 0.5521990060806274,
-      "learning_rate": 0.0005489343820708974,
-      "loss": 3.8781,
+      "epoch": 0.862533692722372,
+      "grad_norm": 0.5939817428588867,
+      "learning_rate": 0.0005488526713437669,
+      "loss": 3.8634,
       "step": 8000
     },
     {
-      "epoch": 0.8610483263373157,
-      "eval_accuracy": 0.33836769245495874,
-      "eval_loss": 3.808518409729004,
-      "eval_runtime": 186.4389,
-      "eval_samples_per_second": 96.605,
-      "eval_steps_per_second": 6.04,
+      "epoch": 0.862533692722372,
+      "eval_accuracy": 0.3365711169549463,
+      "eval_loss": 3.825241804122925,
+      "eval_runtime": 184.429,
+      "eval_samples_per_second": 97.658,
+      "eval_steps_per_second": 6.105,
       "step": 8000
     },
     {
-      "epoch": 0.8664298783769239,
-      "grad_norm": 0.7532389163970947,
-      "learning_rate": 0.0005486111410408361,
-      "loss": 3.8919,
+      "epoch": 0.8679245283018868,
+      "grad_norm": 0.6615479588508606,
+      "learning_rate": 0.0005485288720992984,
+      "loss": 3.9032,
       "step": 8050
     },
     {
-      "epoch": 0.8718114304165321,
-      "grad_norm": 0.4985487461090088,
-      "learning_rate": 0.0005482879000107746,
-      "loss": 3.8845,
+      "epoch": 0.8733153638814016,
+      "grad_norm": 0.6856676936149597,
+      "learning_rate": 0.0005482050728548299,
+      "loss": 3.8819,
       "step": 8100
     },
     {
-      "epoch": 0.8771929824561403,
-      "grad_norm": 0.5289489030838013,
-      "learning_rate": 0.0005479646589807133,
-      "loss": 3.8701,
+      "epoch": 0.8787061994609164,
+      "grad_norm": 0.6637502312660217,
+      "learning_rate": 0.0005478812736103615,
+      "loss": 3.8865,
       "step": 8150
     },
     {
-      "epoch": 0.8825745344957485,
-      "grad_norm": 0.5698693990707397,
-      "learning_rate": 0.0005476414179506518,
-      "loss": 3.8722,
+      "epoch": 0.8840970350404312,
+      "grad_norm": 0.6695827841758728,
+      "learning_rate": 0.0005475574743658931,
+      "loss": 3.8805,
       "step": 8200
     },
     {
-      "epoch": 0.8879560865353568,
-      "grad_norm": 0.5225383639335632,
-      "learning_rate": 0.0005473181769205904,
-      "loss": 3.8707,
+      "epoch": 0.889487870619946,
+      "grad_norm": 0.6298319101333618,
+      "learning_rate": 0.0005472336751214246,
+      "loss": 3.8768,
       "step": 8250
     },
     {
-      "epoch": 0.8933376385749651,
-      "grad_norm": 0.5693679451942444,
-      "learning_rate": 0.000546994935890529,
-      "loss": 3.8796,
+      "epoch": 0.894878706199461,
+      "grad_norm": 0.6043828725814819,
+      "learning_rate": 0.0005469098758769562,
+      "loss": 3.8959,
       "step": 8300
     },
     {
-      "epoch": 0.8987191906145733,
-      "grad_norm": 0.5720039010047913,
-      "learning_rate": 0.0005466716948604677,
-      "loss": 3.8676,
+      "epoch": 0.9002695417789758,
+      "grad_norm": 0.5677790641784668,
+      "learning_rate": 0.0005465860766324878,
+      "loss": 3.8661,
       "step": 8350
     },
     {
-      "epoch": 0.9041007426541815,
-      "grad_norm": 0.5858473777770996,
-      "learning_rate": 0.0005463484538304062,
-      "loss": 3.8574,
+      "epoch": 0.9056603773584906,
+      "grad_norm": 0.6001612544059753,
+      "learning_rate": 0.0005462622773880194,
+      "loss": 3.8729,
       "step": 8400
     },
     {
-      "epoch": 0.9094822946937897,
-      "grad_norm": 0.6164777874946594,
-      "learning_rate": 0.0005460252128003447,
-      "loss": 3.8595,
+      "epoch": 0.9110512129380054,
+      "grad_norm": 0.5244486331939697,
+      "learning_rate": 0.000545938478143551,
+      "loss": 3.8667,
       "step": 8450
     },
     {
-      "epoch": 0.9148638467333979,
-      "grad_norm": 0.5653186440467834,
-      "learning_rate": 0.0005457019717702833,
-      "loss": 3.8554,
+      "epoch": 0.9164420485175202,
+      "grad_norm": 0.5915636420249939,
+      "learning_rate": 0.0005456146788990825,
+      "loss": 3.8724,
       "step": 8500
     },
     {
-      "epoch": 0.9202453987730062,
-      "grad_norm": 0.5244986414909363,
-      "learning_rate": 0.0005453787307402219,
-      "loss": 3.8548,
+      "epoch": 0.921832884097035,
+      "grad_norm": 0.5782058238983154,
+      "learning_rate": 0.000545290879654614,
+      "loss": 3.8669,
       "step": 8550
     },
     {
-      "epoch": 0.9256269508126144,
-      "grad_norm": 0.5365890860557556,
-      "learning_rate": 0.0005450554897101605,
-      "loss": 3.8624,
+      "epoch": 0.9272237196765498,
+      "grad_norm": 0.5795342922210693,
+      "learning_rate": 0.0005449670804101457,
+      "loss": 3.8738,
       "step": 8600
     },
     {
-      "epoch": 0.9310085028522226,
-      "grad_norm": 0.5696502923965454,
-      "learning_rate": 0.0005447322486800991,
-      "loss": 3.8549,
+      "epoch": 0.9326145552560647,
+      "grad_norm": 0.5499997735023499,
+      "learning_rate": 0.0005446432811656773,
+      "loss": 3.8516,
       "step": 8650
     },
     {
-      "epoch": 0.9363900548918308,
-      "grad_norm": 0.5654961466789246,
-      "learning_rate": 0.0005444090076500377,
-      "loss": 3.8445,
+      "epoch": 0.9380053908355795,
+      "grad_norm": 0.5812600255012512,
+      "learning_rate": 0.0005443194819212088,
+      "loss": 3.8582,
       "step": 8700
     },
     {
-      "epoch": 0.941771606931439,
-      "grad_norm": 0.5529045462608337,
-      "learning_rate": 0.0005440857666199763,
-      "loss": 3.8314,
+      "epoch": 0.9433962264150944,
+      "grad_norm": 0.5550045371055603,
+      "learning_rate": 0.0005439956826767404,
+      "loss": 3.8556,
       "step": 8750
     },
     {
-      "epoch": 0.9471531589710472,
-      "grad_norm": 0.6090762615203857,
-      "learning_rate": 0.0005437625255899148,
-      "loss": 3.8606,
+      "epoch": 0.9487870619946092,
+      "grad_norm": 0.5164715051651001,
+      "learning_rate": 0.000543671883432272,
+      "loss": 3.858,
       "step": 8800
     },
     {
-      "epoch": 0.9525347110106555,
-      "grad_norm": 0.6306180953979492,
-      "learning_rate": 0.0005434392845598534,
-      "loss": 3.844,
+      "epoch": 0.954177897574124,
+      "grad_norm": 0.5920612215995789,
+      "learning_rate": 0.0005433480841878035,
+      "loss": 3.8529,
       "step": 8850
     },
     {
-      "epoch": 0.9579162630502637,
-      "grad_norm": 0.6025968790054321,
-      "learning_rate": 0.000543116043529792,
-      "loss": 3.853,
+      "epoch": 0.9595687331536388,
+      "grad_norm": 0.546668529510498,
+      "learning_rate": 0.000543024284943335,
+      "loss": 3.8446,
       "step": 8900
     },
     {
-      "epoch": 0.9632978150898719,
-      "grad_norm": 0.48994097113609314,
-      "learning_rate": 0.0005427928024997306,
-      "loss": 3.8427,
+      "epoch": 0.9649595687331537,
+      "grad_norm": 0.618562638759613,
+      "learning_rate": 0.0005427004856988667,
+      "loss": 3.8383,
       "step": 8950
     },
     {
-      "epoch": 0.9686793671294801,
-      "grad_norm": 0.5213954448699951,
-      "learning_rate": 0.0005424695614696692,
-      "loss": 3.8317,
+      "epoch": 0.9703504043126685,
+      "grad_norm": 0.5533706545829773,
+      "learning_rate": 0.0005423766864543982,
+      "loss": 3.8628,
       "step": 9000
     },
     {
-      "epoch": 0.9686793671294801,
-      "eval_accuracy": 0.341810141899585,
-      "eval_loss": 3.771684169769287,
-      "eval_runtime": 186.5572,
-      "eval_samples_per_second": 96.544,
-      "eval_steps_per_second": 6.036,
+      "epoch": 0.9703504043126685,
+      "eval_accuracy": 0.34025760078560385,
+      "eval_loss": 3.7860257625579834,
+      "eval_runtime": 184.3313,
+      "eval_samples_per_second": 97.71,
+      "eval_steps_per_second": 6.109,
       "step": 9000
     },
     {
-      "epoch": 0.9740609191690883,
-      "grad_norm": 0.5642579197883606,
-      "learning_rate": 0.0005421463204396078,
-      "loss": 3.8271,
+      "epoch": 0.9757412398921833,
+      "grad_norm": 0.5909618139266968,
+      "learning_rate": 0.0005420528872099298,
+      "loss": 3.8762,
       "step": 9050
     },
     {
-      "epoch": 0.9794424712086965,
-      "grad_norm": 0.5330118536949158,
-      "learning_rate": 0.0005418230794095463,
-      "loss": 3.8396,
+      "epoch": 0.9811320754716981,
+      "grad_norm": 0.5813995599746704,
+      "learning_rate": 0.0005417290879654613,
+      "loss": 3.8348,
       "step": 9100
     },
     {
-      "epoch": 0.9848240232483048,
-      "grad_norm": 0.5284038782119751,
-      "learning_rate": 0.000541499838379485,
-      "loss": 3.8295,
+      "epoch": 0.9865229110512129,
+      "grad_norm": 0.6170182824134827,
+      "learning_rate": 0.000541405288720993,
+      "loss": 3.8487,
       "step": 9150
     },
     {
-      "epoch": 0.9902055752879131,
-      "grad_norm": 0.5889008641242981,
-      "learning_rate": 0.0005411765973494235,
-      "loss": 3.8466,
+      "epoch": 0.9919137466307277,
+      "grad_norm": 0.5542091131210327,
+      "learning_rate": 0.0005410814894765245,
+      "loss": 3.8397,
       "step": 9200
     },
     {
-      "epoch": 0.9955871273275213,
-      "grad_norm": 0.5475260615348816,
-      "learning_rate": 0.0005408533563193621,
-      "loss": 3.8387,
+      "epoch": 0.9973045822102425,
+      "grad_norm": 0.5739024877548218,
+      "learning_rate": 0.0005407576902320561,
+      "loss": 3.8539,
       "step": 9250
     },
     {
-      "epoch": 1.0009686793671295,
-      "grad_norm": 0.5391682982444763,
-      "learning_rate": 0.0005405301152893007,
-      "loss": 3.8118,
+      "epoch": 1.0026954177897573,
+      "grad_norm": 0.6098498702049255,
+      "learning_rate": 0.0005404338909875876,
+      "loss": 3.8054,
       "step": 9300
     },
     {
-      "epoch": 1.0063502314067376,
-      "grad_norm": 0.5280771851539612,
-      "learning_rate": 0.0005402068742592392,
-      "loss": 3.7804,
+      "epoch": 1.0080862533692723,
+      "grad_norm": 0.5732975602149963,
+      "learning_rate": 0.0005401100917431192,
+      "loss": 3.7687,
       "step": 9350
     },
     {
-      "epoch": 1.011731783446346,
-      "grad_norm": 0.6984241008758545,
-      "learning_rate": 0.0005398836332291779,
-      "loss": 3.7839,
+      "epoch": 1.013477088948787,
+      "grad_norm": 0.5974531769752502,
+      "learning_rate": 0.0005397862924986508,
+      "loss": 3.7647,
       "step": 9400
     },
     {
-      "epoch": 1.017113335485954,
-      "grad_norm": 0.5374875068664551,
-      "learning_rate": 0.0005395603921991164,
-      "loss": 3.7713,
+      "epoch": 1.0188679245283019,
+      "grad_norm": 0.5727161765098572,
+      "learning_rate": 0.0005394624932541824,
+      "loss": 3.7626,
       "step": 9450
     },
     {
-      "epoch": 1.0224948875255624,
-      "grad_norm": 0.547314465045929,
-      "learning_rate": 0.0005392371511690551,
-      "loss": 3.7466,
+      "epoch": 1.0242587601078168,
+      "grad_norm": 0.588323712348938,
+      "learning_rate": 0.0005391386940097139,
+      "loss": 3.768,
       "step": 9500
     },
     {
-      "epoch": 1.0278764395651705,
-      "grad_norm": 0.5270159244537354,
-      "learning_rate": 0.0005389139101389936,
-      "loss": 3.7633,
+      "epoch": 1.0296495956873315,
+      "grad_norm": 0.5632597804069519,
+      "learning_rate": 0.0005388148947652455,
+      "loss": 3.7779,
       "step": 9550
     },
     {
-      "epoch": 1.0332579916047788,
-      "grad_norm": 0.5631979703903198,
-      "learning_rate": 0.0005385906691089321,
-      "loss": 3.7464,
+      "epoch": 1.0350404312668464,
+      "grad_norm": 0.6062238216400146,
+      "learning_rate": 0.000538491095520777,
+      "loss": 3.7764,
       "step": 9600
     },
     {
-      "epoch": 1.0386395436443872,
-      "grad_norm": 0.6160620450973511,
-      "learning_rate": 0.0005382674280788708,
-      "loss": 3.7739,
+      "epoch": 1.0404312668463611,
+      "grad_norm": 0.6252755522727966,
+      "learning_rate": 0.0005381672962763086,
+      "loss": 3.781,
       "step": 9650
     },
     {
-      "epoch": 1.0440210956839953,
-      "grad_norm": 0.6897755861282349,
-      "learning_rate": 0.0005379506518694106,
-      "loss": 3.7622,
+      "epoch": 1.045822102425876,
+      "grad_norm": 0.5532248020172119,
+      "learning_rate": 0.0005378434970318403,
+      "loss": 3.7632,
       "step": 9700
     },
     {
-      "epoch": 1.0494026477236036,
-      "grad_norm": 0.5001075863838196,
-      "learning_rate": 0.0005376274108393491,
-      "loss": 3.7406,
+      "epoch": 1.0512129380053907,
+      "grad_norm": 0.6396112442016602,
+      "learning_rate": 0.0005375196977873718,
+      "loss": 3.7612,
       "step": 9750
     },
     {
-      "epoch": 1.0547841997632117,
-      "grad_norm": 0.615739643573761,
-      "learning_rate": 0.0005373041698092877,
-      "loss": 3.7592,
+      "epoch": 1.0566037735849056,
+      "grad_norm": 0.5881681442260742,
+      "learning_rate": 0.0005371958985429034,
+      "loss": 3.7888,
       "step": 9800
     },
     {
-      "epoch": 1.06016575180282,
-      "grad_norm": 0.5007633566856384,
-      "learning_rate": 0.0005369809287792263,
-      "loss": 3.7521,
+      "epoch": 1.0619946091644206,
+      "grad_norm": 0.5552713871002197,
+      "learning_rate": 0.0005368720992984349,
+      "loss": 3.7903,
       "step": 9850
     },
     {
-      "epoch": 1.0655473038424281,
-      "grad_norm": 0.5301292538642883,
-      "learning_rate": 0.000536657687749165,
-      "loss": 3.7546,
+      "epoch": 1.0673854447439353,
+      "grad_norm": 0.5522390604019165,
+      "learning_rate": 0.0005365483000539665,
+      "loss": 3.7837,
       "step": 9900
     },
     {
-      "epoch": 1.0709288558820365,
-      "grad_norm": 0.5346594452857971,
-      "learning_rate": 0.0005363344467191035,
-      "loss": 3.7703,
+      "epoch": 1.0727762803234502,
+      "grad_norm": 0.5725672245025635,
+      "learning_rate": 0.0005362245008094981,
+      "loss": 3.7932,
       "step": 9950
     },
     {
-      "epoch": 1.0763104079216446,
-      "grad_norm": 0.5654758810997009,
-      "learning_rate": 0.000536011205689042,
-      "loss": 3.765,
+      "epoch": 1.0781671159029649,
+      "grad_norm": 0.5486573576927185,
+      "learning_rate": 0.0005359007015650297,
+      "loss": 3.7724,
       "step": 10000
     },
     {
-      "epoch": 1.0763104079216446,
-      "eval_accuracy": 0.3452888814087412,
-      "eval_loss": 3.7403595447540283,
-      "eval_runtime": 186.5509,
-      "eval_samples_per_second": 96.547,
-      "eval_steps_per_second": 6.036,
+      "epoch": 1.0781671159029649,
+      "eval_accuracy": 0.3436660418763441,
+      "eval_loss": 3.754814863204956,
+      "eval_runtime": 184.2442,
+      "eval_samples_per_second": 97.756,
+      "eval_steps_per_second": 6.111,
       "step": 10000
     },
     {
-      "epoch": 1.081691959961253,
-      "grad_norm": 0.5386353135108948,
-      "learning_rate": 0.0005356879646589807,
-      "loss": 3.7644,
+      "epoch": 1.0835579514824798,
+      "grad_norm": 0.5739654302597046,
+      "learning_rate": 0.0005355769023205612,
+      "loss": 3.7649,
       "step": 10050
     },
     {
-      "epoch": 1.087073512000861,
-      "grad_norm": 0.6020763516426086,
-      "learning_rate": 0.0005353647236289192,
-      "loss": 3.7552,
+      "epoch": 1.0889487870619945,
+      "grad_norm": 0.6266982555389404,
+      "learning_rate": 0.0005352531030760928,
+      "loss": 3.7678,
       "step": 10100
     },
     {
-      "epoch": 1.0924550640404693,
-      "grad_norm": 0.5746710300445557,
-      "learning_rate": 0.0005350414825988579,
-      "loss": 3.7718,
+      "epoch": 1.0943396226415094,
+      "grad_norm": 0.5478909015655518,
+      "learning_rate": 0.0005349293038316244,
+      "loss": 3.7609,
       "step": 10150
     },
     {
-      "epoch": 1.0978366160800774,
-      "grad_norm": 0.6303889751434326,
-      "learning_rate": 0.0005347182415687964,
-      "loss": 3.7454,
+      "epoch": 1.0997304582210243,
+      "grad_norm": 0.6404117345809937,
+      "learning_rate": 0.0005346055045871559,
+      "loss": 3.7746,
       "step": 10200
     },
     {
-      "epoch": 1.1032181681196858,
-      "grad_norm": 0.5959435701370239,
-      "learning_rate": 0.000534395000538735,
-      "loss": 3.758,
+      "epoch": 1.105121293800539,
+      "grad_norm": 0.5925765037536621,
+      "learning_rate": 0.0005342817053426874,
+      "loss": 3.7638,
       "step": 10250
     },
     {
-      "epoch": 1.1085997201592939,
-      "grad_norm": 0.5945746302604675,
-      "learning_rate": 0.0005340717595086736,
-      "loss": 3.7573,
+      "epoch": 1.110512129380054,
+      "grad_norm": 0.6467453241348267,
+      "learning_rate": 0.0005339579060982191,
+      "loss": 3.7705,
       "step": 10300
     },
     {
-      "epoch": 1.1139812721989022,
-      "grad_norm": 0.5508937239646912,
-      "learning_rate": 0.0005337485184786122,
-      "loss": 3.7403,
+      "epoch": 1.1159029649595686,
+      "grad_norm": 0.5718551278114319,
+      "learning_rate": 0.0005336341068537506,
+      "loss": 3.7647,
       "step": 10350
     },
     {
-      "epoch": 1.1193628242385103,
-      "grad_norm": 0.6239654421806335,
-      "learning_rate": 0.0005334252774485507,
-      "loss": 3.7508,
+      "epoch": 1.1212938005390836,
+      "grad_norm": 0.5561963319778442,
+      "learning_rate": 0.0005333103076092822,
+      "loss": 3.7701,
       "step": 10400
     },
     {
-      "epoch": 1.1247443762781186,
-      "grad_norm": 0.5784809589385986,
-      "learning_rate": 0.0005331020364184894,
-      "loss": 3.7482,
+      "epoch": 1.1266846361185983,
+      "grad_norm": 0.5496252179145813,
+      "learning_rate": 0.0005329865083648137,
+      "loss": 3.7685,
       "step": 10450
     },
     {
-      "epoch": 1.1301259283177267,
-      "grad_norm": 0.5604040622711182,
-      "learning_rate": 0.0005327787953884279,
-      "loss": 3.7481,
+      "epoch": 1.1320754716981132,
+      "grad_norm": 0.5853784680366516,
+      "learning_rate": 0.0005326627091203454,
+      "loss": 3.7707,
       "step": 10500
     },
     {
-      "epoch": 1.135507480357335,
-      "grad_norm": 0.6120591759681702,
-      "learning_rate": 0.0005324555543583665,
-      "loss": 3.7478,
+      "epoch": 1.137466307277628,
+      "grad_norm": 0.6111279726028442,
+      "learning_rate": 0.0005323389098758769,
+      "loss": 3.7752,
       "step": 10550
     },
     {
-      "epoch": 1.1408890323969434,
-      "grad_norm": 0.5750361680984497,
-      "learning_rate": 0.0005321323133283051,
-      "loss": 3.7274,
+      "epoch": 1.1428571428571428,
+      "grad_norm": 0.6025037169456482,
+      "learning_rate": 0.0005320151106314085,
+      "loss": 3.7647,
       "step": 10600
     },
     {
-      "epoch": 1.1462705844365515,
-      "grad_norm": 0.5619457960128784,
-      "learning_rate": 0.0005318090722982436,
-      "loss": 3.752,
+      "epoch": 1.1482479784366577,
+      "grad_norm": 0.6000508069992065,
+      "learning_rate": 0.00053169131138694,
+      "loss": 3.7485,
       "step": 10650
     },
     {
-      "epoch": 1.1516521364761596,
-      "grad_norm": 0.5420694947242737,
-      "learning_rate": 0.0005314858312681823,
-      "loss": 3.7496,
+      "epoch": 1.1536388140161726,
+      "grad_norm": 0.5954493880271912,
+      "learning_rate": 0.0005313675121424716,
+      "loss": 3.7614,
       "step": 10700
     },
     {
-      "epoch": 1.157033688515768,
-      "grad_norm": 0.5764556527137756,
-      "learning_rate": 0.0005311625902381209,
-      "loss": 3.7556,
+      "epoch": 1.1590296495956873,
+      "grad_norm": 0.5783281922340393,
+      "learning_rate": 0.0005310437128980032,
+      "loss": 3.7429,
       "step": 10750
     },
     {
-      "epoch": 1.1624152405553763,
-      "grad_norm": 0.548437237739563,
-      "learning_rate": 0.0005308393492080595,
-      "loss": 3.7368,
+      "epoch": 1.1644204851752022,
+      "grad_norm": 0.609627366065979,
+      "learning_rate": 0.0005307199136535348,
+      "loss": 3.7474,
       "step": 10800
     },
     {
-      "epoch": 1.1677967925949844,
-      "grad_norm": 0.5925754308700562,
-      "learning_rate": 0.000530516108177998,
-      "loss": 3.7335,
+      "epoch": 1.169811320754717,
+      "grad_norm": 0.5700644254684448,
+      "learning_rate": 0.0005304025903939556,
+      "loss": 3.7391,
       "step": 10850
     },
     {
-      "epoch": 1.1731783446345927,
-      "grad_norm": 0.5371476411819458,
-      "learning_rate": 0.0005301928671479365,
-      "loss": 3.7456,
+      "epoch": 1.1752021563342319,
+      "grad_norm": 0.5731554627418518,
+      "learning_rate": 0.0005300787911494873,
+      "loss": 3.753,
       "step": 10900
     },
     {
-      "epoch": 1.1785598966742008,
-      "grad_norm": 0.5599814653396606,
-      "learning_rate": 0.0005298696261178752,
-      "loss": 3.737,
+      "epoch": 1.1805929919137466,
+      "grad_norm": 0.538571834564209,
+      "learning_rate": 0.0005297549919050189,
+      "loss": 3.7667,
       "step": 10950
     },
     {
-      "epoch": 1.1839414487138091,
-      "grad_norm": 0.5617986917495728,
-      "learning_rate": 0.0005295463850878138,
-      "loss": 3.7212,
+      "epoch": 1.1859838274932615,
+      "grad_norm": 0.5856531262397766,
+      "learning_rate": 0.0005294311926605504,
+      "loss": 3.7772,
       "step": 11000
     },
     {
-      "epoch": 1.1839414487138091,
-      "eval_accuracy": 0.3478470049994453,
-      "eval_loss": 3.718867540359497,
-      "eval_runtime": 186.4225,
-      "eval_samples_per_second": 96.614,
-      "eval_steps_per_second": 6.04,
+      "epoch": 1.1859838274932615,
+      "eval_accuracy": 0.34641843683046053,
+      "eval_loss": 3.7294209003448486,
+      "eval_runtime": 184.3558,
+      "eval_samples_per_second": 97.697,
+      "eval_steps_per_second": 6.108,
       "step": 11000
     },
     {
-      "epoch": 1.1893230007534172,
-      "grad_norm": 0.6061155200004578,
-      "learning_rate": 0.0005292231440577524,
-      "loss": 3.7606,
+      "epoch": 1.1913746630727764,
+      "grad_norm": 0.5894326567649841,
+      "learning_rate": 0.000529107393416082,
+      "loss": 3.7635,
       "step": 11050
     },
     {
-      "epoch": 1.1947045527930256,
-      "grad_norm": 0.57025545835495,
-      "learning_rate": 0.0005288999030276909,
-      "loss": 3.7587,
+      "epoch": 1.196765498652291,
+      "grad_norm": 0.5577927231788635,
+      "learning_rate": 0.0005287835941716135,
+      "loss": 3.7489,
       "step": 11100
     },
     {
-      "epoch": 1.2000861048326337,
-      "grad_norm": 0.594207763671875,
-      "learning_rate": 0.0005285766619976295,
-      "loss": 3.733,
+      "epoch": 1.202156334231806,
+      "grad_norm": 0.5895690321922302,
+      "learning_rate": 0.0005284597949271452,
+      "loss": 3.7468,
       "step": 11150
     },
     {
-      "epoch": 1.205467656872242,
-      "grad_norm": 0.5179318785667419,
-      "learning_rate": 0.0005282534209675681,
-      "loss": 3.7304,
+      "epoch": 1.2075471698113207,
+      "grad_norm": 0.5373809933662415,
+      "learning_rate": 0.0005281359956826767,
+      "loss": 3.7701,
       "step": 11200
     },
     {
-      "epoch": 1.21084920891185,
-      "grad_norm": 0.5588968396186829,
-      "learning_rate": 0.0005279301799375066,
-      "loss": 3.7313,
+      "epoch": 1.2129380053908356,
+      "grad_norm": 0.5922164916992188,
+      "learning_rate": 0.0005278121964382083,
+      "loss": 3.7576,
       "step": 11250
     },
     {
-      "epoch": 1.2162307609514584,
-      "grad_norm": 0.6667935252189636,
-      "learning_rate": 0.0005276069389074453,
-      "loss": 3.7395,
+      "epoch": 1.2183288409703503,
+      "grad_norm": 0.5997561812400818,
+      "learning_rate": 0.0005274883971937398,
+      "loss": 3.7436,
       "step": 11300
     },
     {
-      "epoch": 1.2216123129910665,
-      "grad_norm": 0.5639368295669556,
-      "learning_rate": 0.0005272836978773838,
-      "loss": 3.7491,
+      "epoch": 1.2237196765498652,
+      "grad_norm": 0.5441973805427551,
+      "learning_rate": 0.0005271645979492714,
+      "loss": 3.7388,
       "step": 11350
     },
     {
-      "epoch": 1.2269938650306749,
-      "grad_norm": 0.6160590052604675,
-      "learning_rate": 0.0005269604568473225,
-      "loss": 3.7305,
+      "epoch": 1.2291105121293802,
+      "grad_norm": 0.7957209944725037,
+      "learning_rate": 0.0005268472746896923,
+      "loss": 3.755,
       "step": 11400
     },
     {
-      "epoch": 1.232375417070283,
-      "grad_norm": 0.5744423866271973,
-      "learning_rate": 0.000526637215817261,
-      "loss": 3.7358,
+      "epoch": 1.2345013477088949,
+      "grad_norm": 0.534511923789978,
+      "learning_rate": 0.000526523475445224,
+      "loss": 3.7653,
       "step": 11450
     },
     {
-      "epoch": 1.2377569691098913,
-      "grad_norm": 0.5147489309310913,
-      "learning_rate": 0.0005263139747871996,
-      "loss": 3.7276,
+      "epoch": 1.2398921832884098,
+      "grad_norm": 0.5863783359527588,
+      "learning_rate": 0.0005261996762007554,
+      "loss": 3.7513,
       "step": 11500
     },
     {
-      "epoch": 1.2431385211494996,
-      "grad_norm": 0.5641859769821167,
-      "learning_rate": 0.0005259907337571381,
-      "loss": 3.7207,
+      "epoch": 1.2452830188679245,
+      "grad_norm": 0.5853481888771057,
+      "learning_rate": 0.0005258758769562871,
+      "loss": 3.7474,
       "step": 11550
     },
     {
-      "epoch": 1.2485200731891077,
-      "grad_norm": 0.5240283608436584,
-      "learning_rate": 0.0005256674927270768,
-      "loss": 3.7198,
+      "epoch": 1.2506738544474394,
+      "grad_norm": 0.5571786761283875,
+      "learning_rate": 0.0005255520777118186,
+      "loss": 3.7592,
       "step": 11600
     },
     {
-      "epoch": 1.2539016252287158,
-      "grad_norm": 0.5902089476585388,
-      "learning_rate": 0.0005253442516970154,
-      "loss": 3.7365,
+      "epoch": 1.256064690026954,
+      "grad_norm": 0.5264196395874023,
+      "learning_rate": 0.0005252282784673502,
+      "loss": 3.7395,
       "step": 11650
     },
     {
-      "epoch": 1.2592831772683242,
-      "grad_norm": 0.5724839568138123,
-      "learning_rate": 0.000525021010666954,
-      "loss": 3.7365,
+      "epoch": 1.261455525606469,
+      "grad_norm": 0.5181690454483032,
+      "learning_rate": 0.0005249044792228817,
+      "loss": 3.7494,
       "step": 11700
     },
     {
-      "epoch": 1.2646647293079325,
-      "grad_norm": 0.6671346426010132,
-      "learning_rate": 0.0005246977696368925,
-      "loss": 3.7442,
+      "epoch": 1.266846361185984,
+      "grad_norm": 0.5544186234474182,
+      "learning_rate": 0.0005245806799784133,
+      "loss": 3.75,
       "step": 11750
     },
     {
-      "epoch": 1.2700462813475406,
-      "grad_norm": 0.5624138712882996,
-      "learning_rate": 0.0005243809934274323,
-      "loss": 3.7269,
+      "epoch": 1.2722371967654986,
+      "grad_norm": 0.6025522947311401,
+      "learning_rate": 0.0005242568807339449,
+      "loss": 3.764,
       "step": 11800
     },
     {
-      "epoch": 1.275427833387149,
-      "grad_norm": 0.5766282677650452,
-      "learning_rate": 0.0005240577523973709,
-      "loss": 3.7126,
+      "epoch": 1.2776280323450135,
+      "grad_norm": 0.5848909020423889,
+      "learning_rate": 0.0005239330814894765,
+      "loss": 3.7459,
       "step": 11850
     },
     {
-      "epoch": 1.280809385426757,
-      "grad_norm": 0.6318398714065552,
-      "learning_rate": 0.0005237345113673095,
-      "loss": 3.7123,
+      "epoch": 1.2830188679245282,
+      "grad_norm": 0.5673113465309143,
+      "learning_rate": 0.000523609282245008,
+      "loss": 3.7416,
       "step": 11900
     },
     {
-      "epoch": 1.2861909374663654,
-      "grad_norm": 0.5170403122901917,
-      "learning_rate": 0.0005234112703372481,
-      "loss": 3.7201,
+      "epoch": 1.2884097035040432,
+      "grad_norm": 0.6251968741416931,
+      "learning_rate": 0.0005232854830005396,
+      "loss": 3.7245,
       "step": 11950
     },
     {
-      "epoch": 1.2915724895059735,
-      "grad_norm": 0.5587390065193176,
-      "learning_rate": 0.0005230880293071867,
-      "loss": 3.7228,
+      "epoch": 1.2938005390835579,
+      "grad_norm": 0.5748421549797058,
+      "learning_rate": 0.0005229616837560712,
+      "loss": 3.7384,
       "step": 12000
     },
     {
-      "epoch": 1.2915724895059735,
-      "eval_accuracy": 0.3496730254320555,
-      "eval_loss": 3.6965837478637695,
-      "eval_runtime": 186.4424,
-      "eval_samples_per_second": 96.604,
-      "eval_steps_per_second": 6.039,
+      "epoch": 1.2938005390835579,
+      "eval_accuracy": 0.34850728859869395,
+      "eval_loss": 3.7043049335479736,
+      "eval_runtime": 184.3362,
+      "eval_samples_per_second": 97.707,
+      "eval_steps_per_second": 6.108,
       "step": 12000
     },
     {
-      "epoch": 1.2969540415455818,
-      "grad_norm": 0.6743207573890686,
-      "learning_rate": 0.0005227647882771253,
-      "loss": 3.7156,
+      "epoch": 1.2991913746630728,
+      "grad_norm": 0.5955175161361694,
+      "learning_rate": 0.0005226378845116028,
+      "loss": 3.7529,
       "step": 12050
     },
     {
-      "epoch": 1.30233559358519,
-      "grad_norm": 0.5941646099090576,
-      "learning_rate": 0.0005224415472470639,
-      "loss": 3.7279,
+      "epoch": 1.3045822102425877,
+      "grad_norm": 0.6381511092185974,
+      "learning_rate": 0.0005223140852671344,
+      "loss": 3.7469,
       "step": 12100
     },
     {
-      "epoch": 1.3077171456247982,
-      "grad_norm": 0.5375372171401978,
-      "learning_rate": 0.0005221183062170024,
-      "loss": 3.7188,
+      "epoch": 1.3099730458221024,
+      "grad_norm": 0.5906614661216736,
+      "learning_rate": 0.0005219902860226659,
+      "loss": 3.7366,
       "step": 12150
     },
     {
-      "epoch": 1.3130986976644063,
-      "grad_norm": 0.5272995829582214,
-      "learning_rate": 0.0005217950651869409,
-      "loss": 3.7284,
+      "epoch": 1.3153638814016173,
+      "grad_norm": 0.5654398202896118,
+      "learning_rate": 0.0005216664867781975,
+      "loss": 3.7518,
       "step": 12200
     },
     {
-      "epoch": 1.3184802497040147,
-      "grad_norm": 0.5427494645118713,
-      "learning_rate": 0.0005214718241568796,
-      "loss": 3.742,
+      "epoch": 1.320754716981132,
+      "grad_norm": 0.612305223941803,
+      "learning_rate": 0.000521342687533729,
+      "loss": 3.727,
       "step": 12250
     },
     {
-      "epoch": 1.3238618017436228,
-      "grad_norm": 0.5961765646934509,
-      "learning_rate": 0.0005211485831268182,
-      "loss": 3.6971,
+      "epoch": 1.326145552560647,
+      "grad_norm": 0.6109020113945007,
+      "learning_rate": 0.0005210188882892606,
+      "loss": 3.734,
       "step": 12300
     },
     {
-      "epoch": 1.329243353783231,
-      "grad_norm": 0.555831253528595,
-      "learning_rate": 0.0005208253420967568,
-      "loss": 3.7254,
+      "epoch": 1.3315363881401616,
+      "grad_norm": 0.5165815353393555,
+      "learning_rate": 0.0005206950890447922,
+      "loss": 3.7426,
       "step": 12350
     },
     {
-      "epoch": 1.3346249058228392,
-      "grad_norm": 0.5901808738708496,
-      "learning_rate": 0.0005205021010666953,
-      "loss": 3.7258,
+      "epoch": 1.3369272237196765,
+      "grad_norm": 0.6351564526557922,
+      "learning_rate": 0.0005203712898003238,
+      "loss": 3.7399,
       "step": 12400
     },
     {
-      "epoch": 1.3400064578624475,
-      "grad_norm": 0.5180104374885559,
-      "learning_rate": 0.0005201788600366339,
-      "loss": 3.737,
+      "epoch": 1.3423180592991915,
+      "grad_norm": 0.5947250723838806,
+      "learning_rate": 0.0005200474905558553,
+      "loss": 3.7293,
       "step": 12450
     },
     {
-      "epoch": 1.3453880099020559,
-      "grad_norm": 0.5261190533638,
-      "learning_rate": 0.0005198556190065725,
-      "loss": 3.7211,
+      "epoch": 1.3477088948787062,
+      "grad_norm": 0.536835789680481,
+      "learning_rate": 0.0005197236913113869,
+      "loss": 3.7491,
       "step": 12500
     },
     {
-      "epoch": 1.350769561941664,
-      "grad_norm": 0.570330023765564,
-      "learning_rate": 0.0005195323779765112,
-      "loss": 3.7421,
+      "epoch": 1.353099730458221,
+      "grad_norm": 0.5466864705085754,
+      "learning_rate": 0.0005193998920669184,
+      "loss": 3.7195,
       "step": 12550
     },
     {
-      "epoch": 1.356151113981272,
-      "grad_norm": 0.5671786069869995,
-      "learning_rate": 0.0005192091369464497,
-      "loss": 3.7245,
+      "epoch": 1.3584905660377358,
+      "grad_norm": 0.6107433438301086,
+      "learning_rate": 0.0005190760928224501,
+      "loss": 3.7208,
       "step": 12600
     },
     {
-      "epoch": 1.3615326660208804,
-      "grad_norm": 0.5568457245826721,
-      "learning_rate": 0.0005188858959163882,
-      "loss": 3.7191,
+      "epoch": 1.3638814016172507,
+      "grad_norm": 0.5991501808166504,
+      "learning_rate": 0.0005187522935779816,
+      "loss": 3.7102,
       "step": 12650
     },
     {
-      "epoch": 1.3669142180604887,
-      "grad_norm": 0.5448923707008362,
-      "learning_rate": 0.0005185626548863269,
-      "loss": 3.7105,
+      "epoch": 1.3692722371967654,
+      "grad_norm": 0.5714552402496338,
+      "learning_rate": 0.0005184284943335132,
+      "loss": 3.7246,
       "step": 12700
     },
     {
-      "epoch": 1.3722957701000968,
-      "grad_norm": 0.5400291085243225,
-      "learning_rate": 0.0005182394138562654,
-      "loss": 3.7091,
+      "epoch": 1.3746630727762803,
+      "grad_norm": 0.5692049860954285,
+      "learning_rate": 0.0005181046950890447,
+      "loss": 3.7156,
       "step": 12750
     },
     {
-      "epoch": 1.3776773221397052,
-      "grad_norm": 0.5717939734458923,
-      "learning_rate": 0.0005179161728262041,
-      "loss": 3.7085,
+      "epoch": 1.3800539083557952,
+      "grad_norm": 0.5899141430854797,
+      "learning_rate": 0.0005177808958445764,
+      "loss": 3.7167,
       "step": 12800
     },
     {
-      "epoch": 1.3830588741793133,
-      "grad_norm": 0.6186015009880066,
-      "learning_rate": 0.0005175929317961426,
-      "loss": 3.7125,
+      "epoch": 1.38544474393531,
+      "grad_norm": 0.6727328896522522,
+      "learning_rate": 0.0005174570966001078,
+      "loss": 3.7269,
       "step": 12850
     },
     {
-      "epoch": 1.3884404262189216,
-      "grad_norm": 0.5692641139030457,
-      "learning_rate": 0.0005172696907660812,
-      "loss": 3.7134,
+      "epoch": 1.3908355795148248,
+      "grad_norm": 0.6033250689506531,
+      "learning_rate": 0.0005171332973556395,
+      "loss": 3.7255,
       "step": 12900
     },
     {
-      "epoch": 1.3938219782585297,
-      "grad_norm": 0.6157675385475159,
-      "learning_rate": 0.0005169464497360198,
-      "loss": 3.7023,
+      "epoch": 1.3962264150943398,
+      "grad_norm": 0.534835159778595,
+      "learning_rate": 0.000516809498111171,
+      "loss": 3.7309,
       "step": 12950
     },
     {
-      "epoch": 1.399203530298138,
-      "grad_norm": 0.5926107168197632,
-      "learning_rate": 0.0005166232087059583,
-      "loss": 3.7021,
+      "epoch": 1.4016172506738545,
+      "grad_norm": 0.5884051322937012,
+      "learning_rate": 0.0005164856988667026,
+      "loss": 3.7141,
       "step": 13000
     },
     {
-      "epoch": 1.399203530298138,
-      "eval_accuracy": 0.3519149691192195,
-      "eval_loss": 3.6726531982421875,
-      "eval_runtime": 186.754,
-      "eval_samples_per_second": 96.442,
-      "eval_steps_per_second": 6.029,
+      "epoch": 1.4016172506738545,
+      "eval_accuracy": 0.3509202519312779,
+      "eval_loss": 3.680180788040161,
+      "eval_runtime": 184.2686,
+      "eval_samples_per_second": 97.743,
+      "eval_steps_per_second": 6.111,
       "step": 13000
     },
     {
-      "epoch": 1.4045850823377461,
-      "grad_norm": 0.5519713759422302,
-      "learning_rate": 0.0005162999676758969,
-      "loss": 3.7302,
+      "epoch": 1.4070080862533692,
+      "grad_norm": 0.6032182574272156,
+      "learning_rate": 0.0005161618996222341,
+      "loss": 3.7128,
       "step": 13050
     },
     {
-      "epoch": 1.4099666343773545,
-      "grad_norm": 0.5651559233665466,
-      "learning_rate": 0.0005159767266458355,
-      "loss": 3.6969,
+      "epoch": 1.412398921832884,
+      "grad_norm": 0.591998279094696,
+      "learning_rate": 0.0005158381003777657,
+      "loss": 3.709,
       "step": 13100
     },
     {
-      "epoch": 1.4153481864169626,
-      "grad_norm": 0.6367672681808472,
-      "learning_rate": 0.0005156534856157741,
-      "loss": 3.6836,
+      "epoch": 1.417789757412399,
+      "grad_norm": 0.5996487140655518,
+      "learning_rate": 0.0005155143011332973,
+      "loss": 3.7142,
       "step": 13150
     },
     {
-      "epoch": 1.420729738456571,
-      "grad_norm": 0.5546962022781372,
-      "learning_rate": 0.0005153302445857127,
-      "loss": 3.6885,
+      "epoch": 1.4231805929919137,
+      "grad_norm": 0.5706505179405212,
+      "learning_rate": 0.0005151905018888289,
+      "loss": 3.7135,
       "step": 13200
     },
     {
-      "epoch": 1.426111290496179,
-      "grad_norm": 0.5732678771018982,
-      "learning_rate": 0.0005150070035556513,
-      "loss": 3.6982,
+      "epoch": 1.4285714285714286,
+      "grad_norm": 0.5638548135757446,
+      "learning_rate": 0.0005148667026443604,
+      "loss": 3.7247,
       "step": 13250
     },
     {
-      "epoch": 1.4314928425357873,
-      "grad_norm": 0.5779805183410645,
-      "learning_rate": 0.0005146837625255898,
-      "loss": 3.691,
+      "epoch": 1.4339622641509435,
+      "grad_norm": 0.5824941992759705,
+      "learning_rate": 0.000514542903399892,
+      "loss": 3.6962,
       "step": 13300
     },
     {
-      "epoch": 1.4368743945753955,
-      "grad_norm": 0.5563697814941406,
-      "learning_rate": 0.0005143605214955285,
-      "loss": 3.6854,
+      "epoch": 1.4393530997304582,
+      "grad_norm": 0.5614493489265442,
+      "learning_rate": 0.0005142191041554237,
+      "loss": 3.7153,
       "step": 13350
     },
     {
-      "epoch": 1.4422559466150038,
-      "grad_norm": 0.5216091871261597,
-      "learning_rate": 0.0005140372804654671,
-      "loss": 3.6785,
+      "epoch": 1.444743935309973,
+      "grad_norm": 0.578725278377533,
+      "learning_rate": 0.0005138953049109552,
+      "loss": 3.7308,
       "step": 13400
     },
     {
-      "epoch": 1.447637498654612,
-      "grad_norm": 0.5843927264213562,
-      "learning_rate": 0.0005137140394354056,
-      "loss": 3.7227,
+      "epoch": 1.4501347708894878,
+      "grad_norm": 0.5532973408699036,
+      "learning_rate": 0.0005135715056664868,
+      "loss": 3.7349,
       "step": 13450
     },
     {
-      "epoch": 1.4530190506942202,
-      "grad_norm": 0.5472906231880188,
-      "learning_rate": 0.0005133907984053442,
-      "loss": 3.6968,
+      "epoch": 1.4555256064690028,
+      "grad_norm": 0.6558372974395752,
+      "learning_rate": 0.0005132477064220183,
+      "loss": 3.726,
       "step": 13500
     },
     {
-      "epoch": 1.4584006027338283,
-      "grad_norm": 0.5616341233253479,
-      "learning_rate": 0.0005130675573752827,
-      "loss": 3.6979,
+      "epoch": 1.4609164420485174,
+      "grad_norm": 0.5940355062484741,
+      "learning_rate": 0.0005129239071775499,
+      "loss": 3.7144,
       "step": 13550
     },
     {
-      "epoch": 1.4637821547734367,
-      "grad_norm": 0.5248051285743713,
-      "learning_rate": 0.0005127443163452214,
-      "loss": 3.6886,
+      "epoch": 1.4663072776280324,
+      "grad_norm": 0.6997793316841125,
+      "learning_rate": 0.0005126001079330814,
+      "loss": 3.7325,
       "step": 13600
     },
     {
-      "epoch": 1.469163706813045,
-      "grad_norm": 0.5378375053405762,
-      "learning_rate": 0.00051242107531516,
-      "loss": 3.6987,
+      "epoch": 1.4716981132075473,
+      "grad_norm": 0.5935999751091003,
+      "learning_rate": 0.000512276308688613,
+      "loss": 3.7205,
       "step": 13650
     },
     {
-      "epoch": 1.474545258852653,
-      "grad_norm": 0.6293401718139648,
-      "learning_rate": 0.0005120978342850986,
-      "loss": 3.6921,
+      "epoch": 1.477088948787062,
+      "grad_norm": 0.6040553450584412,
+      "learning_rate": 0.0005119525094441446,
+      "loss": 3.7059,
       "step": 13700
     },
     {
-      "epoch": 1.4799268108922612,
-      "grad_norm": 0.5358719825744629,
-      "learning_rate": 0.0005117745932550371,
-      "loss": 3.7102,
+      "epoch": 1.482479784366577,
+      "grad_norm": 0.5976484417915344,
+      "learning_rate": 0.0005116287101996762,
+      "loss": 3.709,
       "step": 13750
     },
     {
-      "epoch": 1.4853083629318695,
-      "grad_norm": 0.5491902232170105,
-      "learning_rate": 0.0005114513522249758,
-      "loss": 3.7058,
+      "epoch": 1.4878706199460916,
+      "grad_norm": 0.5414652824401855,
+      "learning_rate": 0.0005113049109552077,
+      "loss": 3.7117,
       "step": 13800
     },
     {
-      "epoch": 1.4906899149714778,
-      "grad_norm": 0.5653431415557861,
-      "learning_rate": 0.0005111281111949143,
-      "loss": 3.6886,
+      "epoch": 1.4932614555256065,
+      "grad_norm": 0.5687587857246399,
+      "learning_rate": 0.0005109811117107393,
+      "loss": 3.7153,
       "step": 13850
     },
     {
-      "epoch": 1.496071467011086,
-      "grad_norm": 0.5465991497039795,
-      "learning_rate": 0.0005108048701648528,
-      "loss": 3.6838,
+      "epoch": 1.4986522911051212,
+      "grad_norm": 0.6240153312683105,
+      "learning_rate": 0.0005106573124662708,
+      "loss": 3.6957,
       "step": 13900
     },
     {
-      "epoch": 1.501453019050694,
-      "grad_norm": 0.5228298306465149,
-      "learning_rate": 0.0005104816291347915,
-      "loss": 3.6874,
+      "epoch": 1.5040431266846361,
+      "grad_norm": 0.6018204092979431,
+      "learning_rate": 0.0005103335132218025,
+      "loss": 3.6976,
       "step": 13950
     },
     {
-      "epoch": 1.5068345710903024,
-      "grad_norm": 0.5415006875991821,
-      "learning_rate": 0.0005101648529253313,
-      "loss": 3.6887,
+      "epoch": 1.509433962264151,
+      "grad_norm": 0.5469267964363098,
+      "learning_rate": 0.000510009713977334,
+      "loss": 3.7107,
       "step": 14000
     },
     {
-      "epoch": 1.5068345710903024,
-      "eval_accuracy": 0.3541161679734531,
-      "eval_loss": 3.6511662006378174,
-      "eval_runtime": 186.4847,
-      "eval_samples_per_second": 96.582,
-      "eval_steps_per_second": 6.038,
+      "epoch": 1.509433962264151,
+      "eval_accuracy": 0.35278582201505254,
+      "eval_loss": 3.6626014709472656,
+      "eval_runtime": 183.9793,
+      "eval_samples_per_second": 97.897,
+      "eval_steps_per_second": 6.12,
       "step": 14000
     },
     {
-      "epoch": 1.5122161231299107,
-      "grad_norm": 0.5534867644309998,
-      "learning_rate": 0.0005098416118952699,
-      "loss": 3.6952,
+      "epoch": 1.5148247978436657,
+      "grad_norm": 0.8756197094917297,
+      "learning_rate": 0.0005096859147328656,
+      "loss": 3.7094,
       "step": 14050
     },
     {
-      "epoch": 1.5175976751695188,
-      "grad_norm": 0.5454338192939758,
-      "learning_rate": 0.0005095183708652085,
-      "loss": 3.6986,
+      "epoch": 1.5202156334231804,
+      "grad_norm": 0.5950367450714111,
+      "learning_rate": 0.0005093621154883971,
+      "loss": 3.7092,
       "step": 14100
     },
     {
-      "epoch": 1.5229792272091272,
-      "grad_norm": 0.5306920409202576,
-      "learning_rate": 0.000509195129835147,
-      "loss": 3.6781,
+      "epoch": 1.5256064690026954,
+      "grad_norm": 0.5225749015808105,
+      "learning_rate": 0.0005090383162439288,
+      "loss": 3.7051,
       "step": 14150
     },
     {
-      "epoch": 1.5283607792487355,
-      "grad_norm": 0.5203532576560974,
-      "learning_rate": 0.0005088718888050856,
-      "loss": 3.6771,
+      "epoch": 1.5309973045822103,
+      "grad_norm": 0.5934370160102844,
+      "learning_rate": 0.0005087145169994602,
+      "loss": 3.7092,
       "step": 14200
     },
     {
-      "epoch": 1.5337423312883436,
-      "grad_norm": 0.545259952545166,
-      "learning_rate": 0.0005085486477750242,
-      "loss": 3.6875,
+      "epoch": 1.536388140161725,
+      "grad_norm": 0.6141777634620667,
+      "learning_rate": 0.0005083907177549918,
+      "loss": 3.7076,
       "step": 14250
     },
     {
-      "epoch": 1.5391238833279517,
-      "grad_norm": 0.5470886826515198,
-      "learning_rate": 0.0005082254067449629,
-      "loss": 3.6841,
+      "epoch": 1.54177897574124,
+      "grad_norm": 0.531351625919342,
+      "learning_rate": 0.0005080669185105234,
+      "loss": 3.7042,
       "step": 14300
     },
     {
-      "epoch": 1.54450543536756,
-      "grad_norm": 0.5590439438819885,
-      "learning_rate": 0.0005079021657149014,
-      "loss": 3.6796,
+      "epoch": 1.5471698113207548,
+      "grad_norm": 0.5487028360366821,
+      "learning_rate": 0.000507743119266055,
+      "loss": 3.6947,
       "step": 14350
     },
     {
-      "epoch": 1.5498869874071683,
-      "grad_norm": 0.553065836429596,
-      "learning_rate": 0.0005075789246848399,
-      "loss": 3.6749,
+      "epoch": 1.5525606469002695,
+      "grad_norm": 0.5891349911689758,
+      "learning_rate": 0.0005074193200215865,
+      "loss": 3.6973,
       "step": 14400
     },
     {
-      "epoch": 1.5552685394467765,
-      "grad_norm": 0.6136184334754944,
-      "learning_rate": 0.0005072556836547785,
-      "loss": 3.6851,
+      "epoch": 1.5579514824797842,
+      "grad_norm": 0.5347645282745361,
+      "learning_rate": 0.0005070955207771181,
+      "loss": 3.6999,
       "step": 14450
     },
     {
-      "epoch": 1.5606500914863846,
-      "grad_norm": 0.5715840458869934,
-      "learning_rate": 0.0005069324426247171,
-      "loss": 3.6904,
+      "epoch": 1.5633423180592994,
+      "grad_norm": 0.5652821063995361,
+      "learning_rate": 0.0005067717215326498,
+      "loss": 3.6993,
       "step": 14500
     },
     {
-      "epoch": 1.566031643525993,
-      "grad_norm": 0.5454801321029663,
-      "learning_rate": 0.0005066092015946557,
-      "loss": 3.6801,
+      "epoch": 1.568733153638814,
+      "grad_norm": 0.5672808885574341,
+      "learning_rate": 0.0005064479222881813,
+      "loss": 3.6977,
       "step": 14550
     },
     {
-      "epoch": 1.5714131955656012,
-      "grad_norm": 0.5784483551979065,
-      "learning_rate": 0.0005062859605645943,
-      "loss": 3.6955,
+      "epoch": 1.5741239892183287,
+      "grad_norm": 0.576740562915802,
+      "learning_rate": 0.0005061241230437129,
+      "loss": 3.6831,
       "step": 14600
     },
     {
-      "epoch": 1.5767947476052093,
-      "grad_norm": 0.5951915383338928,
-      "learning_rate": 0.0005059627195345329,
-      "loss": 3.6899,
+      "epoch": 1.5795148247978437,
+      "grad_norm": 0.5609301924705505,
+      "learning_rate": 0.0005058003237992444,
+      "loss": 3.6872,
       "step": 14650
     },
     {
-      "epoch": 1.5821762996448174,
-      "grad_norm": 0.5582926869392395,
-      "learning_rate": 0.0005056394785044715,
-      "loss": 3.6662,
+      "epoch": 1.5849056603773586,
+      "grad_norm": 0.5302152037620544,
+      "learning_rate": 0.000505476524554776,
+      "loss": 3.6938,
       "step": 14700
     },
     {
-      "epoch": 1.5875578516844258,
-      "grad_norm": 0.5636453628540039,
-      "learning_rate": 0.00050531623747441,
-      "loss": 3.6812,
+      "epoch": 1.5902964959568733,
+      "grad_norm": 0.6070558428764343,
+      "learning_rate": 0.0005051527253103076,
+      "loss": 3.6747,
       "step": 14750
     },
     {
-      "epoch": 1.592939403724034,
-      "grad_norm": 0.5663139820098877,
-      "learning_rate": 0.0005049929964443486,
-      "loss": 3.6863,
+      "epoch": 1.595687331536388,
+      "grad_norm": 0.5361658334732056,
+      "learning_rate": 0.0005048289260658392,
+      "loss": 3.6654,
       "step": 14800
     },
     {
-      "epoch": 1.5983209557636422,
-      "grad_norm": 0.531947672367096,
-      "learning_rate": 0.0005046697554142871,
-      "loss": 3.6731,
+      "epoch": 1.6010781671159031,
+      "grad_norm": 0.5976514220237732,
+      "learning_rate": 0.0005045051268213707,
+      "loss": 3.6901,
       "step": 14850
     },
     {
-      "epoch": 1.6037025078032503,
-      "grad_norm": 0.5767591595649719,
-      "learning_rate": 0.0005043465143842258,
-      "loss": 3.66,
+      "epoch": 1.6064690026954178,
+      "grad_norm": 0.5398532152175903,
+      "learning_rate": 0.0005041813275769023,
+      "loss": 3.6962,
       "step": 14900
     },
     {
-      "epoch": 1.6090840598428586,
-      "grad_norm": 0.5688928961753845,
-      "learning_rate": 0.0005040232733541644,
-      "loss": 3.6792,
+      "epoch": 1.6118598382749325,
+      "grad_norm": 0.6030893325805664,
+      "learning_rate": 0.0005038575283324338,
+      "loss": 3.6812,
       "step": 14950
     },
     {
-      "epoch": 1.614465611882467,
-      "grad_norm": 0.5142151117324829,
-      "learning_rate": 0.000503700032324103,
-      "loss": 3.6817,
+      "epoch": 1.6172506738544474,
+      "grad_norm": 0.5295268297195435,
+      "learning_rate": 0.0005035337290879654,
+      "loss": 3.674,
       "step": 15000
     },
     {
-      "epoch": 1.614465611882467,
-      "eval_accuracy": 0.35604127983975004,
-      "eval_loss": 3.6317059993743896,
-      "eval_runtime": 186.4447,
-      "eval_samples_per_second": 96.602,
-      "eval_steps_per_second": 6.039,
+      "epoch": 1.6172506738544474,
+      "eval_accuracy": 0.3550096206699515,
+      "eval_loss": 3.6437854766845703,
+      "eval_runtime": 184.3989,
+      "eval_samples_per_second": 97.674,
+      "eval_steps_per_second": 6.106,
       "step": 15000
     },
     {
-      "epoch": 1.619847163922075,
-      "grad_norm": 0.5600343942642212,
-      "learning_rate": 0.0005033767912940415,
-      "loss": 3.6775,
+      "epoch": 1.6226415094339623,
+      "grad_norm": 0.5891487002372742,
+      "learning_rate": 0.000503209929843497,
+      "loss": 3.6709,
       "step": 15050
     },
     {
-      "epoch": 1.6252287159616834,
-      "grad_norm": 0.5435324311256409,
-      "learning_rate": 0.0005030535502639802,
-      "loss": 3.6679,
+      "epoch": 1.628032345013477,
+      "grad_norm": 0.5549376010894775,
+      "learning_rate": 0.0005028861305990286,
+      "loss": 3.6831,
       "step": 15100
     },
     {
-      "epoch": 1.6306102680012917,
-      "grad_norm": 0.5788407921791077,
-      "learning_rate": 0.0005027303092339187,
-      "loss": 3.7,
+      "epoch": 1.633423180592992,
+      "grad_norm": 0.582349956035614,
+      "learning_rate": 0.0005025623313545601,
+      "loss": 3.702,
       "step": 15150
     },
     {
-      "epoch": 1.6359918200408998,
-      "grad_norm": 0.5337195992469788,
-      "learning_rate": 0.0005024070682038573,
-      "loss": 3.682,
+      "epoch": 1.6388140161725069,
+      "grad_norm": 0.5355322957038879,
+      "learning_rate": 0.0005022385321100917,
+      "loss": 3.6882,
       "step": 15200
     },
     {
-      "epoch": 1.641373372080508,
-      "grad_norm": 0.6084492802619934,
-      "learning_rate": 0.0005020838271737959,
-      "loss": 3.6528,
+      "epoch": 1.6442048517520216,
+      "grad_norm": 0.5241622924804688,
+      "learning_rate": 0.0005019147328656232,
+      "loss": 3.6822,
       "step": 15250
     },
     {
-      "epoch": 1.6467549241201163,
-      "grad_norm": 0.5530157685279846,
-      "learning_rate": 0.0005017605861437344,
-      "loss": 3.6691,
+      "epoch": 1.6495956873315363,
+      "grad_norm": 0.5959563255310059,
+      "learning_rate": 0.0005015909336211549,
+      "loss": 3.678,
       "step": 15300
     },
     {
-      "epoch": 1.6521364761597246,
-      "grad_norm": 0.5181704163551331,
-      "learning_rate": 0.0005014373451136731,
-      "loss": 3.6746,
+      "epoch": 1.6549865229110512,
+      "grad_norm": 0.5305055975914001,
+      "learning_rate": 0.0005012671343766864,
+      "loss": 3.6654,
       "step": 15350
     },
     {
-      "epoch": 1.6575180281993327,
-      "grad_norm": 0.572228193283081,
-      "learning_rate": 0.0005011141040836116,
-      "loss": 3.683,
+      "epoch": 1.6603773584905661,
+      "grad_norm": 0.6248850226402283,
+      "learning_rate": 0.000500943335132218,
+      "loss": 3.6849,
       "step": 15400
     },
     {
-      "epoch": 1.6628995802389408,
-      "grad_norm": 0.5499255061149597,
-      "learning_rate": 0.0005007908630535503,
-      "loss": 3.6722,
+      "epoch": 1.6657681940700808,
+      "grad_norm": 0.6201460957527161,
+      "learning_rate": 0.0005006195358877495,
+      "loss": 3.6723,
       "step": 15450
     },
     {
-      "epoch": 1.6682811322785491,
-      "grad_norm": 0.559100866317749,
-      "learning_rate": 0.0005004676220234888,
-      "loss": 3.6535,
+      "epoch": 1.6711590296495957,
+      "grad_norm": 0.5390592217445374,
+      "learning_rate": 0.0005002957366432812,
+      "loss": 3.6742,
       "step": 15500
     },
     {
-      "epoch": 1.6736626843181575,
-      "grad_norm": 0.5993514060974121,
-      "learning_rate": 0.0005001443809934273,
-      "loss": 3.6549,
+      "epoch": 1.6765498652291106,
+      "grad_norm": 0.5546064972877502,
+      "learning_rate": 0.0004999719373988127,
+      "loss": 3.6594,
       "step": 15550
     },
     {
-      "epoch": 1.6790442363577656,
-      "grad_norm": 0.5286920666694641,
-      "learning_rate": 0.000499821139963366,
-      "loss": 3.6639,
+      "epoch": 1.6819407008086253,
+      "grad_norm": 0.5827458500862122,
+      "learning_rate": 0.0004996546141392336,
+      "loss": 3.6794,
       "step": 15600
     },
     {
-      "epoch": 1.6844257883973737,
-      "grad_norm": 0.5905039310455322,
-      "learning_rate": 0.0004994978989333045,
-      "loss": 3.6772,
+      "epoch": 1.68733153638814,
+      "grad_norm": 0.5636492371559143,
+      "learning_rate": 0.0004993308148947651,
+      "loss": 3.6829,
       "step": 15650
     },
     {
-      "epoch": 1.689807340436982,
-      "grad_norm": 0.5641158223152161,
-      "learning_rate": 0.0004991746579032431,
-      "loss": 3.6771,
+      "epoch": 1.692722371967655,
+      "grad_norm": 0.6371291875839233,
+      "learning_rate": 0.0004990070156502968,
+      "loss": 3.6791,
       "step": 15700
     },
     {
-      "epoch": 1.6951888924765903,
-      "grad_norm": 0.5468423962593079,
-      "learning_rate": 0.0004988514168731817,
-      "loss": 3.6633,
+      "epoch": 1.6981132075471699,
+      "grad_norm": 0.5599915385246277,
+      "learning_rate": 0.0004986832164058284,
+      "loss": 3.6613,
       "step": 15750
     },
     {
-      "epoch": 1.7005704445161984,
-      "grad_norm": 0.5505973100662231,
-      "learning_rate": 0.0004985281758431204,
-      "loss": 3.6676,
+      "epoch": 1.7035040431266846,
+      "grad_norm": 0.5805423855781555,
+      "learning_rate": 0.0004983594171613599,
+      "loss": 3.6731,
       "step": 15800
     },
     {
-      "epoch": 1.7059519965558065,
-      "grad_norm": 0.572822630405426,
-      "learning_rate": 0.0004982049348130589,
-      "loss": 3.6628,
+      "epoch": 1.7088948787061995,
+      "grad_norm": 0.5574281215667725,
+      "learning_rate": 0.0004980356179168915,
+      "loss": 3.688,
       "step": 15850
     },
     {
-      "epoch": 1.7113335485954149,
-      "grad_norm": 0.631182074546814,
-      "learning_rate": 0.0004978816937829975,
-      "loss": 3.6654,
+      "epoch": 1.7142857142857144,
+      "grad_norm": 0.5729790329933167,
+      "learning_rate": 0.000497711818672423,
+      "loss": 3.6879,
       "step": 15900
     },
     {
-      "epoch": 1.7167151006350232,
-      "grad_norm": 0.5868000388145447,
-      "learning_rate": 0.000497558452752936,
-      "loss": 3.6509,
+      "epoch": 1.719676549865229,
+      "grad_norm": 0.5762771368026733,
+      "learning_rate": 0.0004973880194279547,
+      "loss": 3.677,
       "step": 15950
     },
     {
-      "epoch": 1.7220966526746313,
-      "grad_norm": 0.586867094039917,
-      "learning_rate": 0.0004972352117228746,
-      "loss": 3.6788,
+      "epoch": 1.7250673854447438,
+      "grad_norm": 0.5976433157920837,
+      "learning_rate": 0.0004970642201834862,
+      "loss": 3.672,
       "step": 16000
     },
     {
-      "epoch": 1.7220966526746313,
-      "eval_accuracy": 0.35772311789023037,
-      "eval_loss": 3.614586591720581,
-      "eval_runtime": 186.4406,
-      "eval_samples_per_second": 96.604,
-      "eval_steps_per_second": 6.039,
+      "epoch": 1.7250673854447438,
+      "eval_accuracy": 0.3565820452624027,
+      "eval_loss": 3.6253254413604736,
+      "eval_runtime": 184.1532,
+      "eval_samples_per_second": 97.804,
+      "eval_steps_per_second": 6.114,
       "step": 16000
     },
     {
-      "epoch": 1.7274782047142396,
-      "grad_norm": 0.5538934469223022,
-      "learning_rate": 0.0004969119706928133,
-      "loss": 3.6573,
+      "epoch": 1.7304582210242587,
+      "grad_norm": 0.5788641571998596,
+      "learning_rate": 0.0004967404209390178,
+      "loss": 3.6599,
       "step": 16050
     },
     {
-      "epoch": 1.732859756753848,
-      "grad_norm": 0.5432226061820984,
-      "learning_rate": 0.0004965951944833531,
-      "loss": 3.6803,
+      "epoch": 1.7358490566037736,
+      "grad_norm": 0.5980125069618225,
+      "learning_rate": 0.0004964166216945493,
+      "loss": 3.6594,
       "step": 16100
     },
     {
-      "epoch": 1.738241308793456,
-      "grad_norm": 0.5792766213417053,
-      "learning_rate": 0.0004962719534532916,
-      "loss": 3.6563,
+      "epoch": 1.7412398921832883,
+      "grad_norm": 0.6294205188751221,
+      "learning_rate": 0.000496092822450081,
+      "loss": 3.668,
       "step": 16150
     },
     {
-      "epoch": 1.7436228608330642,
-      "grad_norm": 0.5571629405021667,
-      "learning_rate": 0.0004959487124232302,
-      "loss": 3.6578,
+      "epoch": 1.7466307277628033,
+      "grad_norm": 0.6452693939208984,
+      "learning_rate": 0.0004957690232056125,
+      "loss": 3.6757,
       "step": 16200
     },
     {
-      "epoch": 1.7490044128726725,
-      "grad_norm": 0.5888702869415283,
-      "learning_rate": 0.0004956254713931688,
-      "loss": 3.6534,
+      "epoch": 1.7520215633423182,
+      "grad_norm": 0.6337955594062805,
+      "learning_rate": 0.0004954452239611441,
+      "loss": 3.6567,
       "step": 16250
     },
     {
-      "epoch": 1.7543859649122808,
-      "grad_norm": 0.5217219591140747,
-      "learning_rate": 0.0004953022303631074,
-      "loss": 3.6584,
+      "epoch": 1.7574123989218329,
+      "grad_norm": 0.5349404215812683,
+      "learning_rate": 0.0004951214247166756,
+      "loss": 3.6525,
       "step": 16300
     },
     {
-      "epoch": 1.759767516951889,
-      "grad_norm": 0.5816513895988464,
-      "learning_rate": 0.0004949789893330459,
-      "loss": 3.6558,
+      "epoch": 1.7628032345013476,
+      "grad_norm": 0.5445868968963623,
+      "learning_rate": 0.0004947976254722072,
+      "loss": 3.676,
       "step": 16350
     },
     {
-      "epoch": 1.765149068991497,
-      "grad_norm": 0.5748602747917175,
-      "learning_rate": 0.0004946557483029846,
-      "loss": 3.6309,
+      "epoch": 1.7681940700808625,
+      "grad_norm": 0.525174081325531,
+      "learning_rate": 0.0004944738262277387,
+      "loss": 3.6534,
       "step": 16400
     },
     {
-      "epoch": 1.7705306210311054,
-      "grad_norm": 0.6380993723869324,
-      "learning_rate": 0.0004943325072729231,
-      "loss": 3.6687,
+      "epoch": 1.7735849056603774,
+      "grad_norm": 0.58821702003479,
+      "learning_rate": 0.0004941500269832703,
+      "loss": 3.6649,
       "step": 16450
     },
     {
-      "epoch": 1.7759121730707137,
-      "grad_norm": 0.5963583588600159,
-      "learning_rate": 0.0004940092662428617,
-      "loss": 3.6838,
+      "epoch": 1.778975741239892,
+      "grad_norm": 0.5536784529685974,
+      "learning_rate": 0.0004938262277388019,
+      "loss": 3.6523,
       "step": 16500
     },
     {
-      "epoch": 1.7812937251103218,
-      "grad_norm": 0.55112624168396,
-      "learning_rate": 0.0004936860252128003,
-      "loss": 3.6462,
+      "epoch": 1.784366576819407,
+      "grad_norm": 0.5388889312744141,
+      "learning_rate": 0.0004935024284943335,
+      "loss": 3.6578,
       "step": 16550
     },
     {
-      "epoch": 1.78667527714993,
-      "grad_norm": 0.5294162631034851,
-      "learning_rate": 0.0004933627841827388,
-      "loss": 3.6824,
+      "epoch": 1.789757412398922,
+      "grad_norm": 0.5487968325614929,
+      "learning_rate": 0.000493178629249865,
+      "loss": 3.6703,
       "step": 16600
     },
     {
-      "epoch": 1.7920568291895382,
-      "grad_norm": 0.6394523978233337,
-      "learning_rate": 0.0004930395431526775,
-      "loss": 3.6558,
+      "epoch": 1.7951482479784366,
+      "grad_norm": 0.5447817444801331,
+      "learning_rate": 0.0004928548300053966,
+      "loss": 3.6604,
       "step": 16650
     },
     {
-      "epoch": 1.7974383812291466,
-      "grad_norm": 0.5520370602607727,
-      "learning_rate": 0.0004927163021226161,
-      "loss": 3.6568,
+      "epoch": 1.8005390835579513,
+      "grad_norm": 0.6437905430793762,
+      "learning_rate": 0.0004925310307609282,
+      "loss": 3.668,
       "step": 16700
     },
     {
-      "epoch": 1.8028199332687547,
-      "grad_norm": 0.5403886437416077,
-      "learning_rate": 0.0004923930610925547,
-      "loss": 3.6424,
+      "epoch": 1.8059299191374663,
+      "grad_norm": 0.5301018357276917,
+      "learning_rate": 0.0004922072315164598,
+      "loss": 3.6662,
       "step": 16750
     },
     {
-      "epoch": 1.8082014853083628,
-      "grad_norm": 0.543944776058197,
-      "learning_rate": 0.0004920698200624932,
-      "loss": 3.6417,
+      "epoch": 1.8113207547169812,
+      "grad_norm": 0.5489523410797119,
+      "learning_rate": 0.0004918834322719913,
+      "loss": 3.6753,
       "step": 16800
     },
     {
-      "epoch": 1.813583037347971,
-      "grad_norm": 0.5547634363174438,
-      "learning_rate": 0.0004917465790324317,
-      "loss": 3.6439,
+      "epoch": 1.8167115902964959,
+      "grad_norm": 0.5634409785270691,
+      "learning_rate": 0.0004915596330275229,
+      "loss": 3.6846,
       "step": 16850
     },
     {
-      "epoch": 1.8189645893875794,
-      "grad_norm": 0.5435753464698792,
-      "learning_rate": 0.0004914233380023704,
-      "loss": 3.6422,
+      "epoch": 1.8221024258760108,
+      "grad_norm": 0.5558264255523682,
+      "learning_rate": 0.0004912358337830544,
+      "loss": 3.654,
       "step": 16900
     },
     {
-      "epoch": 1.8243461414271875,
-      "grad_norm": 0.5777159929275513,
-      "learning_rate": 0.0004911000969723089,
-      "loss": 3.6466,
+      "epoch": 1.8274932614555257,
+      "grad_norm": 0.6050113439559937,
+      "learning_rate": 0.000490912034538586,
+      "loss": 3.6671,
       "step": 16950
     },
     {
-      "epoch": 1.8297276934667959,
-      "grad_norm": 0.5340699553489685,
-      "learning_rate": 0.0004907768559422476,
-      "loss": 3.651,
+      "epoch": 1.8328840970350404,
+      "grad_norm": 0.5846980214118958,
+      "learning_rate": 0.0004905882352941175,
+      "loss": 3.6464,
       "step": 17000
     },
     {
-      "epoch": 1.8297276934667959,
-      "eval_accuracy": 0.3591974289249851,
-      "eval_loss": 3.598952054977417,
-      "eval_runtime": 186.3296,
-      "eval_samples_per_second": 96.662,
-      "eval_steps_per_second": 6.043,
+      "epoch": 1.8328840970350404,
+      "eval_accuracy": 0.3580526620989717,
+      "eval_loss": 3.609726667404175,
+      "eval_runtime": 184.6425,
+      "eval_samples_per_second": 97.545,
+      "eval_steps_per_second": 6.098,
       "step": 17000
     },
     {
-      "epoch": 1.8351092455064042,
-      "grad_norm": 0.5523515343666077,
-      "learning_rate": 0.0004904536149121861,
-      "loss": 3.6344,
+      "epoch": 1.838274932614555,
+      "grad_norm": 0.5543476939201355,
+      "learning_rate": 0.0004902644360496492,
+      "loss": 3.6521,
       "step": 17050
     },
     {
-      "epoch": 1.8404907975460123,
-      "grad_norm": 0.6107510924339294,
-      "learning_rate": 0.0004901303738821248,
-      "loss": 3.6431,
+      "epoch": 1.8436657681940702,
+      "grad_norm": 0.5313900709152222,
+      "learning_rate": 0.0004899406368051808,
+      "loss": 3.6417,
       "step": 17100
     },
     {
-      "epoch": 1.8458723495856204,
-      "grad_norm": 0.541286826133728,
-      "learning_rate": 0.0004898071328520633,
-      "loss": 3.6404,
+      "epoch": 1.849056603773585,
+      "grad_norm": 0.5435433387756348,
+      "learning_rate": 0.0004896168375607123,
+      "loss": 3.6479,
       "step": 17150
     },
     {
-      "epoch": 1.8512539016252287,
-      "grad_norm": 0.5887472033500671,
-      "learning_rate": 0.0004894838918220019,
-      "loss": 3.6466,
+      "epoch": 1.8544474393530996,
+      "grad_norm": 0.5575142502784729,
+      "learning_rate": 0.0004892930383162439,
+      "loss": 3.6556,
       "step": 17200
     },
     {
-      "epoch": 1.856635453664837,
-      "grad_norm": 0.6260271668434143,
-      "learning_rate": 0.0004891606507919405,
-      "loss": 3.6581,
+      "epoch": 1.8598382749326146,
+      "grad_norm": 0.549067497253418,
+      "learning_rate": 0.0004889692390717754,
+      "loss": 3.6615,
       "step": 17250
     },
     {
-      "epoch": 1.8620170057044452,
-      "grad_norm": 0.5545384883880615,
-      "learning_rate": 0.000488837409761879,
-      "loss": 3.6341,
+      "epoch": 1.8652291105121295,
+      "grad_norm": 0.5817179679870605,
+      "learning_rate": 0.0004886454398273071,
+      "loss": 3.6629,
       "step": 17300
     },
     {
-      "epoch": 1.8673985577440533,
-      "grad_norm": 0.5130829811096191,
-      "learning_rate": 0.0004885141687318177,
-      "loss": 3.6398,
+      "epoch": 1.8706199460916442,
+      "grad_norm": 0.5488595962524414,
+      "learning_rate": 0.0004883216405828386,
+      "loss": 3.6632,
       "step": 17350
     },
     {
-      "epoch": 1.8727801097836616,
-      "grad_norm": 0.547607958316803,
-      "learning_rate": 0.00048819092770175623,
-      "loss": 3.6339,
+      "epoch": 1.8760107816711589,
+      "grad_norm": 0.5994912385940552,
+      "learning_rate": 0.00048799784133837017,
+      "loss": 3.6574,
       "step": 17400
     },
     {
-      "epoch": 1.87816166182327,
-      "grad_norm": 0.5617174506187439,
-      "learning_rate": 0.0004878676866716948,
-      "loss": 3.6433,
+      "epoch": 1.881401617250674,
+      "grad_norm": 0.6115542054176331,
+      "learning_rate": 0.0004876740420939017,
+      "loss": 3.6454,
       "step": 17450
     },
     {
-      "epoch": 1.883543213862878,
-      "grad_norm": 0.5771898031234741,
-      "learning_rate": 0.00048754444564163337,
-      "loss": 3.6459,
+      "epoch": 1.8867924528301887,
+      "grad_norm": 0.5530582070350647,
+      "learning_rate": 0.0004873502428494333,
+      "loss": 3.6496,
       "step": 17500
     },
     {
-      "epoch": 1.8889247659024861,
-      "grad_norm": 0.6357139348983765,
-      "learning_rate": 0.000487221204611572,
-      "loss": 3.614,
+      "epoch": 1.8921832884097034,
+      "grad_norm": 0.597615659236908,
+      "learning_rate": 0.0004870264436049649,
+      "loss": 3.6463,
       "step": 17550
     },
     {
-      "epoch": 1.8943063179420945,
-      "grad_norm": 0.5490467548370361,
-      "learning_rate": 0.00048689796358151056,
-      "loss": 3.6501,
+      "epoch": 1.8975741239892183,
+      "grad_norm": 0.584467887878418,
+      "learning_rate": 0.00048670264436049643,
+      "loss": 3.6615,
       "step": 17600
     },
     {
-      "epoch": 1.8996878699817028,
-      "grad_norm": 0.6037611365318298,
-      "learning_rate": 0.00048657472255144915,
-      "loss": 3.6563,
+      "epoch": 1.9029649595687332,
+      "grad_norm": 0.5941674113273621,
+      "learning_rate": 0.00048637884511602803,
+      "loss": 3.6503,
       "step": 17650
     },
     {
-      "epoch": 1.905069422021311,
-      "grad_norm": 0.5259382128715515,
-      "learning_rate": 0.00048625148152138775,
-      "loss": 3.6391,
+      "epoch": 1.908355795148248,
+      "grad_norm": 0.5420777201652527,
+      "learning_rate": 0.000486061521856449,
+      "loss": 3.6443,
       "step": 17700
     },
     {
-      "epoch": 1.910450974060919,
-      "grad_norm": 0.6096329689025879,
-      "learning_rate": 0.0004859282404913263,
-      "loss": 3.6324,
+      "epoch": 1.9137466307277629,
+      "grad_norm": 0.5749199986457825,
+      "learning_rate": 0.00048573772261198054,
+      "loss": 3.6652,
       "step": 17750
     },
     {
-      "epoch": 1.9158325261005273,
-      "grad_norm": 0.6676279306411743,
-      "learning_rate": 0.0004856049994612649,
-      "loss": 3.6366,
+      "epoch": 1.9191374663072778,
+      "grad_norm": 0.5945655107498169,
+      "learning_rate": 0.00048541392336751214,
+      "loss": 3.6496,
       "step": 17800
     },
     {
-      "epoch": 1.9212140781401357,
-      "grad_norm": 0.5407010912895203,
-      "learning_rate": 0.00048528175843120353,
-      "loss": 3.6307,
+      "epoch": 1.9245283018867925,
+      "grad_norm": 0.5909385085105896,
+      "learning_rate": 0.0004850901241230437,
+      "loss": 3.6416,
       "step": 17850
     },
     {
-      "epoch": 1.9265956301797438,
-      "grad_norm": 0.6026391386985779,
-      "learning_rate": 0.0004849585174011421,
-      "loss": 3.6292,
+      "epoch": 1.9299191374663072,
+      "grad_norm": 0.5878526568412781,
+      "learning_rate": 0.0004847663248785753,
+      "loss": 3.6539,
       "step": 17900
     },
     {
-      "epoch": 1.931977182219352,
-      "grad_norm": 0.6055847406387329,
-      "learning_rate": 0.00048463527637108067,
-      "loss": 3.6281,
+      "epoch": 1.935309973045822,
+      "grad_norm": 0.5761050581932068,
+      "learning_rate": 0.0004844425256341068,
+      "loss": 3.6362,
       "step": 17950
     },
     {
-      "epoch": 1.9373587342589604,
-      "grad_norm": 0.5859214663505554,
-      "learning_rate": 0.0004843120353410192,
-      "loss": 3.631,
+      "epoch": 1.940700808625337,
+      "grad_norm": 0.5990300178527832,
+      "learning_rate": 0.00048411872638963834,
+      "loss": 3.6394,
       "step": 18000
     },
     {
-      "epoch": 1.9373587342589604,
-      "eval_accuracy": 0.36063762295296603,
-      "eval_loss": 3.5867769718170166,
-      "eval_runtime": 186.6726,
-      "eval_samples_per_second": 96.484,
-      "eval_steps_per_second": 6.032,
+      "epoch": 1.940700808625337,
+      "eval_accuracy": 0.35914929569568327,
+      "eval_loss": 3.5965323448181152,
+      "eval_runtime": 184.1623,
+      "eval_samples_per_second": 97.8,
+      "eval_steps_per_second": 6.114,
       "step": 18000
     },
     {
-      "epoch": 1.9427402862985685,
-      "grad_norm": 0.5761770606040955,
-      "learning_rate": 0.0004839887943109578,
-      "loss": 3.6389,
+      "epoch": 1.9460916442048517,
+      "grad_norm": 0.584472119808197,
+      "learning_rate": 0.00048379492714516995,
+      "loss": 3.6413,
       "step": 18050
     },
     {
-      "epoch": 1.9481218383381766,
-      "grad_norm": 0.6096853017807007,
-      "learning_rate": 0.00048366555328089645,
-      "loss": 3.6308,
+      "epoch": 1.9514824797843666,
+      "grad_norm": 0.584588348865509,
+      "learning_rate": 0.0004834711279007015,
+      "loss": 3.6424,
       "step": 18100
     },
     {
-      "epoch": 1.953503390377785,
-      "grad_norm": 0.61137455701828,
-      "learning_rate": 0.0004833487770714362,
-      "loss": 3.6394,
+      "epoch": 1.9568733153638815,
+      "grad_norm": 0.543335497379303,
+      "learning_rate": 0.0004831473286562331,
+      "loss": 3.6401,
       "step": 18150
     },
     {
-      "epoch": 1.9588849424173933,
-      "grad_norm": 0.6039766669273376,
-      "learning_rate": 0.00048302553604137485,
-      "loss": 3.6234,
+      "epoch": 1.9622641509433962,
+      "grad_norm": 0.5636866688728333,
+      "learning_rate": 0.00048283000539665405,
+      "loss": 3.6294,
       "step": 18200
     },
     {
-      "epoch": 1.9642664944570014,
-      "grad_norm": 0.6795485019683838,
-      "learning_rate": 0.0004827022950113134,
-      "loss": 3.6331,
+      "epoch": 1.967654986522911,
+      "grad_norm": 0.5789482593536377,
+      "learning_rate": 0.0004825062061521856,
+      "loss": 3.6484,
       "step": 18250
     },
     {
-      "epoch": 1.9696480464966095,
-      "grad_norm": 0.5619679689407349,
-      "learning_rate": 0.000482379053981252,
-      "loss": 3.6125,
+      "epoch": 1.9730458221024259,
+      "grad_norm": 0.5915958285331726,
+      "learning_rate": 0.00048218240690771716,
+      "loss": 3.6591,
       "step": 18300
     },
     {
-      "epoch": 1.9750295985362178,
-      "grad_norm": 0.592473030090332,
-      "learning_rate": 0.0004820558129511906,
-      "loss": 3.6311,
+      "epoch": 1.9784366576819408,
+      "grad_norm": 0.5506396889686584,
+      "learning_rate": 0.00048185860766324876,
+      "loss": 3.6442,
       "step": 18350
     },
     {
-      "epoch": 1.9804111505758262,
-      "grad_norm": 0.581325113773346,
-      "learning_rate": 0.0004817325719211291,
-      "loss": 3.6323,
+      "epoch": 1.9838274932614555,
+      "grad_norm": 0.6207454800605774,
+      "learning_rate": 0.0004815348084187803,
+      "loss": 3.643,
       "step": 18400
     },
     {
-      "epoch": 1.9857927026154343,
-      "grad_norm": 0.5376942157745361,
-      "learning_rate": 0.0004814093308910677,
-      "loss": 3.6229,
+      "epoch": 1.9892183288409704,
+      "grad_norm": 0.5647146105766296,
+      "learning_rate": 0.0004812110091743119,
+      "loss": 3.6526,
       "step": 18450
     },
     {
-      "epoch": 1.9911742546550424,
-      "grad_norm": 0.5628843903541565,
-      "learning_rate": 0.00048108608986100637,
-      "loss": 3.6377,
+      "epoch": 1.9946091644204853,
+      "grad_norm": 0.6433171629905701,
+      "learning_rate": 0.00048088720992984347,
+      "loss": 3.6441,
       "step": 18500
     },
     {
-      "epoch": 1.9965558066946507,
-      "grad_norm": 0.5238602161407471,
-      "learning_rate": 0.0004807628488309449,
-      "loss": 3.6192,
+      "epoch": 2.0,
+      "grad_norm": 1.183536410331726,
+      "learning_rate": 0.00048056341068537507,
+      "loss": 3.649,
       "step": 18550
     },
     {
-      "epoch": 2.001937358734259,
-      "grad_norm": 0.6020618677139282,
-      "learning_rate": 0.0004804396078008835,
-      "loss": 3.5992,
+      "epoch": 2.0053908355795147,
+      "grad_norm": 0.5773569345474243,
+      "learning_rate": 0.00048023961144090657,
+      "loss": 3.5788,
       "step": 18600
     },
     {
-      "epoch": 2.007318910773867,
-      "grad_norm": 0.5989746451377869,
-      "learning_rate": 0.00048011636677082204,
-      "loss": 3.5263,
+      "epoch": 2.01078167115903,
+      "grad_norm": 0.5826117396354675,
+      "learning_rate": 0.00047991581219643817,
+      "loss": 3.549,
       "step": 18650
     },
     {
-      "epoch": 2.0127004628134753,
-      "grad_norm": 0.5591849088668823,
-      "learning_rate": 0.00047979312574076064,
-      "loss": 3.5437,
+      "epoch": 2.0161725067385445,
+      "grad_norm": 0.5458110570907593,
+      "learning_rate": 0.0004795920129519697,
+      "loss": 3.5604,
       "step": 18700
     },
     {
-      "epoch": 2.018082014853084,
-      "grad_norm": 0.546755313873291,
-      "learning_rate": 0.0004794698847106992,
-      "loss": 3.5435,
+      "epoch": 2.0215633423180592,
+      "grad_norm": 0.5588856935501099,
+      "learning_rate": 0.0004792682137075013,
+      "loss": 3.5396,
       "step": 18750
     },
     {
-      "epoch": 2.023463566892692,
-      "grad_norm": 0.5691375136375427,
-      "learning_rate": 0.0004791466436806378,
-      "loss": 3.5537,
+      "epoch": 2.026954177897574,
+      "grad_norm": 0.6126614212989807,
+      "learning_rate": 0.0004789444144630329,
+      "loss": 3.5588,
       "step": 18800
     },
     {
-      "epoch": 2.0288451189323,
-      "grad_norm": 0.5947789549827576,
-      "learning_rate": 0.0004788234026505764,
-      "loss": 3.5553,
+      "epoch": 2.032345013477089,
+      "grad_norm": 0.5554583072662354,
+      "learning_rate": 0.00047862061521856443,
+      "loss": 3.5659,
       "step": 18850
     },
     {
-      "epoch": 2.034226670971908,
-      "grad_norm": 0.614838182926178,
-      "learning_rate": 0.00047850016162051496,
-      "loss": 3.5494,
+      "epoch": 2.0377358490566038,
+      "grad_norm": 0.5678772926330566,
+      "learning_rate": 0.00047829681597409603,
+      "loss": 3.5547,
       "step": 18900
     },
     {
-      "epoch": 2.0396082230115167,
-      "grad_norm": 0.581800103187561,
-      "learning_rate": 0.00047817692059045356,
-      "loss": 3.5432,
+      "epoch": 2.0431266846361185,
+      "grad_norm": 0.5880005955696106,
+      "learning_rate": 0.0004779730167296276,
+      "loss": 3.5471,
       "step": 18950
     },
     {
-      "epoch": 2.044989775051125,
-      "grad_norm": 0.5449944138526917,
-      "learning_rate": 0.00047785367956039215,
-      "loss": 3.5611,
+      "epoch": 2.0485175202156336,
+      "grad_norm": 0.621539294719696,
+      "learning_rate": 0.0004776492174851592,
+      "loss": 3.5754,
       "step": 19000
     },
     {
-      "epoch": 2.044989775051125,
-      "eval_accuracy": 0.36166873885832546,
-      "eval_loss": 3.576646089553833,
-      "eval_runtime": 186.3372,
-      "eval_samples_per_second": 96.658,
-      "eval_steps_per_second": 6.043,
+      "epoch": 2.0485175202156336,
+      "eval_accuracy": 0.3607663766250262,
+      "eval_loss": 3.5881307125091553,
+      "eval_runtime": 184.1685,
+      "eval_samples_per_second": 97.796,
+      "eval_steps_per_second": 6.114,
       "step": 19000
     },
     {
-      "epoch": 2.050371327090733,
-      "grad_norm": 0.5662707090377808,
-      "learning_rate": 0.00047753043853033075,
-      "loss": 3.5469,
+      "epoch": 2.0539083557951483,
+      "grad_norm": 0.582294762134552,
+      "learning_rate": 0.00047732541824069074,
+      "loss": 3.5525,
       "step": 19050
     },
     {
-      "epoch": 2.055752879130341,
-      "grad_norm": 0.551207423210144,
-      "learning_rate": 0.00047720719750026934,
-      "loss": 3.5311,
+      "epoch": 2.059299191374663,
+      "grad_norm": 0.5775398015975952,
+      "learning_rate": 0.0004770016189962223,
+      "loss": 3.5751,
       "step": 19100
     },
     {
-      "epoch": 2.0611344311699495,
-      "grad_norm": 0.5451582670211792,
-      "learning_rate": 0.00047688395647020793,
-      "loss": 3.549,
+      "epoch": 2.0646900269541777,
+      "grad_norm": 0.6106521487236023,
+      "learning_rate": 0.0004766778197517539,
+      "loss": 3.569,
       "step": 19150
     },
     {
-      "epoch": 2.0665159832095576,
-      "grad_norm": 0.5709878206253052,
-      "learning_rate": 0.0004765607154401465,
-      "loss": 3.5479,
+      "epoch": 2.070080862533693,
+      "grad_norm": 0.6209040880203247,
+      "learning_rate": 0.0004763540205072854,
+      "loss": 3.5758,
       "step": 19200
     },
     {
-      "epoch": 2.0718975352491658,
-      "grad_norm": 0.6128430366516113,
-      "learning_rate": 0.00047623747441008507,
-      "loss": 3.5679,
+      "epoch": 2.0754716981132075,
+      "grad_norm": 0.6275937557220459,
+      "learning_rate": 0.00047603022126281705,
+      "loss": 3.5573,
       "step": 19250
     },
     {
-      "epoch": 2.0772790872887743,
-      "grad_norm": 0.5695922374725342,
-      "learning_rate": 0.0004759142333800236,
-      "loss": 3.5502,
+      "epoch": 2.0808625336927222,
+      "grad_norm": 0.5772266387939453,
+      "learning_rate": 0.00047570642201834855,
+      "loss": 3.5522,
       "step": 19300
     },
     {
-      "epoch": 2.0826606393283824,
-      "grad_norm": 0.5929347276687622,
-      "learning_rate": 0.00047559099234996226,
-      "loss": 3.5434,
+      "epoch": 2.0862533692722374,
+      "grad_norm": 0.5767107009887695,
+      "learning_rate": 0.00047538262277388015,
+      "loss": 3.5573,
       "step": 19350
     },
     {
-      "epoch": 2.0880421913679905,
-      "grad_norm": 0.5880864262580872,
-      "learning_rate": 0.00047526775131990085,
-      "loss": 3.5476,
+      "epoch": 2.091644204851752,
+      "grad_norm": 0.6045193076133728,
+      "learning_rate": 0.0004750588235294117,
+      "loss": 3.5516,
       "step": 19400
     },
     {
-      "epoch": 2.0934237434075986,
-      "grad_norm": 0.568219780921936,
-      "learning_rate": 0.0004749445102898394,
-      "loss": 3.5383,
+      "epoch": 2.0970350404312668,
+      "grad_norm": 0.8183354735374451,
+      "learning_rate": 0.0004747350242849433,
+      "loss": 3.5617,
       "step": 19450
     },
     {
-      "epoch": 2.098805295447207,
-      "grad_norm": 0.5589820146560669,
-      "learning_rate": 0.000474621269259778,
-      "loss": 3.5527,
+      "epoch": 2.1024258760107815,
+      "grad_norm": 0.5576651692390442,
+      "learning_rate": 0.00047441122504047486,
+      "loss": 3.5533,
       "step": 19500
     },
     {
-      "epoch": 2.1041868474868153,
-      "grad_norm": 0.6112917065620422,
-      "learning_rate": 0.0004742980282297166,
-      "loss": 3.5511,
+      "epoch": 2.1078167115902966,
+      "grad_norm": 0.5935356020927429,
+      "learning_rate": 0.0004740874257960064,
+      "loss": 3.5591,
       "step": 19550
     },
     {
-      "epoch": 2.1095683995264234,
-      "grad_norm": 0.5517949461936951,
-      "learning_rate": 0.0004739747871996551,
-      "loss": 3.5544,
+      "epoch": 2.1132075471698113,
+      "grad_norm": 0.6061331629753113,
+      "learning_rate": 0.000473763626551538,
+      "loss": 3.5571,
       "step": 19600
     },
     {
-      "epoch": 2.1149499515660315,
-      "grad_norm": 0.5628524422645569,
-      "learning_rate": 0.00047365154616959377,
-      "loss": 3.5554,
+      "epoch": 2.118598382749326,
+      "grad_norm": 0.5980098843574524,
+      "learning_rate": 0.00047343982730706956,
+      "loss": 3.578,
       "step": 19650
     },
     {
-      "epoch": 2.12033150360564,
-      "grad_norm": 0.5876952409744263,
-      "learning_rate": 0.00047332830513953237,
-      "loss": 3.5664,
+      "epoch": 2.123989218328841,
+      "grad_norm": 0.6133925318717957,
+      "learning_rate": 0.00047311602806260117,
+      "loss": 3.5497,
       "step": 19700
     },
     {
-      "epoch": 2.125713055645248,
-      "grad_norm": 0.6220678687095642,
-      "learning_rate": 0.0004730050641094709,
-      "loss": 3.5506,
+      "epoch": 2.129380053908356,
+      "grad_norm": 0.5455171465873718,
+      "learning_rate": 0.0004727922288181327,
+      "loss": 3.5697,
       "step": 19750
     },
     {
-      "epoch": 2.1310946076848563,
-      "grad_norm": 0.5798251628875732,
-      "learning_rate": 0.0004726818230794095,
-      "loss": 3.5393,
+      "epoch": 2.1347708894878705,
+      "grad_norm": 0.5884529948234558,
+      "learning_rate": 0.0004724684295736643,
+      "loss": 3.5617,
       "step": 19800
     },
     {
-      "epoch": 2.1364761597244644,
-      "grad_norm": 0.5333768129348755,
-      "learning_rate": 0.00047235858204934804,
-      "loss": 3.5541,
+      "epoch": 2.1401617250673857,
+      "grad_norm": 0.5963370203971863,
+      "learning_rate": 0.0004721446303291959,
+      "loss": 3.5558,
       "step": 19850
     },
     {
-      "epoch": 2.141857711764073,
-      "grad_norm": 0.5997352600097656,
-      "learning_rate": 0.0004720353410192867,
-      "loss": 3.5608,
+      "epoch": 2.1455525606469004,
+      "grad_norm": 0.6389753222465515,
+      "learning_rate": 0.0004718208310847275,
+      "loss": 3.5659,
       "step": 19900
     },
     {
-      "epoch": 2.147239263803681,
-      "grad_norm": 0.5617119073867798,
-      "learning_rate": 0.0004717120999892253,
-      "loss": 3.5493,
+      "epoch": 2.150943396226415,
+      "grad_norm": 0.5952791571617126,
+      "learning_rate": 0.000471497031840259,
+      "loss": 3.5592,
       "step": 19950
     },
     {
-      "epoch": 2.152620815843289,
-      "grad_norm": 0.5545929074287415,
-      "learning_rate": 0.0004713888589591638,
-      "loss": 3.5688,
+      "epoch": 2.1563342318059298,
+      "grad_norm": 0.6527063846588135,
+      "learning_rate": 0.0004711732325957905,
+      "loss": 3.5828,
       "step": 20000
     },
     {
-      "epoch": 2.152620815843289,
-      "eval_accuracy": 0.3632254087820435,
-      "eval_loss": 3.5676701068878174,
-      "eval_runtime": 186.7931,
-      "eval_samples_per_second": 96.422,
-      "eval_steps_per_second": 6.028,
+      "epoch": 2.1563342318059298,
+      "eval_accuracy": 0.3613796135238511,
+      "eval_loss": 3.5782346725463867,
+      "eval_runtime": 183.9907,
+      "eval_samples_per_second": 97.891,
+      "eval_steps_per_second": 6.12,
       "step": 20000
     },
     {
-      "epoch": 2.1580023678828972,
-      "grad_norm": 0.5706044435501099,
-      "learning_rate": 0.0004710656179291024,
-      "loss": 3.5375,
+      "epoch": 2.161725067385445,
+      "grad_norm": 0.6027877926826477,
+      "learning_rate": 0.00047084943335132213,
+      "loss": 3.5846,
       "step": 20050
     },
     {
-      "epoch": 2.163383919922506,
-      "grad_norm": 0.6058392524719238,
-      "learning_rate": 0.000470742376899041,
-      "loss": 3.5597,
+      "epoch": 2.1671159029649596,
+      "grad_norm": 0.5578047037124634,
+      "learning_rate": 0.0004705256341068537,
+      "loss": 3.5827,
       "step": 20100
     },
     {
-      "epoch": 2.168765471962114,
-      "grad_norm": 0.5709688067436218,
-      "learning_rate": 0.00047041913586897956,
-      "loss": 3.5626,
+      "epoch": 2.1725067385444743,
+      "grad_norm": 0.5658228993415833,
+      "learning_rate": 0.0004702018348623853,
+      "loss": 3.5783,
       "step": 20150
     },
     {
-      "epoch": 2.174147024001722,
-      "grad_norm": 0.5546327233314514,
-      "learning_rate": 0.0004700958948389182,
-      "loss": 3.5639,
+      "epoch": 2.177897574123989,
+      "grad_norm": 0.569828450679779,
+      "learning_rate": 0.00046987803561791684,
+      "loss": 3.5747,
       "step": 20200
     },
     {
-      "epoch": 2.1795285760413305,
-      "grad_norm": 0.5731617212295532,
-      "learning_rate": 0.0004697726538088568,
-      "loss": 3.5279,
+      "epoch": 2.183288409703504,
+      "grad_norm": 0.5543139576911926,
+      "learning_rate": 0.00046955423637344844,
+      "loss": 3.5751,
       "step": 20250
     },
     {
-      "epoch": 2.1849101280809387,
-      "grad_norm": 0.5344586968421936,
-      "learning_rate": 0.0004694558775993966,
-      "loss": 3.5316,
+      "epoch": 2.188679245283019,
+      "grad_norm": 0.5759474039077759,
+      "learning_rate": 0.00046923043712898,
+      "loss": 3.566,
       "step": 20300
     },
     {
-      "epoch": 2.1902916801205468,
-      "grad_norm": 0.5539819598197937,
-      "learning_rate": 0.00046913263656933515,
-      "loss": 3.551,
+      "epoch": 2.1940700808625335,
+      "grad_norm": 0.5713962316513062,
+      "learning_rate": 0.0004689066378845116,
+      "loss": 3.5816,
       "step": 20350
     },
     {
-      "epoch": 2.195673232160155,
-      "grad_norm": 0.5484054684638977,
-      "learning_rate": 0.00046880939553927374,
-      "loss": 3.5564,
+      "epoch": 2.1994609164420487,
+      "grad_norm": 0.6075260043144226,
+      "learning_rate": 0.00046858283864004315,
+      "loss": 3.5799,
       "step": 20400
     },
     {
-      "epoch": 2.2010547841997634,
-      "grad_norm": 0.5522842407226562,
-      "learning_rate": 0.00046848615450921234,
-      "loss": 3.5538,
+      "epoch": 2.2048517520215634,
+      "grad_norm": 0.5758059620857239,
+      "learning_rate": 0.0004682590393955747,
+      "loss": 3.5779,
       "step": 20450
     },
     {
-      "epoch": 2.2064363362393715,
-      "grad_norm": 0.7291058301925659,
-      "learning_rate": 0.0004681629134791509,
-      "loss": 3.5492,
+      "epoch": 2.210242587601078,
+      "grad_norm": 0.5641762614250183,
+      "learning_rate": 0.0004679352401511063,
+      "loss": 3.5503,
       "step": 20500
     },
     {
-      "epoch": 2.2118178882789796,
-      "grad_norm": 0.5805173516273499,
-      "learning_rate": 0.00046783967244908947,
-      "loss": 3.564,
+      "epoch": 2.215633423180593,
+      "grad_norm": 0.6047466397285461,
+      "learning_rate": 0.0004676114409066378,
+      "loss": 3.5775,
       "step": 20550
     },
     {
-      "epoch": 2.2171994403185877,
-      "grad_norm": 0.5900067687034607,
-      "learning_rate": 0.0004675164314190281,
-      "loss": 3.5614,
+      "epoch": 2.221024258760108,
+      "grad_norm": 0.6605554819107056,
+      "learning_rate": 0.00046728764166216946,
+      "loss": 3.5539,
       "step": 20600
     },
     {
-      "epoch": 2.2225809923581963,
-      "grad_norm": 0.5646765828132629,
-      "learning_rate": 0.00046719319038896666,
-      "loss": 3.5451,
+      "epoch": 2.2264150943396226,
+      "grad_norm": 0.5884596109390259,
+      "learning_rate": 0.00046696384241770095,
+      "loss": 3.56,
       "step": 20650
     },
     {
-      "epoch": 2.2279625443978044,
-      "grad_norm": 0.5644922256469727,
-      "learning_rate": 0.00046686994935890526,
-      "loss": 3.5542,
+      "epoch": 2.2318059299191373,
+      "grad_norm": 0.5671321749687195,
+      "learning_rate": 0.00046664004317323256,
+      "loss": 3.5587,
       "step": 20700
     },
     {
-      "epoch": 2.2333440964374125,
-      "grad_norm": 0.5582818984985352,
-      "learning_rate": 0.0004665467083288438,
-      "loss": 3.5584,
+      "epoch": 2.2371967654986524,
+      "grad_norm": 0.5857590436935425,
+      "learning_rate": 0.0004663162439287641,
+      "loss": 3.5643,
       "step": 20750
     },
     {
-      "epoch": 2.2387256484770206,
-      "grad_norm": 0.585395097732544,
-      "learning_rate": 0.0004662234672987824,
-      "loss": 3.5622,
+      "epoch": 2.242587601078167,
+      "grad_norm": 0.6296424865722656,
+      "learning_rate": 0.00046599244468429566,
+      "loss": 3.5597,
       "step": 20800
     },
     {
-      "epoch": 2.244107200516629,
-      "grad_norm": 0.543573796749115,
-      "learning_rate": 0.00046590022626872104,
-      "loss": 3.543,
+      "epoch": 2.247978436657682,
+      "grad_norm": 0.557126522064209,
+      "learning_rate": 0.00046566864543982726,
+      "loss": 3.5764,
       "step": 20850
     },
     {
-      "epoch": 2.2494887525562373,
-      "grad_norm": 0.5891315937042236,
-      "learning_rate": 0.0004655834500592608,
-      "loss": 3.5704,
+      "epoch": 2.2533692722371965,
+      "grad_norm": 0.5671316385269165,
+      "learning_rate": 0.0004653448461953588,
+      "loss": 3.5689,
       "step": 20900
     },
     {
-      "epoch": 2.2548703045958454,
-      "grad_norm": 0.5713940858840942,
-      "learning_rate": 0.00046526020902919944,
-      "loss": 3.5673,
+      "epoch": 2.2587601078167117,
+      "grad_norm": 0.5766490697860718,
+      "learning_rate": 0.0004650210469508904,
+      "loss": 3.5574,
       "step": 20950
     },
     {
-      "epoch": 2.2602518566354535,
-      "grad_norm": 0.5523359179496765,
-      "learning_rate": 0.000464936967999138,
-      "loss": 3.5606,
+      "epoch": 2.2641509433962264,
+      "grad_norm": 0.5920085906982422,
+      "learning_rate": 0.00046469724770642197,
+      "loss": 3.5716,
       "step": 21000
     },
     {
-      "epoch": 2.2602518566354535,
-      "eval_accuracy": 0.3642741264552289,
-      "eval_loss": 3.559185743331909,
-      "eval_runtime": 186.2773,
-      "eval_samples_per_second": 96.689,
-      "eval_steps_per_second": 6.045,
+      "epoch": 2.2641509433962264,
+      "eval_accuracy": 0.3629849599413101,
+      "eval_loss": 3.5669126510620117,
+      "eval_runtime": 184.1445,
+      "eval_samples_per_second": 97.809,
+      "eval_steps_per_second": 6.115,
       "step": 21000
     },
     {
-      "epoch": 2.265633408675062,
-      "grad_norm": 0.5400150418281555,
-      "learning_rate": 0.0004646137269690766,
-      "loss": 3.5577,
+      "epoch": 2.269541778975741,
+      "grad_norm": 0.572544515132904,
+      "learning_rate": 0.0004643734484619536,
+      "loss": 3.5635,
       "step": 21050
     },
     {
-      "epoch": 2.27101496071467,
-      "grad_norm": 0.5969783067703247,
-      "learning_rate": 0.00046429048593901517,
-      "loss": 3.5389,
+      "epoch": 2.274932614555256,
+      "grad_norm": 0.6115766763687134,
+      "learning_rate": 0.0004640496492174851,
+      "loss": 3.5617,
       "step": 21100
     },
     {
-      "epoch": 2.2763965127542782,
-      "grad_norm": 0.5856836438179016,
-      "learning_rate": 0.0004639672449089537,
-      "loss": 3.5495,
+      "epoch": 2.280323450134771,
+      "grad_norm": 0.6257408857345581,
+      "learning_rate": 0.00046372584997301673,
+      "loss": 3.5835,
       "step": 21150
     },
     {
-      "epoch": 2.281778064793887,
-      "grad_norm": 0.5923604965209961,
-      "learning_rate": 0.0004636440038788923,
-      "loss": 3.5363,
+      "epoch": 2.2857142857142856,
+      "grad_norm": 0.6438502669334412,
+      "learning_rate": 0.0004634020507285483,
+      "loss": 3.5701,
       "step": 21200
     },
     {
-      "epoch": 2.287159616833495,
-      "grad_norm": 0.5832878351211548,
-      "learning_rate": 0.00046332076284883095,
-      "loss": 3.534,
+      "epoch": 2.2911051212938007,
+      "grad_norm": 0.7775608897209167,
+      "learning_rate": 0.0004630782514840798,
+      "loss": 3.5787,
       "step": 21250
     },
     {
-      "epoch": 2.292541168873103,
-      "grad_norm": 0.6145648956298828,
-      "learning_rate": 0.0004629975218187695,
-      "loss": 3.5545,
+      "epoch": 2.2964959568733154,
+      "grad_norm": 0.5988386273384094,
+      "learning_rate": 0.0004627544522396114,
+      "loss": 3.5452,
       "step": 21300
     },
     {
-      "epoch": 2.297922720912711,
-      "grad_norm": 0.6220842003822327,
-      "learning_rate": 0.0004626742807887081,
-      "loss": 3.5633,
+      "epoch": 2.30188679245283,
+      "grad_norm": 0.6791864633560181,
+      "learning_rate": 0.00046243065299514293,
+      "loss": 3.5651,
       "step": 21350
     },
     {
-      "epoch": 2.303304272952319,
-      "grad_norm": 0.6282955408096313,
-      "learning_rate": 0.00046235103975864663,
-      "loss": 3.5426,
+      "epoch": 2.3072776280323453,
+      "grad_norm": 0.5625030994415283,
+      "learning_rate": 0.00046210685375067454,
+      "loss": 3.559,
       "step": 21400
     },
     {
-      "epoch": 2.3086858249919278,
-      "grad_norm": 0.5589568018913269,
-      "learning_rate": 0.0004620277987285852,
-      "loss": 3.5467,
+      "epoch": 2.31266846361186,
+      "grad_norm": 0.5689521431922913,
+      "learning_rate": 0.0004617830545062061,
+      "loss": 3.5834,
       "step": 21450
     },
     {
-      "epoch": 2.314067377031536,
-      "grad_norm": 0.5827752947807312,
-      "learning_rate": 0.0004617045576985239,
-      "loss": 3.5473,
+      "epoch": 2.3180592991913747,
+      "grad_norm": 0.5639389157295227,
+      "learning_rate": 0.0004614592552617377,
+      "loss": 3.5542,
       "step": 21500
     },
     {
-      "epoch": 2.319448929071144,
-      "grad_norm": 0.5686028599739075,
-      "learning_rate": 0.0004613813166684624,
-      "loss": 3.5631,
+      "epoch": 2.3234501347708894,
+      "grad_norm": 0.5893327593803406,
+      "learning_rate": 0.00046113545601726924,
+      "loss": 3.5697,
       "step": 21550
     },
     {
-      "epoch": 2.3248304811107525,
-      "grad_norm": 0.626375675201416,
-      "learning_rate": 0.000461058075638401,
-      "loss": 3.5445,
+      "epoch": 2.3288409703504045,
+      "grad_norm": 0.6247137784957886,
+      "learning_rate": 0.00046081165677280085,
+      "loss": 3.5612,
       "step": 21600
     },
     {
-      "epoch": 2.3302120331503606,
-      "grad_norm": 0.5661760568618774,
-      "learning_rate": 0.00046073483460833955,
-      "loss": 3.557,
+      "epoch": 2.334231805929919,
+      "grad_norm": 0.6741095185279846,
+      "learning_rate": 0.0004604878575283324,
+      "loss": 3.5619,
       "step": 21650
     },
     {
-      "epoch": 2.3355935851899687,
-      "grad_norm": 0.6052250862121582,
-      "learning_rate": 0.00046041159357827814,
-      "loss": 3.5683,
+      "epoch": 2.339622641509434,
+      "grad_norm": 0.6737405061721802,
+      "learning_rate": 0.00046016405828386395,
+      "loss": 3.5814,
       "step": 21700
     },
     {
-      "epoch": 2.340975137229577,
-      "grad_norm": 0.5732718110084534,
-      "learning_rate": 0.00046008835254821674,
-      "loss": 3.5631,
+      "epoch": 2.3450134770889486,
+      "grad_norm": 0.5989480018615723,
+      "learning_rate": 0.00045984025903939555,
+      "loss": 3.5692,
       "step": 21750
     },
     {
-      "epoch": 2.3463566892691854,
-      "grad_norm": 0.5140211582183838,
-      "learning_rate": 0.0004597651115181554,
-      "loss": 3.5487,
+      "epoch": 2.3504043126684637,
+      "grad_norm": 0.6491212844848633,
+      "learning_rate": 0.0004595164597949271,
+      "loss": 3.552,
       "step": 21800
     },
     {
-      "epoch": 2.3517382413087935,
-      "grad_norm": 0.5432984828948975,
-      "learning_rate": 0.00045944187048809393,
-      "loss": 3.5413,
+      "epoch": 2.3557951482479784,
+      "grad_norm": 0.5892617702484131,
+      "learning_rate": 0.0004591926605504587,
+      "loss": 3.5614,
       "step": 21850
     },
     {
-      "epoch": 2.3571197933484016,
-      "grad_norm": 0.5777596235275269,
-      "learning_rate": 0.0004591186294580325,
-      "loss": 3.5418,
+      "epoch": 2.361185983827493,
+      "grad_norm": 0.6541824340820312,
+      "learning_rate": 0.0004588688613059902,
+      "loss": 3.5572,
       "step": 21900
     },
     {
-      "epoch": 2.3625013453880097,
-      "grad_norm": 0.5614270567893982,
-      "learning_rate": 0.00045879538842797106,
-      "loss": 3.5598,
+      "epoch": 2.3665768194070083,
+      "grad_norm": 0.5727041959762573,
+      "learning_rate": 0.00045854506206152186,
+      "loss": 3.5606,
       "step": 21950
     },
     {
-      "epoch": 2.3678828974276183,
-      "grad_norm": 0.6078837513923645,
-      "learning_rate": 0.00045847214739790966,
-      "loss": 3.5513,
+      "epoch": 2.371967654986523,
+      "grad_norm": 0.5565124750137329,
+      "learning_rate": 0.00045822126281705336,
+      "loss": 3.576,
       "step": 22000
     },
     {
-      "epoch": 2.3678828974276183,
-      "eval_accuracy": 0.3652106056953019,
-      "eval_loss": 3.5474283695220947,
-      "eval_runtime": 186.5561,
-      "eval_samples_per_second": 96.545,
-      "eval_steps_per_second": 6.036,
+      "epoch": 2.371967654986523,
+      "eval_accuracy": 0.3642171923420141,
+      "eval_loss": 3.5566956996917725,
+      "eval_runtime": 184.391,
+      "eval_samples_per_second": 97.678,
+      "eval_steps_per_second": 6.107,
       "step": 22000
     },
     {
-      "epoch": 2.3732644494672264,
-      "grad_norm": 0.6337881088256836,
-      "learning_rate": 0.0004581489063678482,
-      "loss": 3.5351,
+      "epoch": 2.3773584905660377,
+      "grad_norm": 0.5506110787391663,
+      "learning_rate": 0.00045789746357258497,
+      "loss": 3.5556,
       "step": 22050
     },
     {
-      "epoch": 2.3786460015068345,
-      "grad_norm": 0.5621966123580933,
-      "learning_rate": 0.00045782566533778685,
-      "loss": 3.5375,
+      "epoch": 2.382749326145553,
+      "grad_norm": 0.6396636366844177,
+      "learning_rate": 0.0004575736643281165,
+      "loss": 3.5576,
       "step": 22100
     },
     {
-      "epoch": 2.384027553546443,
-      "grad_norm": 0.5724130868911743,
-      "learning_rate": 0.00045750242430772544,
-      "loss": 3.5452,
+      "epoch": 2.3881401617250675,
+      "grad_norm": 0.9475221633911133,
+      "learning_rate": 0.00045724986508364807,
+      "loss": 3.5596,
       "step": 22150
     },
     {
-      "epoch": 2.389409105586051,
-      "grad_norm": 0.6550247669219971,
-      "learning_rate": 0.000457179183277664,
-      "loss": 3.5726,
+      "epoch": 2.393530997304582,
+      "grad_norm": 0.593930184841156,
+      "learning_rate": 0.00045692606583917967,
+      "loss": 3.5707,
       "step": 22200
     },
     {
-      "epoch": 2.3947906576256592,
-      "grad_norm": 0.5965059995651245,
-      "learning_rate": 0.0004568559422476026,
-      "loss": 3.5719,
+      "epoch": 2.398921832884097,
+      "grad_norm": 0.5992391109466553,
+      "learning_rate": 0.0004566087425796006,
+      "loss": 3.5759,
       "step": 22250
     },
     {
-      "epoch": 2.4001722096652673,
-      "grad_norm": 0.5822271704673767,
-      "learning_rate": 0.00045653270121754117,
-      "loss": 3.5463,
+      "epoch": 2.404312668463612,
+      "grad_norm": 0.5973031520843506,
+      "learning_rate": 0.0004562849433351322,
+      "loss": 3.5788,
       "step": 22300
     },
     {
-      "epoch": 2.4055537617048754,
-      "grad_norm": 0.6009593605995178,
-      "learning_rate": 0.0004562094601874797,
-      "loss": 3.5527,
+      "epoch": 2.4097035040431267,
+      "grad_norm": 0.603207528591156,
+      "learning_rate": 0.0004559611440906638,
+      "loss": 3.5577,
       "step": 22350
     },
     {
-      "epoch": 2.410935313744484,
-      "grad_norm": 0.5658345818519592,
-      "learning_rate": 0.00045588621915741836,
-      "loss": 3.5515,
+      "epoch": 2.4150943396226414,
+      "grad_norm": 0.5866345763206482,
+      "learning_rate": 0.00045563734484619533,
+      "loss": 3.5796,
       "step": 22400
     },
     {
-      "epoch": 2.416316865784092,
-      "grad_norm": 0.5815855860710144,
-      "learning_rate": 0.00045556297812735696,
-      "loss": 3.5443,
+      "epoch": 2.420485175202156,
+      "grad_norm": 0.6198456883430481,
+      "learning_rate": 0.0004553135456017269,
+      "loss": 3.5699,
       "step": 22450
     },
     {
-      "epoch": 2.4216984178237,
-      "grad_norm": 0.5852071046829224,
-      "learning_rate": 0.0004552397370972955,
-      "loss": 3.5473,
+      "epoch": 2.4258760107816713,
+      "grad_norm": 0.5642961263656616,
+      "learning_rate": 0.0004549897463572585,
+      "loss": 3.5644,
       "step": 22500
     },
     {
-      "epoch": 2.4270799698633088,
-      "grad_norm": 0.5829253196716309,
-      "learning_rate": 0.0004549164960672341,
-      "loss": 3.5402,
+      "epoch": 2.431266846361186,
+      "grad_norm": 0.5960258841514587,
+      "learning_rate": 0.00045466594711279,
+      "loss": 3.5583,
       "step": 22550
     },
     {
-      "epoch": 2.432461521902917,
-      "grad_norm": 0.5852131843566895,
-      "learning_rate": 0.00045459325503717263,
-      "loss": 3.5496,
+      "epoch": 2.4366576819407006,
+      "grad_norm": 0.5633063316345215,
+      "learning_rate": 0.00045434214786832164,
+      "loss": 3.5587,
       "step": 22600
     },
     {
-      "epoch": 2.437843073942525,
-      "grad_norm": 0.5861826539039612,
-      "learning_rate": 0.0004542700140071113,
-      "loss": 3.556,
+      "epoch": 2.442048517520216,
+      "grad_norm": 0.5709449052810669,
+      "learning_rate": 0.00045401834862385314,
+      "loss": 3.5473,
       "step": 22650
     },
     {
-      "epoch": 2.443224625982133,
-      "grad_norm": 0.5537131428718567,
-      "learning_rate": 0.0004539467729770499,
-      "loss": 3.5444,
+      "epoch": 2.4474393530997305,
+      "grad_norm": 0.6025438904762268,
+      "learning_rate": 0.00045369454937938474,
+      "loss": 3.5499,
       "step": 22700
     },
     {
-      "epoch": 2.4486061780217416,
-      "grad_norm": 0.5770830512046814,
-      "learning_rate": 0.0004536235319469884,
-      "loss": 3.5466,
+      "epoch": 2.452830188679245,
+      "grad_norm": 0.5747271180152893,
+      "learning_rate": 0.0004533707501349163,
+      "loss": 3.5534,
       "step": 22750
     },
     {
-      "epoch": 2.4539877300613497,
-      "grad_norm": 0.5427812337875366,
-      "learning_rate": 0.000453300290916927,
-      "loss": 3.5351,
+      "epoch": 2.4582210242587603,
+      "grad_norm": 0.5328289270401001,
+      "learning_rate": 0.0004530469508904479,
+      "loss": 3.5517,
       "step": 22800
     },
     {
-      "epoch": 2.459369282100958,
-      "grad_norm": 0.5419933199882507,
-      "learning_rate": 0.0004529770498868656,
-      "loss": 3.5458,
+      "epoch": 2.463611859838275,
+      "grad_norm": 0.6371007561683655,
+      "learning_rate": 0.00045272315164597945,
+      "loss": 3.5594,
       "step": 22850
     },
     {
-      "epoch": 2.464750834140566,
-      "grad_norm": 0.5840222239494324,
-      "learning_rate": 0.00045265380885680414,
-      "loss": 3.5407,
+      "epoch": 2.4690026954177897,
+      "grad_norm": 0.6010317206382751,
+      "learning_rate": 0.000452399352401511,
+      "loss": 3.5687,
       "step": 22900
     },
     {
-      "epoch": 2.4701323861801745,
-      "grad_norm": 0.5505115985870361,
-      "learning_rate": 0.0004523305678267428,
-      "loss": 3.5535,
+      "epoch": 2.4743935309973044,
+      "grad_norm": 0.6460767984390259,
+      "learning_rate": 0.0004520755531570426,
+      "loss": 3.5696,
       "step": 22950
     },
     {
-      "epoch": 2.4755139382197826,
-      "grad_norm": 0.5969367623329163,
-      "learning_rate": 0.0004520073267966814,
-      "loss": 3.5652,
+      "epoch": 2.4797843665768196,
+      "grad_norm": 0.5525557994842529,
+      "learning_rate": 0.00045175175391257415,
+      "loss": 3.5453,
       "step": 23000
     },
     {
-      "epoch": 2.4755139382197826,
-      "eval_accuracy": 0.3664607658224953,
-      "eval_loss": 3.5371267795562744,
-      "eval_runtime": 186.112,
-      "eval_samples_per_second": 96.775,
-      "eval_steps_per_second": 6.05,
+      "epoch": 2.4797843665768196,
+      "eval_accuracy": 0.36503632646324746,
+      "eval_loss": 3.544049024581909,
+      "eval_runtime": 184.0204,
+      "eval_samples_per_second": 97.875,
+      "eval_steps_per_second": 6.119,
       "step": 23000
     },
     {
-      "epoch": 2.4808954902593907,
-      "grad_norm": 0.5622343420982361,
-      "learning_rate": 0.00045168408576661993,
-      "loss": 3.5457,
+      "epoch": 2.4851752021563343,
+      "grad_norm": 0.5829729437828064,
+      "learning_rate": 0.00045142795466810576,
+      "loss": 3.549,
       "step": 23050
     },
     {
-      "epoch": 2.4862770422989993,
-      "grad_norm": 0.5596908330917358,
-      "learning_rate": 0.0004513608447365585,
-      "loss": 3.5303,
+      "epoch": 2.490566037735849,
+      "grad_norm": 0.6070571541786194,
+      "learning_rate": 0.0004511041554236373,
+      "loss": 3.5662,
       "step": 23100
     },
     {
-      "epoch": 2.4916585943386074,
-      "grad_norm": 0.5408734083175659,
-      "learning_rate": 0.00045103760370649706,
-      "loss": 3.5314,
+      "epoch": 2.4959568733153636,
+      "grad_norm": 0.5864456295967102,
+      "learning_rate": 0.0004507803561791689,
+      "loss": 3.5647,
       "step": 23150
     },
     {
-      "epoch": 2.4970401463782155,
-      "grad_norm": 0.5756362676620483,
-      "learning_rate": 0.00045071436267643566,
-      "loss": 3.5367,
+      "epoch": 2.501347708894879,
+      "grad_norm": 0.588222861289978,
+      "learning_rate": 0.00045045655693470046,
+      "loss": 3.5599,
       "step": 23200
     },
     {
-      "epoch": 2.5024216984178236,
-      "grad_norm": 0.5337349772453308,
-      "learning_rate": 0.0004503911216463743,
-      "loss": 3.5426,
+      "epoch": 2.5067385444743935,
+      "grad_norm": 0.5665920376777649,
+      "learning_rate": 0.00045013275769023207,
+      "loss": 3.5535,
       "step": 23250
     },
     {
-      "epoch": 2.5078032504574317,
-      "grad_norm": 0.5681710839271545,
-      "learning_rate": 0.00045006788061631285,
-      "loss": 3.5481,
+      "epoch": 2.512129380053908,
+      "grad_norm": 0.6056466698646545,
+      "learning_rate": 0.00044980895844576356,
+      "loss": 3.5534,
       "step": 23300
     },
     {
-      "epoch": 2.5131848024970402,
-      "grad_norm": 0.5921882390975952,
-      "learning_rate": 0.00044974463958625144,
-      "loss": 3.5448,
+      "epoch": 2.5175202156334233,
+      "grad_norm": 0.6521342992782593,
+      "learning_rate": 0.0004494851592012951,
+      "loss": 3.5481,
       "step": 23350
     },
     {
-      "epoch": 2.5185663545366483,
-      "grad_norm": 0.5589230060577393,
-      "learning_rate": 0.00044942139855619004,
-      "loss": 3.5522,
+      "epoch": 2.522911051212938,
+      "grad_norm": 0.5913186073303223,
+      "learning_rate": 0.0004491613599568267,
+      "loss": 3.5393,
       "step": 23400
     },
     {
-      "epoch": 2.5239479065762565,
-      "grad_norm": 0.57844477891922,
-      "learning_rate": 0.0004490981575261286,
-      "loss": 3.5336,
+      "epoch": 2.5283018867924527,
+      "grad_norm": 0.5629101395606995,
+      "learning_rate": 0.00044884403669724767,
+      "loss": 3.5532,
       "step": 23450
     },
     {
-      "epoch": 2.529329458615865,
-      "grad_norm": 0.6227934956550598,
-      "learning_rate": 0.0004487749164960672,
-      "loss": 3.5706,
+      "epoch": 2.533692722371968,
+      "grad_norm": 0.5822175145149231,
+      "learning_rate": 0.0004485202374527792,
+      "loss": 3.5478,
       "step": 23500
     },
     {
-      "epoch": 2.534711010655473,
-      "grad_norm": 0.5843015313148499,
-      "learning_rate": 0.0004484516754660058,
-      "loss": 3.5292,
+      "epoch": 2.5390835579514826,
+      "grad_norm": 0.5633822083473206,
+      "learning_rate": 0.00044819643820831083,
+      "loss": 3.5597,
       "step": 23550
     },
     {
-      "epoch": 2.540092562695081,
-      "grad_norm": 0.5650943517684937,
-      "learning_rate": 0.00044812843443594436,
-      "loss": 3.5369,
+      "epoch": 2.5444743935309972,
+      "grad_norm": 0.597368061542511,
+      "learning_rate": 0.0004478726389638424,
+      "loss": 3.557,
       "step": 23600
     },
     {
-      "epoch": 2.5454741147346893,
-      "grad_norm": 0.5545888543128967,
-      "learning_rate": 0.00044780519340588296,
-      "loss": 3.541,
+      "epoch": 2.5498652291105124,
+      "grad_norm": 0.6209232211112976,
+      "learning_rate": 0.00044754883971937393,
+      "loss": 3.563,
       "step": 23650
     },
     {
-      "epoch": 2.550855666774298,
-      "grad_norm": 0.5571039319038391,
-      "learning_rate": 0.0004474819523758215,
-      "loss": 3.5518,
+      "epoch": 2.555256064690027,
+      "grad_norm": 0.5570722818374634,
+      "learning_rate": 0.00044722504047490553,
+      "loss": 3.5715,
       "step": 23700
     },
     {
-      "epoch": 2.556237218813906,
-      "grad_norm": 0.6183195114135742,
-      "learning_rate": 0.0004471587113457601,
-      "loss": 3.5557,
+      "epoch": 2.560646900269542,
+      "grad_norm": 0.5941867232322693,
+      "learning_rate": 0.0004469012412304371,
+      "loss": 3.5777,
       "step": 23750
     },
     {
-      "epoch": 2.561618770853514,
-      "grad_norm": 0.6002625823020935,
-      "learning_rate": 0.00044683547031569874,
-      "loss": 3.5347,
+      "epoch": 2.5660377358490565,
+      "grad_norm": 0.6936694383621216,
+      "learning_rate": 0.0004465774419859687,
+      "loss": 3.5533,
       "step": 23800
     },
     {
-      "epoch": 2.567000322893122,
-      "grad_norm": 0.5983108282089233,
-      "learning_rate": 0.0004465122292856373,
-      "loss": 3.5511,
+      "epoch": 2.571428571428571,
+      "grad_norm": 0.6152774691581726,
+      "learning_rate": 0.00044625364274150024,
+      "loss": 3.567,
       "step": 23850
     },
     {
-      "epoch": 2.5723818749327307,
-      "grad_norm": 0.6042992472648621,
-      "learning_rate": 0.0004461889882555759,
-      "loss": 3.556,
+      "epoch": 2.5768194070080863,
+      "grad_norm": 0.6012356281280518,
+      "learning_rate": 0.00044592984349703184,
+      "loss": 3.5503,
       "step": 23900
     },
     {
-      "epoch": 2.577763426972339,
-      "grad_norm": 0.6030369400978088,
-      "learning_rate": 0.00044586574722551447,
-      "loss": 3.547,
+      "epoch": 2.582210242587601,
+      "grad_norm": 0.5413028001785278,
+      "learning_rate": 0.00044560604425256334,
+      "loss": 3.5552,
       "step": 23950
     },
     {
-      "epoch": 2.583144979011947,
-      "grad_norm": 0.5858563184738159,
-      "learning_rate": 0.000445542506195453,
-      "loss": 3.537,
+      "epoch": 2.5876010781671157,
+      "grad_norm": 0.618615984916687,
+      "learning_rate": 0.000445282245008095,
+      "loss": 3.5399,
       "step": 24000
     },
     {
-      "epoch": 2.583144979011947,
-      "eval_accuracy": 0.3673491118332665,
-      "eval_loss": 3.5251595973968506,
-      "eval_runtime": 186.7742,
-      "eval_samples_per_second": 96.432,
-      "eval_steps_per_second": 6.029,
+      "epoch": 2.5876010781671157,
+      "eval_accuracy": 0.36634917930670985,
+      "eval_loss": 3.534419298171997,
+      "eval_runtime": 184.3338,
+      "eval_samples_per_second": 97.709,
+      "eval_steps_per_second": 6.108,
       "step": 24000
     },
     {
-      "epoch": 2.5885265310515555,
-      "grad_norm": 0.5575674772262573,
-      "learning_rate": 0.0004452192651653916,
-      "loss": 3.5281,
+      "epoch": 2.592991913746631,
+      "grad_norm": 0.5585067272186279,
+      "learning_rate": 0.0004449584457636265,
+      "loss": 3.5518,
       "step": 24050
     },
     {
-      "epoch": 2.5939080830911636,
-      "grad_norm": 0.5835270285606384,
-      "learning_rate": 0.00044489602413533025,
-      "loss": 3.5237,
+      "epoch": 2.5983827493261455,
+      "grad_norm": 0.6337313055992126,
+      "learning_rate": 0.00044463464651915805,
+      "loss": 3.5448,
       "step": 24100
     },
     {
-      "epoch": 2.5992896351307717,
-      "grad_norm": 0.5514588356018066,
-      "learning_rate": 0.0004445727831052688,
-      "loss": 3.5504,
+      "epoch": 2.6037735849056602,
+      "grad_norm": 0.5886608958244324,
+      "learning_rate": 0.00044431084727468965,
+      "loss": 3.573,
       "step": 24150
     },
     {
-      "epoch": 2.60467118717038,
-      "grad_norm": 0.5781415700912476,
-      "learning_rate": 0.0004442495420752074,
-      "loss": 3.5462,
+      "epoch": 2.6091644204851754,
+      "grad_norm": 0.6241118311882019,
+      "learning_rate": 0.0004439870480302212,
+      "loss": 3.5376,
       "step": 24200
     },
     {
-      "epoch": 2.610052739209988,
-      "grad_norm": 0.5353069305419922,
-      "learning_rate": 0.00044392630104514593,
-      "loss": 3.5722,
+      "epoch": 2.61455525606469,
+      "grad_norm": 0.6280027627944946,
+      "learning_rate": 0.0004436632487857528,
+      "loss": 3.5471,
       "step": 24250
     },
     {
-      "epoch": 2.6154342912495965,
-      "grad_norm": 0.5822466015815735,
-      "learning_rate": 0.0004436030600150845,
-      "loss": 3.535,
+      "epoch": 2.6199460916442048,
+      "grad_norm": 0.5847042202949524,
+      "learning_rate": 0.00044333944954128436,
+      "loss": 3.5562,
       "step": 24300
     },
     {
-      "epoch": 2.6208158432892046,
-      "grad_norm": 0.5951100587844849,
-      "learning_rate": 0.0004432798189850231,
-      "loss": 3.5347,
+      "epoch": 2.62533692722372,
+      "grad_norm": 0.5288570523262024,
+      "learning_rate": 0.00044301565029681596,
+      "loss": 3.57,
       "step": 24350
     },
     {
-      "epoch": 2.6261973953288127,
-      "grad_norm": 0.5432897806167603,
-      "learning_rate": 0.0004429565779549617,
-      "loss": 3.5184,
+      "epoch": 2.6307277628032346,
+      "grad_norm": 0.59012371301651,
+      "learning_rate": 0.0004426918510523475,
+      "loss": 3.5334,
       "step": 24400
     },
     {
-      "epoch": 2.6315789473684212,
-      "grad_norm": 0.5444647669792175,
-      "learning_rate": 0.0004426333369249003,
-      "loss": 3.5355,
+      "epoch": 2.6361185983827493,
+      "grad_norm": 0.5806400179862976,
+      "learning_rate": 0.0004423680518078791,
+      "loss": 3.551,
       "step": 24450
     },
     {
-      "epoch": 2.6369604994080293,
-      "grad_norm": 0.5889618992805481,
-      "learning_rate": 0.0004423100958948389,
-      "loss": 3.5454,
+      "epoch": 2.641509433962264,
+      "grad_norm": 0.5785780549049377,
+      "learning_rate": 0.00044204425256341067,
+      "loss": 3.5669,
       "step": 24500
     },
     {
-      "epoch": 2.6423420514476375,
-      "grad_norm": 0.5954281687736511,
-      "learning_rate": 0.00044198685486477744,
-      "loss": 3.543,
+      "epoch": 2.6469002695417787,
+      "grad_norm": 0.5869790315628052,
+      "learning_rate": 0.0004417204533189422,
+      "loss": 3.535,
       "step": 24550
     },
     {
-      "epoch": 2.6477236034872456,
-      "grad_norm": 0.5864471793174744,
-      "learning_rate": 0.00044166361383471604,
-      "loss": 3.5413,
+      "epoch": 2.652291105121294,
+      "grad_norm": 0.5677444338798523,
+      "learning_rate": 0.0004413966540744738,
+      "loss": 3.5402,
       "step": 24600
     },
     {
-      "epoch": 2.653105155526854,
-      "grad_norm": 0.5580281615257263,
-      "learning_rate": 0.0004413403728046547,
-      "loss": 3.5457,
+      "epoch": 2.6576819407008085,
+      "grad_norm": 0.556079089641571,
+      "learning_rate": 0.0004410728548300053,
+      "loss": 3.5434,
       "step": 24650
     },
     {
-      "epoch": 2.658486707566462,
-      "grad_norm": 0.5743160247802734,
-      "learning_rate": 0.0004410171317745932,
-      "loss": 3.5406,
+      "epoch": 2.6630727762803232,
+      "grad_norm": 0.5939511060714722,
+      "learning_rate": 0.0004407490555855369,
+      "loss": 3.5485,
       "step": 24700
     },
     {
-      "epoch": 2.6638682596060703,
-      "grad_norm": 0.5307379364967346,
-      "learning_rate": 0.0004406938907445318,
-      "loss": 3.536,
+      "epoch": 2.6684636118598384,
+      "grad_norm": 0.6414326429367065,
+      "learning_rate": 0.0004404252563410685,
+      "loss": 3.5461,
       "step": 24750
     },
     {
-      "epoch": 2.6692498116456784,
-      "grad_norm": 0.6348727941513062,
-      "learning_rate": 0.00044037064971447036,
-      "loss": 3.5258,
+      "epoch": 2.673854447439353,
+      "grad_norm": 0.5826203227043152,
+      "learning_rate": 0.0004401014570966001,
+      "loss": 3.5242,
       "step": 24800
     },
     {
-      "epoch": 2.674631363685287,
-      "grad_norm": 0.5537613034248352,
-      "learning_rate": 0.00044004740868440896,
-      "loss": 3.5351,
+      "epoch": 2.6792452830188678,
+      "grad_norm": 0.5997711420059204,
+      "learning_rate": 0.00043977765785213163,
+      "loss": 3.5213,
       "step": 24850
     },
     {
-      "epoch": 2.680012915724895,
-      "grad_norm": 0.5886291265487671,
-      "learning_rate": 0.00043972416765434755,
-      "loss": 3.5389,
+      "epoch": 2.684636118598383,
+      "grad_norm": 0.575883150100708,
+      "learning_rate": 0.0004394538586076632,
+      "loss": 3.5609,
       "step": 24900
     },
     {
-      "epoch": 2.685394467764503,
-      "grad_norm": 0.6198649406433105,
-      "learning_rate": 0.00043940092662428615,
-      "loss": 3.5461,
+      "epoch": 2.6900269541778976,
+      "grad_norm": 0.6031562685966492,
+      "learning_rate": 0.0004391300593631948,
+      "loss": 3.5447,
       "step": 24950
     },
     {
-      "epoch": 2.6907760198041117,
-      "grad_norm": 0.6090601086616516,
-      "learning_rate": 0.00043907768559422474,
-      "loss": 3.5369,
+      "epoch": 2.6954177897574123,
+      "grad_norm": 0.586550772190094,
+      "learning_rate": 0.00043880626011872634,
+      "loss": 3.5352,
       "step": 25000
     },
     {
-      "epoch": 2.6907760198041117,
-      "eval_accuracy": 0.36792855768398114,
-      "eval_loss": 3.518705368041992,
-      "eval_runtime": 186.1223,
-      "eval_samples_per_second": 96.77,
-      "eval_steps_per_second": 6.05,
+      "epoch": 2.6954177897574123,
+      "eval_accuracy": 0.3670280425497747,
+      "eval_loss": 3.525153875350952,
+      "eval_runtime": 184.0939,
+      "eval_samples_per_second": 97.836,
+      "eval_steps_per_second": 6.116,
       "step": 25000
     },
     {
-      "epoch": 2.69615757184372,
-      "grad_norm": 0.6166929602622986,
-      "learning_rate": 0.00043875444456416334,
-      "loss": 3.5363,
+      "epoch": 2.7008086253369274,
+      "grad_norm": 0.5802037119865417,
+      "learning_rate": 0.00043848246087425794,
+      "loss": 3.5446,
       "step": 25050
     },
     {
-      "epoch": 2.701539123883328,
-      "grad_norm": 0.5990306735038757,
-      "learning_rate": 0.00043843766835470314,
-      "loss": 3.5467,
+      "epoch": 2.706199460916442,
+      "grad_norm": 0.5433179140090942,
+      "learning_rate": 0.0004381586616297895,
+      "loss": 3.5309,
       "step": 25100
     },
     {
-      "epoch": 2.706920675922936,
-      "grad_norm": 0.558195948600769,
-      "learning_rate": 0.0004381144273246417,
-      "loss": 3.5298,
+      "epoch": 2.711590296495957,
+      "grad_norm": 0.5975437760353088,
+      "learning_rate": 0.0004378348623853211,
+      "loss": 3.5202,
       "step": 25150
     },
     {
-      "epoch": 2.712302227962544,
-      "grad_norm": 0.5733847618103027,
-      "learning_rate": 0.0004377911862945803,
-      "loss": 3.5479,
+      "epoch": 2.7169811320754715,
+      "grad_norm": 0.5662147998809814,
+      "learning_rate": 0.00043751106314085265,
+      "loss": 3.5257,
       "step": 25200
     },
     {
-      "epoch": 2.7176837800021527,
-      "grad_norm": 0.5749377012252808,
-      "learning_rate": 0.00043746794526451887,
-      "loss": 3.531,
+      "epoch": 2.7223719676549867,
+      "grad_norm": 0.6542898416519165,
+      "learning_rate": 0.00043718726389638425,
+      "loss": 3.5596,
       "step": 25250
     },
     {
-      "epoch": 2.723065332041761,
-      "grad_norm": 0.6088876128196716,
-      "learning_rate": 0.00043714470423445747,
-      "loss": 3.5262,
+      "epoch": 2.7277628032345014,
+      "grad_norm": 0.5950943827629089,
+      "learning_rate": 0.00043686346465191575,
+      "loss": 3.5186,
       "step": 25300
     },
     {
-      "epoch": 2.728446884081369,
-      "grad_norm": 0.5567477941513062,
-      "learning_rate": 0.00043682146320439606,
-      "loss": 3.5497,
+      "epoch": 2.733153638814016,
+      "grad_norm": 0.6437334418296814,
+      "learning_rate": 0.0004365396654074473,
+      "loss": 3.5431,
       "step": 25350
     },
     {
-      "epoch": 2.7338284361209775,
-      "grad_norm": 0.569877028465271,
-      "learning_rate": 0.00043649822217433466,
-      "loss": 3.5193,
+      "epoch": 2.7385444743935308,
+      "grad_norm": 0.6011313796043396,
+      "learning_rate": 0.0004362158661629789,
+      "loss": 3.5453,
       "step": 25400
     },
     {
-      "epoch": 2.7392099881605856,
-      "grad_norm": 0.6627097129821777,
-      "learning_rate": 0.0004361749811442732,
-      "loss": 3.5294,
+      "epoch": 2.743935309973046,
+      "grad_norm": 0.5815936326980591,
+      "learning_rate": 0.00043589206691851045,
+      "loss": 3.5161,
       "step": 25450
     },
     {
-      "epoch": 2.7445915402001937,
-      "grad_norm": 0.5579299330711365,
-      "learning_rate": 0.0004358517401142118,
-      "loss": 3.5532,
+      "epoch": 2.7493261455525606,
+      "grad_norm": 0.5691853165626526,
+      "learning_rate": 0.00043556826767404206,
+      "loss": 3.5771,
       "step": 25500
     },
     {
-      "epoch": 2.749973092239802,
-      "grad_norm": 0.5688357353210449,
-      "learning_rate": 0.00043552849908415033,
-      "loss": 3.525,
+      "epoch": 2.7547169811320753,
+      "grad_norm": 0.5867630839347839,
+      "learning_rate": 0.0004352444684295736,
+      "loss": 3.5647,
       "step": 25550
     },
     {
-      "epoch": 2.7553546442794103,
-      "grad_norm": 0.5659778714179993,
-      "learning_rate": 0.000435205258054089,
-      "loss": 3.5468,
+      "epoch": 2.7601078167115904,
+      "grad_norm": 0.6391850709915161,
+      "learning_rate": 0.0004349206691851052,
+      "loss": 3.5414,
       "step": 25600
     },
     {
-      "epoch": 2.7607361963190185,
-      "grad_norm": 0.6074382066726685,
-      "learning_rate": 0.0004348820170240276,
-      "loss": 3.5307,
+      "epoch": 2.765498652291105,
+      "grad_norm": 0.661693274974823,
+      "learning_rate": 0.00043459686994063676,
+      "loss": 3.5314,
       "step": 25650
     },
     {
-      "epoch": 2.7661177483586266,
-      "grad_norm": 0.5789358615875244,
-      "learning_rate": 0.0004345587759939661,
-      "loss": 3.5151,
+      "epoch": 2.77088948787062,
+      "grad_norm": 0.6009679436683655,
+      "learning_rate": 0.00043427307069616837,
+      "loss": 3.549,
       "step": 25700
     },
     {
-      "epoch": 2.7714993003982347,
-      "grad_norm": 0.5493772029876709,
-      "learning_rate": 0.0004342355349639047,
-      "loss": 3.5222,
+      "epoch": 2.776280323450135,
+      "grad_norm": 0.6037169098854065,
+      "learning_rate": 0.0004339492714516999,
+      "loss": 3.5369,
       "step": 25750
     },
     {
-      "epoch": 2.776880852437843,
-      "grad_norm": 0.6333332061767578,
-      "learning_rate": 0.0004339122939338433,
-      "loss": 3.5393,
+      "epoch": 2.7816711590296497,
+      "grad_norm": 0.5839231014251709,
+      "learning_rate": 0.00043362547220723147,
+      "loss": 3.5419,
       "step": 25800
     },
     {
-      "epoch": 2.7822624044774513,
-      "grad_norm": 0.5923963189125061,
-      "learning_rate": 0.00043358905290378184,
-      "loss": 3.5243,
+      "epoch": 2.7870619946091644,
+      "grad_norm": 0.5970977544784546,
+      "learning_rate": 0.0004333016729627631,
+      "loss": 3.5301,
       "step": 25850
     },
     {
-      "epoch": 2.7876439565170594,
-      "grad_norm": 0.5590444207191467,
-      "learning_rate": 0.0004332658118737205,
-      "loss": 3.53,
+      "epoch": 2.7924528301886795,
+      "grad_norm": 0.6259117126464844,
+      "learning_rate": 0.0004329778737182946,
+      "loss": 3.537,
       "step": 25900
     },
     {
-      "epoch": 2.793025508556668,
-      "grad_norm": 0.5920925736427307,
-      "learning_rate": 0.0004329425708436591,
-      "loss": 3.528,
+      "epoch": 2.797843665768194,
+      "grad_norm": 0.5748756527900696,
+      "learning_rate": 0.00043265407447382623,
+      "loss": 3.5275,
       "step": 25950
     },
     {
-      "epoch": 2.798407060596276,
-      "grad_norm": 0.5606895089149475,
-      "learning_rate": 0.00043261932981359763,
-      "loss": 3.5182,
+      "epoch": 2.803234501347709,
+      "grad_norm": 0.6004160642623901,
+      "learning_rate": 0.0004323302752293577,
+      "loss": 3.5378,
       "step": 26000
     },
     {
-      "epoch": 2.798407060596276,
-      "eval_accuracy": 0.36917252459656913,
-      "eval_loss": 3.5083131790161133,
-      "eval_runtime": 186.7819,
-      "eval_samples_per_second": 96.428,
-      "eval_steps_per_second": 6.028,
+      "epoch": 2.803234501347709,
+      "eval_accuracy": 0.3678886820741532,
+      "eval_loss": 3.516509771347046,
+      "eval_runtime": 184.4425,
+      "eval_samples_per_second": 97.651,
+      "eval_steps_per_second": 6.105,
       "step": 26000
     },
     {
-      "epoch": 2.803788612635884,
-      "grad_norm": 0.5823130011558533,
-      "learning_rate": 0.0004322960887835362,
-      "loss": 3.5147,
+      "epoch": 2.8086253369272236,
+      "grad_norm": 0.5950242280960083,
+      "learning_rate": 0.00043200647598488933,
+      "loss": 3.5357,
       "step": 26050
     },
     {
-      "epoch": 2.8091701646754923,
-      "grad_norm": 0.5810118317604065,
-      "learning_rate": 0.00043197284775347476,
-      "loss": 3.5203,
+      "epoch": 2.8140161725067383,
+      "grad_norm": 0.5505948662757874,
+      "learning_rate": 0.0004316826767404209,
+      "loss": 3.512,
       "step": 26100
     },
     {
-      "epoch": 2.8145517167151004,
-      "grad_norm": 0.6111186146736145,
-      "learning_rate": 0.00043164960672341336,
-      "loss": 3.5223,
+      "epoch": 2.8194070080862534,
+      "grad_norm": 0.5727338790893555,
+      "learning_rate": 0.0004313588774959525,
+      "loss": 3.5129,
       "step": 26150
     },
     {
-      "epoch": 2.819933268754709,
-      "grad_norm": 0.6507933735847473,
-      "learning_rate": 0.000431326365693352,
-      "loss": 3.5276,
+      "epoch": 2.824797843665768,
+      "grad_norm": 0.618218719959259,
+      "learning_rate": 0.00043103507825148404,
+      "loss": 3.5384,
       "step": 26200
     },
     {
-      "epoch": 2.825314820794317,
-      "grad_norm": 0.6045193672180176,
-      "learning_rate": 0.00043100312466329055,
-      "loss": 3.533,
+      "epoch": 2.830188679245283,
+      "grad_norm": 0.6344850659370422,
+      "learning_rate": 0.0004307112790070156,
+      "loss": 3.5512,
       "step": 26250
     },
     {
-      "epoch": 2.830696372833925,
-      "grad_norm": 0.6073702573776245,
-      "learning_rate": 0.00043067988363322914,
-      "loss": 3.5191,
+      "epoch": 2.835579514824798,
+      "grad_norm": 0.5566301941871643,
+      "learning_rate": 0.0004303874797625472,
+      "loss": 3.5668,
       "step": 26300
     },
     {
-      "epoch": 2.8360779248735337,
-      "grad_norm": 0.612769603729248,
-      "learning_rate": 0.00043035664260316774,
-      "loss": 3.5533,
+      "epoch": 2.8409703504043127,
+      "grad_norm": 0.758602499961853,
+      "learning_rate": 0.00043006368051807874,
+      "loss": 3.5369,
       "step": 26350
     },
     {
-      "epoch": 2.841459476913142,
-      "grad_norm": 0.5725610256195068,
-      "learning_rate": 0.0004300334015731063,
-      "loss": 3.512,
+      "epoch": 2.8463611859838274,
+      "grad_norm": 0.6691704988479614,
+      "learning_rate": 0.00042973988127361035,
+      "loss": 3.5462,
       "step": 26400
     },
     {
-      "epoch": 2.84684102895275,
-      "grad_norm": 0.648580014705658,
-      "learning_rate": 0.0004297101605430449,
-      "loss": 3.5198,
+      "epoch": 2.8517520215633425,
+      "grad_norm": 0.6524878144264221,
+      "learning_rate": 0.0004294160820291419,
+      "loss": 3.5508,
       "step": 26450
     },
     {
-      "epoch": 2.852222580992358,
-      "grad_norm": 0.536857008934021,
-      "learning_rate": 0.0004293869195129835,
-      "loss": 3.501,
+      "epoch": 2.857142857142857,
+      "grad_norm": 0.6020094752311707,
+      "learning_rate": 0.0004290922827846735,
+      "loss": 3.5225,
       "step": 26500
     },
     {
-      "epoch": 2.857604133031966,
-      "grad_norm": 0.5986590385437012,
-      "learning_rate": 0.00042906367848292206,
-      "loss": 3.529,
+      "epoch": 2.862533692722372,
+      "grad_norm": 0.5683682560920715,
+      "learning_rate": 0.00042876848354020505,
+      "loss": 3.5395,
       "step": 26550
     },
     {
-      "epoch": 2.8629856850715747,
-      "grad_norm": 0.5748111605644226,
-      "learning_rate": 0.00042874043745286066,
-      "loss": 3.5054,
+      "epoch": 2.867924528301887,
+      "grad_norm": 0.6499101519584656,
+      "learning_rate": 0.00042844468429573655,
+      "loss": 3.5475,
       "step": 26600
     },
     {
-      "epoch": 2.868367237111183,
-      "grad_norm": 0.6125283241271973,
-      "learning_rate": 0.0004284171964227992,
-      "loss": 3.5158,
+      "epoch": 2.8733153638814017,
+      "grad_norm": 0.6190754771232605,
+      "learning_rate": 0.00042812088505126815,
+      "loss": 3.5531,
       "step": 26650
     },
     {
-      "epoch": 2.873748789150791,
-      "grad_norm": 0.5569992065429688,
-      "learning_rate": 0.0004280939553927378,
-      "loss": 3.5242,
+      "epoch": 2.8787061994609164,
+      "grad_norm": 0.5818064212799072,
+      "learning_rate": 0.0004277970858067997,
+      "loss": 3.5385,
       "step": 26700
     },
     {
-      "epoch": 2.8791303411903995,
-      "grad_norm": 0.585707426071167,
-      "learning_rate": 0.00042777071436267644,
-      "loss": 3.5242,
+      "epoch": 2.884097035040431,
+      "grad_norm": 0.6224560737609863,
+      "learning_rate": 0.0004274732865623313,
+      "loss": 3.5503,
       "step": 26750
     },
     {
-      "epoch": 2.8845118932300076,
-      "grad_norm": 0.5798590183258057,
-      "learning_rate": 0.000427447473332615,
-      "loss": 3.533,
+      "epoch": 2.889487870619946,
+      "grad_norm": 0.6236782073974609,
+      "learning_rate": 0.00042714948731786286,
+      "loss": 3.5279,
       "step": 26800
     },
     {
-      "epoch": 2.8898934452696157,
-      "grad_norm": 0.5846250057220459,
-      "learning_rate": 0.0004271242323025536,
-      "loss": 3.52,
+      "epoch": 2.894878706199461,
+      "grad_norm": 0.6322186589241028,
+      "learning_rate": 0.00042682568807339447,
+      "loss": 3.5492,
       "step": 26850
     },
     {
-      "epoch": 2.895274997309224,
-      "grad_norm": 0.617246687412262,
-      "learning_rate": 0.00042680099127249217,
-      "loss": 3.5223,
+      "epoch": 2.9002695417789757,
+      "grad_norm": 0.6208633184432983,
+      "learning_rate": 0.000426501888828926,
+      "loss": 3.5154,
       "step": 26900
     },
     {
-      "epoch": 2.9006565493488323,
-      "grad_norm": 0.560801088809967,
-      "learning_rate": 0.0004264777502424307,
-      "loss": 3.5315,
+      "epoch": 2.9056603773584904,
+      "grad_norm": 0.6330786943435669,
+      "learning_rate": 0.0004261780895844576,
+      "loss": 3.5374,
       "step": 26950
     },
     {
-      "epoch": 2.9060381013884404,
-      "grad_norm": 0.5934939384460449,
-      "learning_rate": 0.0004261545092123693,
-      "loss": 3.5331,
+      "epoch": 2.9110512129380055,
+      "grad_norm": 0.556624174118042,
+      "learning_rate": 0.00042585429033998917,
+      "loss": 3.5517,
       "step": 27000
     },
     {
-      "epoch": 2.9060381013884404,
-      "eval_accuracy": 0.3697983652303805,
-      "eval_loss": 3.501643657684326,
-      "eval_runtime": 186.5576,
-      "eval_samples_per_second": 96.544,
-      "eval_steps_per_second": 6.036,
+      "epoch": 2.9110512129380055,
+      "eval_accuracy": 0.3689546755565011,
+      "eval_loss": 3.509303092956543,
+      "eval_runtime": 184.2913,
+      "eval_samples_per_second": 97.731,
+      "eval_steps_per_second": 6.11,
       "step": 27000
     },
     {
-      "epoch": 2.9114196534280485,
-      "grad_norm": 0.6355244517326355,
-      "learning_rate": 0.00042583126818230795,
-      "loss": 3.515,
+      "epoch": 2.91644204851752,
+      "grad_norm": 0.6224827170372009,
+      "learning_rate": 0.0004255304910955207,
+      "loss": 3.5358,
       "step": 27050
     },
     {
-      "epoch": 2.9168012054676566,
-      "grad_norm": 0.596826434135437,
-      "learning_rate": 0.0004255080271522465,
-      "loss": 3.5263,
+      "epoch": 2.921832884097035,
+      "grad_norm": 0.5939812064170837,
+      "learning_rate": 0.0004252066918510523,
+      "loss": 3.5405,
       "step": 27100
     },
     {
-      "epoch": 2.922182757507265,
-      "grad_norm": 0.6337636709213257,
-      "learning_rate": 0.0004251847861221851,
-      "loss": 3.5234,
+      "epoch": 2.92722371967655,
+      "grad_norm": 0.7123465538024902,
+      "learning_rate": 0.0004248828926065839,
+      "loss": 3.5368,
       "step": 27150
     },
     {
-      "epoch": 2.9275643095468733,
-      "grad_norm": 0.5752792358398438,
-      "learning_rate": 0.00042486154509212363,
-      "loss": 3.5197,
+      "epoch": 2.9326145552560647,
+      "grad_norm": 0.6271578669548035,
+      "learning_rate": 0.0004245590933621155,
+      "loss": 3.5232,
       "step": 27200
     },
     {
-      "epoch": 2.9329458615864814,
-      "grad_norm": 0.5513163208961487,
-      "learning_rate": 0.0004245383040620622,
-      "loss": 3.52,
+      "epoch": 2.9380053908355794,
+      "grad_norm": 0.6323043704032898,
+      "learning_rate": 0.00042423529411764703,
+      "loss": 3.5364,
       "step": 27250
     },
     {
-      "epoch": 2.93832741362609,
-      "grad_norm": 0.5749354362487793,
-      "learning_rate": 0.00042422152785260203,
-      "loss": 3.5245,
+      "epoch": 2.9433962264150946,
+      "grad_norm": 0.5798599123954773,
+      "learning_rate": 0.00042391149487317864,
+      "loss": 3.5404,
       "step": 27300
     },
     {
-      "epoch": 2.943708965665698,
-      "grad_norm": 0.6305385231971741,
-      "learning_rate": 0.0004238982868225406,
-      "loss": 3.5186,
+      "epoch": 2.9487870619946093,
+      "grad_norm": 0.6347798705101013,
+      "learning_rate": 0.00042358769562871013,
+      "loss": 3.526,
       "step": 27350
     },
     {
-      "epoch": 2.949090517705306,
-      "grad_norm": 0.5567276477813721,
-      "learning_rate": 0.0004235750457924793,
-      "loss": 3.5232,
+      "epoch": 2.954177897574124,
+      "grad_norm": 0.7197145819664001,
+      "learning_rate": 0.00042326389638424174,
+      "loss": 3.5176,
       "step": 27400
     },
     {
-      "epoch": 2.9544720697449143,
-      "grad_norm": 0.6510251760482788,
-      "learning_rate": 0.0004232518047624178,
-      "loss": 3.5379,
+      "epoch": 2.9595687331536387,
+      "grad_norm": 0.593965470790863,
+      "learning_rate": 0.0004229400971397733,
+      "loss": 3.5218,
       "step": 27450
     },
     {
-      "epoch": 2.9598536217845224,
-      "grad_norm": 0.5292083621025085,
-      "learning_rate": 0.0004229285637323564,
-      "loss": 3.5181,
+      "epoch": 2.964959568733154,
+      "grad_norm": 0.6081401109695435,
+      "learning_rate": 0.00042262277388019424,
+      "loss": 3.5354,
       "step": 27500
     },
     {
-      "epoch": 2.965235173824131,
-      "grad_norm": 0.6097820401191711,
-      "learning_rate": 0.00042260532270229495,
-      "loss": 3.5184,
+      "epoch": 2.9703504043126685,
+      "grad_norm": 0.5586152076721191,
+      "learning_rate": 0.0004222989746357258,
+      "loss": 3.5244,
       "step": 27550
     },
     {
-      "epoch": 2.970616725863739,
-      "grad_norm": 0.6079690456390381,
-      "learning_rate": 0.00042228208167223354,
-      "loss": 3.5342,
+      "epoch": 2.975741239892183,
+      "grad_norm": 0.5793421268463135,
+      "learning_rate": 0.0004219751753912574,
+      "loss": 3.5331,
       "step": 27600
     },
     {
-      "epoch": 2.975998277903347,
-      "grad_norm": 0.6019097566604614,
-      "learning_rate": 0.00042195884064217214,
-      "loss": 3.5326,
+      "epoch": 2.981132075471698,
+      "grad_norm": 0.6181196570396423,
+      "learning_rate": 0.00042165137614678895,
+      "loss": 3.5123,
       "step": 27650
     },
     {
-      "epoch": 2.9813798299429557,
-      "grad_norm": 0.568587601184845,
-      "learning_rate": 0.00042163559961211073,
-      "loss": 3.5355,
+      "epoch": 2.986522911051213,
+      "grad_norm": 0.6094911098480225,
+      "learning_rate": 0.00042132757690232055,
+      "loss": 3.5238,
       "step": 27700
     },
     {
-      "epoch": 2.986761381982564,
-      "grad_norm": 0.5844939947128296,
-      "learning_rate": 0.00042131235858204933,
-      "loss": 3.5106,
+      "epoch": 2.9919137466307277,
+      "grad_norm": 0.6401978731155396,
+      "learning_rate": 0.0004210037776578521,
+      "loss": 3.5285,
       "step": 27750
     },
     {
-      "epoch": 2.992142934022172,
-      "grad_norm": 0.5808882713317871,
-      "learning_rate": 0.0004209891175519879,
-      "loss": 3.4946,
+      "epoch": 2.9973045822102424,
+      "grad_norm": 0.6615415811538696,
+      "learning_rate": 0.00042067997841338365,
+      "loss": 3.5306,
       "step": 27800
     },
     {
-      "epoch": 2.9975244860617805,
-      "grad_norm": 0.6142293810844421,
-      "learning_rate": 0.00042066587652192646,
-      "loss": 3.5295,
+      "epoch": 3.0026954177897576,
+      "grad_norm": 0.6077088713645935,
+      "learning_rate": 0.00042035617916891526,
+      "loss": 3.4936,
       "step": 27850
     },
     {
-      "epoch": 3.0029060381013886,
-      "grad_norm": 0.5733931064605713,
-      "learning_rate": 0.00042034263549186506,
-      "loss": 3.4664,
+      "epoch": 3.0080862533692723,
+      "grad_norm": 0.6928874850273132,
+      "learning_rate": 0.0004200323799244468,
+      "loss": 3.4344,
       "step": 27900
     },
     {
-      "epoch": 3.0082875901409967,
-      "grad_norm": 0.6103217005729675,
-      "learning_rate": 0.0004200193944618036,
-      "loss": 3.4296,
+      "epoch": 3.013477088948787,
+      "grad_norm": 0.6022824048995972,
+      "learning_rate": 0.0004197085806799784,
+      "loss": 3.4388,
       "step": 27950
     },
     {
-      "epoch": 3.0136691421806048,
-      "grad_norm": 0.5486161112785339,
-      "learning_rate": 0.00041969615343174225,
-      "loss": 3.4241,
+      "epoch": 3.018867924528302,
+      "grad_norm": 0.6410669684410095,
+      "learning_rate": 0.0004193847814355099,
+      "loss": 3.4514,
       "step": 28000
     },
     {
-      "epoch": 3.0136691421806048,
-      "eval_accuracy": 0.3709459570314771,
-      "eval_loss": 3.4968576431274414,
-      "eval_runtime": 186.5664,
-      "eval_samples_per_second": 96.539,
-      "eval_steps_per_second": 6.035,
+      "epoch": 3.018867924528302,
+      "eval_accuracy": 0.3695786690912196,
+      "eval_loss": 3.5046586990356445,
+      "eval_runtime": 184.0963,
+      "eval_samples_per_second": 97.835,
+      "eval_steps_per_second": 6.116,
       "step": 28000
     },
     {
-      "epoch": 3.0190506942202133,
-      "grad_norm": 0.552299439907074,
-      "learning_rate": 0.00041937291240168084,
-      "loss": 3.4282,
+      "epoch": 3.024258760107817,
+      "grad_norm": 0.6208857893943787,
+      "learning_rate": 0.0004190609821910415,
+      "loss": 3.438,
       "step": 28050
     },
     {
-      "epoch": 3.0244322462598214,
-      "grad_norm": 0.6174763441085815,
-      "learning_rate": 0.0004190496713716194,
-      "loss": 3.4234,
+      "epoch": 3.0296495956873315,
+      "grad_norm": 0.6808158159255981,
+      "learning_rate": 0.00041873718294657306,
+      "loss": 3.4323,
       "step": 28100
     },
     {
-      "epoch": 3.0298137982994295,
-      "grad_norm": 0.6053746938705444,
-      "learning_rate": 0.000418726430341558,
-      "loss": 3.4377,
+      "epoch": 3.035040431266846,
+      "grad_norm": 0.6023147702217102,
+      "learning_rate": 0.00041841338370210467,
+      "loss": 3.4713,
       "step": 28150
     },
     {
-      "epoch": 3.0351953503390376,
-      "grad_norm": 0.5536657571792603,
-      "learning_rate": 0.00041840318931149657,
-      "loss": 3.4153,
+      "epoch": 3.0404312668463613,
+      "grad_norm": 0.6498099565505981,
+      "learning_rate": 0.0004180895844576362,
+      "loss": 3.4522,
       "step": 28200
     },
     {
-      "epoch": 3.040576902378646,
-      "grad_norm": 0.5788959860801697,
-      "learning_rate": 0.00041807994828143517,
-      "loss": 3.4299,
+      "epoch": 3.045822102425876,
+      "grad_norm": 0.6043527722358704,
+      "learning_rate": 0.00041776578521316777,
+      "loss": 3.4625,
       "step": 28250
     },
     {
-      "epoch": 3.0459584544182543,
-      "grad_norm": 0.5770339965820312,
-      "learning_rate": 0.00041775670725137376,
-      "loss": 3.4354,
+      "epoch": 3.0512129380053907,
+      "grad_norm": 0.5966095924377441,
+      "learning_rate": 0.0004174419859686994,
+      "loss": 3.4584,
       "step": 28300
     },
     {
-      "epoch": 3.0513400064578624,
-      "grad_norm": 0.5935100317001343,
-      "learning_rate": 0.00041743346622131236,
-      "loss": 3.4253,
+      "epoch": 3.056603773584906,
+      "grad_norm": 0.6195217370986938,
+      "learning_rate": 0.0004171181867242309,
+      "loss": 3.4512,
       "step": 28350
     },
     {
-      "epoch": 3.0567215584974705,
-      "grad_norm": 0.5614299774169922,
-      "learning_rate": 0.0004171102251912509,
-      "loss": 3.431,
+      "epoch": 3.0619946091644206,
+      "grad_norm": 0.587759792804718,
+      "learning_rate": 0.00041679438747976253,
+      "loss": 3.4296,
       "step": 28400
     },
     {
-      "epoch": 3.062103110537079,
-      "grad_norm": 0.5908504128456116,
-      "learning_rate": 0.0004167869841611895,
-      "loss": 3.4449,
+      "epoch": 3.0673854447439353,
+      "grad_norm": 0.6292561292648315,
+      "learning_rate": 0.0004164705882352941,
+      "loss": 3.471,
       "step": 28450
     },
     {
-      "epoch": 3.067484662576687,
-      "grad_norm": 0.5854935050010681,
-      "learning_rate": 0.00041646374313112803,
-      "loss": 3.4305,
+      "epoch": 3.07277628032345,
+      "grad_norm": 0.6752822399139404,
+      "learning_rate": 0.0004161467889908257,
+      "loss": 3.4651,
       "step": 28500
     },
     {
-      "epoch": 3.0728662146162953,
-      "grad_norm": 0.5865063667297363,
-      "learning_rate": 0.0004161405021010667,
-      "loss": 3.4402,
+      "epoch": 3.078167115902965,
+      "grad_norm": 0.6305340528488159,
+      "learning_rate": 0.00041582298974635724,
+      "loss": 3.4505,
       "step": 28550
     },
     {
-      "epoch": 3.0782477666559034,
-      "grad_norm": 0.5648619532585144,
-      "learning_rate": 0.0004158172610710053,
-      "loss": 3.4306,
+      "epoch": 3.08355795148248,
+      "grad_norm": 0.5910475850105286,
+      "learning_rate": 0.00041549919050188884,
+      "loss": 3.4477,
       "step": 28600
     },
     {
-      "epoch": 3.083629318695512,
-      "grad_norm": 0.5741128325462341,
-      "learning_rate": 0.0004154940200409438,
-      "loss": 3.4448,
+      "epoch": 3.0889487870619945,
+      "grad_norm": 0.6657103896141052,
+      "learning_rate": 0.00041517539125742034,
+      "loss": 3.4481,
       "step": 28650
     },
     {
-      "epoch": 3.08901087073512,
-      "grad_norm": 0.600200891494751,
-      "learning_rate": 0.0004151707790108824,
-      "loss": 3.4281,
+      "epoch": 3.0943396226415096,
+      "grad_norm": 0.5596718192100525,
+      "learning_rate": 0.0004148515920129519,
+      "loss": 3.4585,
       "step": 28700
     },
     {
-      "epoch": 3.094392422774728,
-      "grad_norm": 0.6200466752052307,
-      "learning_rate": 0.00041484753798082095,
-      "loss": 3.4415,
+      "epoch": 3.0997304582210243,
+      "grad_norm": 0.6167557239532471,
+      "learning_rate": 0.0004145277927684835,
+      "loss": 3.4469,
       "step": 28750
     },
     {
-      "epoch": 3.0997739748143363,
-      "grad_norm": 0.649884819984436,
-      "learning_rate": 0.00041452429695075955,
-      "loss": 3.4648,
+      "epoch": 3.105121293800539,
+      "grad_norm": 0.6317213773727417,
+      "learning_rate": 0.00041420399352401504,
+      "loss": 3.447,
       "step": 28800
     },
     {
-      "epoch": 3.105155526853945,
-      "grad_norm": 0.5909411311149597,
-      "learning_rate": 0.0004142010559206982,
-      "loss": 3.4405,
+      "epoch": 3.1105121293800537,
+      "grad_norm": 0.6201804280281067,
+      "learning_rate": 0.00041388019427954665,
+      "loss": 3.4318,
       "step": 28850
     },
     {
-      "epoch": 3.110537078893553,
-      "grad_norm": 0.6185323596000671,
-      "learning_rate": 0.00041387781489063673,
-      "loss": 3.4551,
+      "epoch": 3.115902964959569,
+      "grad_norm": 0.6122051477432251,
+      "learning_rate": 0.0004135563950350782,
+      "loss": 3.4495,
       "step": 28900
     },
     {
-      "epoch": 3.115918630933161,
-      "grad_norm": 0.6023309826850891,
-      "learning_rate": 0.00041355457386057533,
-      "loss": 3.4361,
+      "epoch": 3.1212938005390836,
+      "grad_norm": 0.6710070371627808,
+      "learning_rate": 0.0004132325957906098,
+      "loss": 3.4673,
       "step": 28950
     },
     {
-      "epoch": 3.121300182972769,
-      "grad_norm": 0.5937264561653137,
-      "learning_rate": 0.0004132313328305139,
-      "loss": 3.4403,
+      "epoch": 3.1266846361185983,
+      "grad_norm": 0.6424208283424377,
+      "learning_rate": 0.00041290879654614135,
+      "loss": 3.4374,
       "step": 29000
     },
     {
-      "epoch": 3.121300182972769,
-      "eval_accuracy": 0.37116684835240393,
-      "eval_loss": 3.494439125061035,
-      "eval_runtime": 186.2689,
-      "eval_samples_per_second": 96.694,
-      "eval_steps_per_second": 6.045,
+      "epoch": 3.1266846361185983,
+      "eval_accuracy": 0.37038965424586695,
+      "eval_loss": 3.501832962036133,
+      "eval_runtime": 184.0616,
+      "eval_samples_per_second": 97.853,
+      "eval_steps_per_second": 6.118,
       "step": 29000
     },
     {
-      "epoch": 3.1266817350123777,
-      "grad_norm": 0.5779401063919067,
-      "learning_rate": 0.00041290809180045246,
-      "loss": 3.4429,
+      "epoch": 3.1320754716981134,
+      "grad_norm": 0.6676577925682068,
+      "learning_rate": 0.00041258499730167296,
+      "loss": 3.4583,
       "step": 29050
     },
     {
-      "epoch": 3.132063287051986,
-      "grad_norm": 0.6609095931053162,
-      "learning_rate": 0.0004125848507703911,
-      "loss": 3.4252,
+      "epoch": 3.137466307277628,
+      "grad_norm": 0.6498281359672546,
+      "learning_rate": 0.0004122611980572045,
+      "loss": 3.448,
       "step": 29100
     },
     {
-      "epoch": 3.137444839091594,
-      "grad_norm": 0.6140174865722656,
-      "learning_rate": 0.0004122616097403297,
-      "loss": 3.4487,
+      "epoch": 3.142857142857143,
+      "grad_norm": 0.6448217630386353,
+      "learning_rate": 0.00041193739881273606,
+      "loss": 3.4656,
       "step": 29150
     },
     {
-      "epoch": 3.1428263911312024,
-      "grad_norm": 0.5952047109603882,
-      "learning_rate": 0.00041193836871026825,
-      "loss": 3.4334,
+      "epoch": 3.1482479784366575,
+      "grad_norm": 0.5762954950332642,
+      "learning_rate": 0.00041161359956826766,
+      "loss": 3.4665,
       "step": 29200
     },
     {
-      "epoch": 3.1482079431708105,
-      "grad_norm": 0.6305395364761353,
-      "learning_rate": 0.00041161512768020684,
-      "loss": 3.4466,
+      "epoch": 3.1536388140161726,
+      "grad_norm": 0.6261320114135742,
+      "learning_rate": 0.0004112898003237992,
+      "loss": 3.4521,
       "step": 29250
     },
     {
-      "epoch": 3.1535894952104186,
-      "grad_norm": 0.6519289016723633,
-      "learning_rate": 0.0004112918866501454,
-      "loss": 3.4466,
+      "epoch": 3.1590296495956873,
+      "grad_norm": 0.6095337867736816,
+      "learning_rate": 0.0004109660010793308,
+      "loss": 3.4603,
       "step": 29300
     },
     {
-      "epoch": 3.1589710472500268,
-      "grad_norm": 0.6156654953956604,
-      "learning_rate": 0.00041097511044068524,
-      "loss": 3.4478,
+      "epoch": 3.164420485175202,
+      "grad_norm": 0.6465772986412048,
+      "learning_rate": 0.0004106422018348623,
+      "loss": 3.4718,
       "step": 29350
     },
     {
-      "epoch": 3.1643525992896353,
-      "grad_norm": 0.5677369832992554,
-      "learning_rate": 0.0004106518694106238,
-      "loss": 3.4467,
+      "epoch": 3.169811320754717,
+      "grad_norm": 0.5672667622566223,
+      "learning_rate": 0.0004103184025903939,
+      "loss": 3.4662,
       "step": 29400
     },
     {
-      "epoch": 3.1697341513292434,
-      "grad_norm": 0.5640089511871338,
-      "learning_rate": 0.0004103286283805624,
-      "loss": 3.4452,
+      "epoch": 3.175202156334232,
+      "grad_norm": 0.632469892501831,
+      "learning_rate": 0.00040999460334592547,
+      "loss": 3.4763,
       "step": 29450
     },
     {
-      "epoch": 3.1751157033688515,
-      "grad_norm": 0.5955638885498047,
-      "learning_rate": 0.00041000538735050103,
-      "loss": 3.4637,
+      "epoch": 3.1805929919137466,
+      "grad_norm": 0.6090266108512878,
+      "learning_rate": 0.000409670804101457,
+      "loss": 3.4732,
       "step": 29500
     },
     {
-      "epoch": 3.1804972554084596,
-      "grad_norm": 0.5567012429237366,
-      "learning_rate": 0.00040968214632043957,
-      "loss": 3.4558,
+      "epoch": 3.1859838274932613,
+      "grad_norm": 0.6065264940261841,
+      "learning_rate": 0.000409353480841878,
+      "loss": 3.4458,
       "step": 29550
     },
     {
-      "epoch": 3.185878807448068,
-      "grad_norm": 0.5742025375366211,
-      "learning_rate": 0.00040935890529037816,
-      "loss": 3.4426,
+      "epoch": 3.1913746630727764,
+      "grad_norm": 0.5902264714241028,
+      "learning_rate": 0.0004090296815974096,
+      "loss": 3.468,
       "step": 29600
     },
     {
-      "epoch": 3.1912603594876763,
-      "grad_norm": 0.5888547897338867,
-      "learning_rate": 0.00040903566426031676,
-      "loss": 3.4463,
+      "epoch": 3.196765498652291,
+      "grad_norm": 0.5910829305648804,
+      "learning_rate": 0.00040870588235294113,
+      "loss": 3.4555,
       "step": 29650
     },
     {
-      "epoch": 3.1966419115272844,
-      "grad_norm": 0.650274395942688,
-      "learning_rate": 0.0004087124232302553,
-      "loss": 3.4362,
+      "epoch": 3.202156334231806,
+      "grad_norm": 0.5923267006874084,
+      "learning_rate": 0.00040838208310847273,
+      "loss": 3.4618,
       "step": 29700
     },
     {
-      "epoch": 3.2020234635668925,
-      "grad_norm": 0.6139590740203857,
-      "learning_rate": 0.0004083891822001939,
-      "loss": 3.4598,
+      "epoch": 3.207547169811321,
+      "grad_norm": 0.6040782332420349,
+      "learning_rate": 0.0004080582838640043,
+      "loss": 3.4696,
       "step": 29750
     },
     {
-      "epoch": 3.207405015606501,
-      "grad_norm": 0.5943894386291504,
-      "learning_rate": 0.00040806594117013254,
-      "loss": 3.4361,
+      "epoch": 3.2129380053908356,
+      "grad_norm": 0.6052514910697937,
+      "learning_rate": 0.0004077344846195359,
+      "loss": 3.4507,
       "step": 29800
     },
     {
-      "epoch": 3.212786567646109,
-      "grad_norm": 0.5968701839447021,
-      "learning_rate": 0.0004077427001400711,
-      "loss": 3.4541,
+      "epoch": 3.2183288409703503,
+      "grad_norm": 0.6231287717819214,
+      "learning_rate": 0.00040741068537506744,
+      "loss": 3.4733,
       "step": 29850
     },
     {
-      "epoch": 3.2181681196857173,
-      "grad_norm": 0.6620115637779236,
-      "learning_rate": 0.0004074194591100097,
-      "loss": 3.4598,
+      "epoch": 3.223719676549865,
+      "grad_norm": 0.621620774269104,
+      "learning_rate": 0.000407086886130599,
+      "loss": 3.4804,
       "step": 29900
     },
     {
-      "epoch": 3.2235496717253254,
-      "grad_norm": 0.5594594478607178,
-      "learning_rate": 0.0004070962180799482,
-      "loss": 3.4586,
+      "epoch": 3.22911051212938,
+      "grad_norm": 0.5981627106666565,
+      "learning_rate": 0.0004067630868861306,
+      "loss": 3.471,
       "step": 29950
     },
     {
-      "epoch": 3.228931223764934,
-      "grad_norm": 0.6292806267738342,
-      "learning_rate": 0.0004067729770498868,
-      "loss": 3.4606,
+      "epoch": 3.234501347708895,
+      "grad_norm": 0.6375696659088135,
+      "learning_rate": 0.0004064392876416621,
+      "loss": 3.4584,
       "step": 30000
     },
     {
-      "epoch": 3.228931223764934,
-      "eval_accuracy": 0.37190231975002014,
-      "eval_loss": 3.48602294921875,
-      "eval_runtime": 186.7407,
-      "eval_samples_per_second": 96.449,
-      "eval_steps_per_second": 6.03,
+      "epoch": 3.234501347708895,
+      "eval_accuracy": 0.37104635229981764,
+      "eval_loss": 3.494283437728882,
+      "eval_runtime": 184.6457,
+      "eval_samples_per_second": 97.544,
+      "eval_steps_per_second": 6.098,
       "step": 30000
     },
     {
-      "epoch": 3.234312775804542,
-      "grad_norm": 0.6741182804107666,
-      "learning_rate": 0.00040644973601982546,
-      "loss": 3.4655,
+      "epoch": 3.2398921832884096,
+      "grad_norm": 0.5990398526191711,
+      "learning_rate": 0.0004061154883971937,
+      "loss": 3.4554,
       "step": 30050
     },
     {
-      "epoch": 3.23969432784415,
-      "grad_norm": 0.6115264296531677,
-      "learning_rate": 0.000406126494989764,
-      "loss": 3.4511,
+      "epoch": 3.2452830188679247,
+      "grad_norm": 0.6470616459846497,
+      "learning_rate": 0.00040579816513761465,
+      "loss": 3.4561,
       "step": 30100
     },
     {
-      "epoch": 3.2450758798837587,
-      "grad_norm": 0.6215344667434692,
-      "learning_rate": 0.0004058032539597026,
-      "loss": 3.4334,
+      "epoch": 3.2506738544474394,
+      "grad_norm": 0.567266583442688,
+      "learning_rate": 0.0004054743658931462,
+      "loss": 3.4571,
       "step": 30150
     },
     {
-      "epoch": 3.250457431923367,
-      "grad_norm": 0.6353278160095215,
-      "learning_rate": 0.00040548001292964114,
-      "loss": 3.4418,
+      "epoch": 3.256064690026954,
+      "grad_norm": 0.6559926867485046,
+      "learning_rate": 0.00040515056664867775,
+      "loss": 3.4756,
       "step": 30200
     },
     {
-      "epoch": 3.255838983962975,
-      "grad_norm": 0.5738556981086731,
-      "learning_rate": 0.00040515677189957973,
-      "loss": 3.4462,
+      "epoch": 3.2614555256064692,
+      "grad_norm": 0.6251115202903748,
+      "learning_rate": 0.00040482676740420935,
+      "loss": 3.4715,
       "step": 30250
     },
     {
-      "epoch": 3.261220536002583,
-      "grad_norm": 0.6045312285423279,
-      "learning_rate": 0.0004048335308695183,
-      "loss": 3.4399,
+      "epoch": 3.266846361185984,
+      "grad_norm": 0.7224171161651611,
+      "learning_rate": 0.0004045029681597409,
+      "loss": 3.4722,
       "step": 30300
     },
     {
-      "epoch": 3.2666020880421915,
-      "grad_norm": 0.5951460003852844,
-      "learning_rate": 0.0004045102898394569,
-      "loss": 3.4543,
+      "epoch": 3.2722371967654986,
+      "grad_norm": 0.6535780429840088,
+      "learning_rate": 0.0004041791689152725,
+      "loss": 3.4469,
       "step": 30350
     },
     {
-      "epoch": 3.2719836400817996,
-      "grad_norm": 0.7059372663497925,
-      "learning_rate": 0.0004041870488093955,
-      "loss": 3.4742,
+      "epoch": 3.2776280323450133,
+      "grad_norm": 0.7087780237197876,
+      "learning_rate": 0.00040385536967080406,
+      "loss": 3.4786,
       "step": 30400
     },
     {
-      "epoch": 3.2773651921214078,
-      "grad_norm": 0.6901097893714905,
-      "learning_rate": 0.0004038638077793341,
-      "loss": 3.4521,
+      "epoch": 3.2830188679245285,
+      "grad_norm": 0.6123559474945068,
+      "learning_rate": 0.00040353157042633567,
+      "loss": 3.4641,
       "step": 30450
     },
     {
-      "epoch": 3.282746744161016,
-      "grad_norm": 0.5631442666053772,
-      "learning_rate": 0.00040354056674927265,
-      "loss": 3.4702,
+      "epoch": 3.288409703504043,
+      "grad_norm": 0.6059187650680542,
+      "learning_rate": 0.0004032077711818672,
+      "loss": 3.4498,
       "step": 30500
     },
     {
-      "epoch": 3.2881282962006244,
-      "grad_norm": 0.643974781036377,
-      "learning_rate": 0.00040321732571921124,
-      "loss": 3.4405,
+      "epoch": 3.293800539083558,
+      "grad_norm": 0.6493521928787231,
+      "learning_rate": 0.0004028839719373988,
+      "loss": 3.4779,
       "step": 30550
     },
     {
-      "epoch": 3.2935098482402325,
-      "grad_norm": 0.6235001683235168,
-      "learning_rate": 0.0004028940846891498,
-      "loss": 3.4392,
+      "epoch": 3.2991913746630726,
+      "grad_norm": 0.7040890455245972,
+      "learning_rate": 0.00040256017269293037,
+      "loss": 3.4684,
       "step": 30600
     },
     {
-      "epoch": 3.2988914002798406,
-      "grad_norm": 0.6189355850219727,
-      "learning_rate": 0.00040257084365908843,
-      "loss": 3.4457,
+      "epoch": 3.3045822102425877,
+      "grad_norm": 0.6316874623298645,
+      "learning_rate": 0.00040223637344846187,
+      "loss": 3.4486,
       "step": 30650
     },
     {
-      "epoch": 3.304272952319449,
-      "grad_norm": 0.5895229578018188,
-      "learning_rate": 0.00040224760262902703,
-      "loss": 3.4525,
+      "epoch": 3.3099730458221024,
+      "grad_norm": 0.6549959778785706,
+      "learning_rate": 0.00040191257420399347,
+      "loss": 3.466,
       "step": 30700
     },
     {
-      "epoch": 3.3096545043590573,
-      "grad_norm": 0.6058889031410217,
-      "learning_rate": 0.00040192436159896557,
-      "loss": 3.4741,
+      "epoch": 3.315363881401617,
+      "grad_norm": 0.602091372013092,
+      "learning_rate": 0.000401588774959525,
+      "loss": 3.4356,
       "step": 30750
     },
     {
-      "epoch": 3.3150360563986654,
-      "grad_norm": 0.596379280090332,
-      "learning_rate": 0.00040160112056890416,
-      "loss": 3.4471,
+      "epoch": 3.3207547169811322,
+      "grad_norm": 0.5951991677284241,
+      "learning_rate": 0.00040126497571505663,
+      "loss": 3.4716,
       "step": 30800
     },
     {
-      "epoch": 3.3204176084382735,
-      "grad_norm": 0.6520100235939026,
-      "learning_rate": 0.00040127787953884276,
-      "loss": 3.4563,
+      "epoch": 3.326145552560647,
+      "grad_norm": 0.6242449879646301,
+      "learning_rate": 0.0004009411764705882,
+      "loss": 3.4514,
       "step": 30850
     },
     {
-      "epoch": 3.3257991604778816,
-      "grad_norm": 0.6291080117225647,
-      "learning_rate": 0.00040095463850878135,
-      "loss": 3.4568,
+      "epoch": 3.3315363881401616,
+      "grad_norm": 0.6377851963043213,
+      "learning_rate": 0.0004006173772261198,
+      "loss": 3.4734,
       "step": 30900
     },
     {
-      "epoch": 3.33118071251749,
-      "grad_norm": 0.5766595602035522,
-      "learning_rate": 0.00040063139747871995,
-      "loss": 3.4679,
+      "epoch": 3.3369272237196768,
+      "grad_norm": 0.6302831768989563,
+      "learning_rate": 0.00040029357798165133,
+      "loss": 3.4509,
       "step": 30950
     },
     {
-      "epoch": 3.3365622645570983,
-      "grad_norm": 0.5974405407905579,
-      "learning_rate": 0.00040030815644865854,
-      "loss": 3.4446,
+      "epoch": 3.3423180592991915,
+      "grad_norm": 0.6283584833145142,
+      "learning_rate": 0.00039996977873718294,
+      "loss": 3.464,
       "step": 31000
     },
     {
-      "epoch": 3.3365622645570983,
-      "eval_accuracy": 0.372818372247184,
-      "eval_loss": 3.4797027111053467,
-      "eval_runtime": 186.1465,
-      "eval_samples_per_second": 96.757,
-      "eval_steps_per_second": 6.049,
+      "epoch": 3.3423180592991915,
+      "eval_accuracy": 0.37161069539912606,
+      "eval_loss": 3.487372636795044,
+      "eval_runtime": 183.981,
+      "eval_samples_per_second": 97.896,
+      "eval_steps_per_second": 6.12,
       "step": 31000
     },
     {
-      "epoch": 3.3419438165967064,
-      "grad_norm": 0.5883570313453674,
-      "learning_rate": 0.0003999849154185971,
-      "loss": 3.4637,
+      "epoch": 3.347708894878706,
+      "grad_norm": 0.7022663950920105,
+      "learning_rate": 0.0003996459794927145,
+      "loss": 3.4656,
       "step": 31050
     },
     {
-      "epoch": 3.347325368636315,
-      "grad_norm": 0.6028373837471008,
-      "learning_rate": 0.0003996616743885357,
-      "loss": 3.4393,
+      "epoch": 3.353099730458221,
+      "grad_norm": 0.5983381271362305,
+      "learning_rate": 0.00039932218024824604,
+      "loss": 3.4608,
       "step": 31100
     },
     {
-      "epoch": 3.352706920675923,
-      "grad_norm": 0.6262359619140625,
-      "learning_rate": 0.0003993384333584742,
-      "loss": 3.4486,
+      "epoch": 3.358490566037736,
+      "grad_norm": 0.6170933246612549,
+      "learning_rate": 0.00039899838100377764,
+      "loss": 3.4714,
       "step": 31150
     },
     {
-      "epoch": 3.358088472715531,
-      "grad_norm": 0.6157032251358032,
-      "learning_rate": 0.00039901519232841287,
-      "loss": 3.4708,
+      "epoch": 3.3638814016172507,
+      "grad_norm": 0.6005626320838928,
+      "learning_rate": 0.0003986745817593092,
+      "loss": 3.4556,
       "step": 31200
     },
     {
-      "epoch": 3.3634700247551392,
-      "grad_norm": 0.6458026170730591,
-      "learning_rate": 0.00039869195129835146,
-      "loss": 3.462,
+      "epoch": 3.3692722371967654,
+      "grad_norm": 0.5886436700820923,
+      "learning_rate": 0.0003983507825148408,
+      "loss": 3.4602,
       "step": 31250
     },
     {
-      "epoch": 3.368851576794748,
-      "grad_norm": 0.6021883487701416,
-      "learning_rate": 0.00039836871026829,
-      "loss": 3.4438,
+      "epoch": 3.37466307277628,
+      "grad_norm": 0.5955742597579956,
+      "learning_rate": 0.00039802698327037235,
+      "loss": 3.4863,
       "step": 31300
     },
     {
-      "epoch": 3.374233128834356,
-      "grad_norm": 0.5775461196899414,
-      "learning_rate": 0.0003980454692382286,
-      "loss": 3.459,
+      "epoch": 3.3800539083557952,
+      "grad_norm": 0.6336953043937683,
+      "learning_rate": 0.00039770318402590396,
+      "loss": 3.456,
       "step": 31350
     },
     {
-      "epoch": 3.379614680873964,
-      "grad_norm": 0.6387748718261719,
-      "learning_rate": 0.0003977286930287684,
-      "loss": 3.4451,
+      "epoch": 3.38544474393531,
+      "grad_norm": 0.629045307636261,
+      "learning_rate": 0.00039737938478143545,
+      "loss": 3.4559,
       "step": 31400
     },
     {
-      "epoch": 3.384996232913572,
-      "grad_norm": 0.5816205143928528,
-      "learning_rate": 0.000397405451998707,
-      "loss": 3.4378,
+      "epoch": 3.3908355795148246,
+      "grad_norm": 0.6489145755767822,
+      "learning_rate": 0.000397055585536967,
+      "loss": 3.4631,
       "step": 31450
     },
     {
-      "epoch": 3.3903777849531807,
-      "grad_norm": 0.6428983211517334,
-      "learning_rate": 0.00039708221096864554,
-      "loss": 3.4762,
+      "epoch": 3.3962264150943398,
+      "grad_norm": 0.6067516803741455,
+      "learning_rate": 0.0003967317862924986,
+      "loss": 3.4549,
       "step": 31500
     },
     {
-      "epoch": 3.3957593369927888,
-      "grad_norm": 0.576107919216156,
-      "learning_rate": 0.00039675896993858413,
-      "loss": 3.4581,
+      "epoch": 3.4016172506738545,
+      "grad_norm": 0.5860952138900757,
+      "learning_rate": 0.00039640798704803016,
+      "loss": 3.4732,
       "step": 31550
     },
     {
-      "epoch": 3.401140889032397,
-      "grad_norm": 0.6172900199890137,
-      "learning_rate": 0.0003964357289085228,
-      "loss": 3.4481,
+      "epoch": 3.407008086253369,
+      "grad_norm": 0.6216514110565186,
+      "learning_rate": 0.00039608418780356176,
+      "loss": 3.4526,
       "step": 31600
     },
     {
-      "epoch": 3.4065224410720054,
-      "grad_norm": 0.6148959994316101,
-      "learning_rate": 0.0003961124878784613,
-      "loss": 3.4502,
+      "epoch": 3.4123989218328843,
+      "grad_norm": 0.6272159814834595,
+      "learning_rate": 0.0003957603885590933,
+      "loss": 3.4537,
       "step": 31650
     },
     {
-      "epoch": 3.4119039931116135,
-      "grad_norm": 0.6322537660598755,
-      "learning_rate": 0.0003957892468483999,
-      "loss": 3.4383,
+      "epoch": 3.417789757412399,
+      "grad_norm": 0.6162016987800598,
+      "learning_rate": 0.0003954365893146249,
+      "loss": 3.48,
       "step": 31700
     },
     {
-      "epoch": 3.4172855451512216,
-      "grad_norm": 0.6025902032852173,
-      "learning_rate": 0.0003954660058183385,
-      "loss": 3.4487,
+      "epoch": 3.4231805929919137,
+      "grad_norm": 0.6524364352226257,
+      "learning_rate": 0.00039511279007015647,
+      "loss": 3.452,
       "step": 31750
     },
     {
-      "epoch": 3.4226670971908297,
-      "grad_norm": 0.6599864959716797,
-      "learning_rate": 0.00039514276478827705,
-      "loss": 3.4533,
+      "epoch": 3.4285714285714284,
+      "grad_norm": 0.6311447620391846,
+      "learning_rate": 0.00039478899082568807,
+      "loss": 3.4581,
       "step": 31800
     },
     {
-      "epoch": 3.428048649230438,
-      "grad_norm": 0.6351486444473267,
-      "learning_rate": 0.0003948195237582157,
-      "loss": 3.4482,
+      "epoch": 3.4339622641509435,
+      "grad_norm": 0.6871687769889832,
+      "learning_rate": 0.0003944651915812196,
+      "loss": 3.488,
       "step": 31850
     },
     {
-      "epoch": 3.4334302012700464,
-      "grad_norm": 0.574647068977356,
-      "learning_rate": 0.0003944962827281543,
-      "loss": 3.4594,
+      "epoch": 3.439353099730458,
+      "grad_norm": 0.6635282039642334,
+      "learning_rate": 0.0003941413923367512,
+      "loss": 3.4677,
       "step": 31900
     },
     {
-      "epoch": 3.4388117533096545,
-      "grad_norm": 0.6054155230522156,
-      "learning_rate": 0.00039417304169809284,
-      "loss": 3.4415,
+      "epoch": 3.444743935309973,
+      "grad_norm": 0.6801772713661194,
+      "learning_rate": 0.0003938175930922828,
+      "loss": 3.449,
       "step": 31950
     },
     {
-      "epoch": 3.4441933053492626,
-      "grad_norm": 0.5832567811012268,
-      "learning_rate": 0.00039384980066803143,
-      "loss": 3.4477,
+      "epoch": 3.450134770889488,
+      "grad_norm": 0.6524240374565125,
+      "learning_rate": 0.0003934937938478143,
+      "loss": 3.464,
       "step": 32000
     },
     {
-      "epoch": 3.4441933053492626,
-      "eval_accuracy": 0.37344855899650803,
-      "eval_loss": 3.470879316329956,
-      "eval_runtime": 186.8889,
-      "eval_samples_per_second": 96.373,
-      "eval_steps_per_second": 6.025,
+      "epoch": 3.450134770889488,
+      "eval_accuracy": 0.3724226584297637,
+      "eval_loss": 3.482102870941162,
+      "eval_runtime": 184.4443,
+      "eval_samples_per_second": 97.65,
+      "eval_steps_per_second": 6.105,
       "step": 32000
     },
     {
-      "epoch": 3.449574857388871,
-      "grad_norm": 0.6063782572746277,
-      "learning_rate": 0.00039352655963796997,
-      "loss": 3.441,
+      "epoch": 3.4555256064690028,
+      "grad_norm": 0.6412952542304993,
+      "learning_rate": 0.0003931699946033459,
+      "loss": 3.4665,
       "step": 32050
     },
     {
-      "epoch": 3.4549564094284793,
-      "grad_norm": 0.5823463797569275,
-      "learning_rate": 0.00039320331860790857,
-      "loss": 3.4633,
+      "epoch": 3.4609164420485174,
+      "grad_norm": 0.6748786568641663,
+      "learning_rate": 0.00039284619535887743,
+      "loss": 3.461,
       "step": 32100
     },
     {
-      "epoch": 3.4603379614680874,
-      "grad_norm": 0.6007035970687866,
-      "learning_rate": 0.0003928800775778472,
-      "loss": 3.4431,
+      "epoch": 3.466307277628032,
+      "grad_norm": 0.6340701580047607,
+      "learning_rate": 0.00039252239611440904,
+      "loss": 3.4737,
       "step": 32150
     },
     {
-      "epoch": 3.4657195135076955,
-      "grad_norm": 0.6317636966705322,
-      "learning_rate": 0.00039255683654778576,
-      "loss": 3.4525,
+      "epoch": 3.4716981132075473,
+      "grad_norm": 0.610893189907074,
+      "learning_rate": 0.0003921985968699406,
+      "loss": 3.4549,
       "step": 32200
     },
     {
-      "epoch": 3.471101065547304,
-      "grad_norm": 0.6009837985038757,
-      "learning_rate": 0.00039223359551772435,
-      "loss": 3.4623,
+      "epoch": 3.477088948787062,
+      "grad_norm": 0.665233314037323,
+      "learning_rate": 0.0003918747976254722,
+      "loss": 3.4716,
       "step": 32250
     },
     {
-      "epoch": 3.476482617586912,
-      "grad_norm": 0.619046151638031,
-      "learning_rate": 0.00039191035448766294,
-      "loss": 3.4446,
+      "epoch": 3.4824797843665767,
+      "grad_norm": 0.5984647870063782,
+      "learning_rate": 0.00039155099838100374,
+      "loss": 3.4822,
       "step": 32300
     },
     {
-      "epoch": 3.4818641696265202,
-      "grad_norm": 0.6180459856987,
-      "learning_rate": 0.0003915871134576015,
-      "loss": 3.468,
+      "epoch": 3.487870619946092,
+      "grad_norm": 0.6423894166946411,
+      "learning_rate": 0.0003912271991365353,
+      "loss": 3.446,
       "step": 32350
     },
     {
-      "epoch": 3.4872457216661283,
-      "grad_norm": 0.5934559106826782,
-      "learning_rate": 0.0003912638724275401,
-      "loss": 3.4538,
+      "epoch": 3.4932614555256065,
+      "grad_norm": 0.6157969832420349,
+      "learning_rate": 0.0003909033998920669,
+      "loss": 3.477,
       "step": 32400
     },
     {
-      "epoch": 3.492627273705737,
-      "grad_norm": 0.6671445965766907,
-      "learning_rate": 0.00039094063139747873,
-      "loss": 3.4423,
+      "epoch": 3.498652291105121,
+      "grad_norm": 0.6338315010070801,
+      "learning_rate": 0.00039057960064759845,
+      "loss": 3.451,
       "step": 32450
     },
     {
-      "epoch": 3.498008825745345,
-      "grad_norm": 0.6049616932868958,
-      "learning_rate": 0.00039061739036741727,
-      "loss": 3.4638,
+      "epoch": 3.5040431266846364,
+      "grad_norm": 0.5970466732978821,
+      "learning_rate": 0.00039025580140313005,
+      "loss": 3.476,
       "step": 32500
     },
     {
-      "epoch": 3.503390377784953,
-      "grad_norm": 0.5912634134292603,
-      "learning_rate": 0.00039029414933735586,
-      "loss": 3.4501,
+      "epoch": 3.509433962264151,
+      "grad_norm": 0.5832914710044861,
+      "learning_rate": 0.0003899320021586616,
+      "loss": 3.4679,
       "step": 32550
     },
     {
-      "epoch": 3.5087719298245617,
-      "grad_norm": 0.5970247983932495,
-      "learning_rate": 0.0003899709083072944,
-      "loss": 3.4447,
+      "epoch": 3.5148247978436657,
+      "grad_norm": 0.6068440675735474,
+      "learning_rate": 0.0003896082029141932,
+      "loss": 3.4633,
       "step": 32600
     },
     {
-      "epoch": 3.5141534818641698,
-      "grad_norm": 0.6333490610122681,
-      "learning_rate": 0.000389647667277233,
-      "loss": 3.461,
+      "epoch": 3.5202156334231804,
+      "grad_norm": 0.6159490942955017,
+      "learning_rate": 0.00038928440366972476,
+      "loss": 3.4723,
       "step": 32650
     },
     {
-      "epoch": 3.519535033903778,
-      "grad_norm": 0.6278321743011475,
-      "learning_rate": 0.00038932442624717165,
-      "loss": 3.4571,
+      "epoch": 3.525606469002695,
+      "grad_norm": 0.6146063804626465,
+      "learning_rate": 0.00038896060442525636,
+      "loss": 3.448,
       "step": 32700
     },
     {
-      "epoch": 3.524916585943386,
-      "grad_norm": 0.6167550683021545,
-      "learning_rate": 0.0003890011852171102,
-      "loss": 3.4511,
+      "epoch": 3.5309973045822103,
+      "grad_norm": 0.6683788895606995,
+      "learning_rate": 0.00038863680518078786,
+      "loss": 3.4579,
       "step": 32750
     },
     {
-      "epoch": 3.530298137982994,
-      "grad_norm": 0.5786373019218445,
-      "learning_rate": 0.0003886779441870488,
-      "loss": 3.4426,
+      "epoch": 3.536388140161725,
+      "grad_norm": 0.6073746681213379,
+      "learning_rate": 0.0003883130059363194,
+      "loss": 3.4702,
       "step": 32800
     },
     {
-      "epoch": 3.5356796900226026,
-      "grad_norm": 0.5911474227905273,
-      "learning_rate": 0.0003883547031569874,
-      "loss": 3.4472,
+      "epoch": 3.5417789757412397,
+      "grad_norm": 0.6339306235313416,
+      "learning_rate": 0.000387989206691851,
+      "loss": 3.4325,
       "step": 32850
     },
     {
-      "epoch": 3.5410612420622107,
-      "grad_norm": 0.5972420573234558,
-      "learning_rate": 0.0003880314621269259,
-      "loss": 3.4564,
+      "epoch": 3.547169811320755,
+      "grad_norm": 0.6130032539367676,
+      "learning_rate": 0.00038766540744738256,
+      "loss": 3.4551,
       "step": 32900
     },
     {
-      "epoch": 3.546442794101819,
-      "grad_norm": 0.6500043272972107,
-      "learning_rate": 0.0003877082210968645,
-      "loss": 3.4488,
+      "epoch": 3.5525606469002695,
+      "grad_norm": 0.6631814241409302,
+      "learning_rate": 0.00038734160820291417,
+      "loss": 3.469,
       "step": 32950
     },
     {
-      "epoch": 3.5518243461414274,
-      "grad_norm": 0.6295835971832275,
-      "learning_rate": 0.00038738498006680316,
-      "loss": 3.4627,
+      "epoch": 3.557951482479784,
+      "grad_norm": 0.6582044363021851,
+      "learning_rate": 0.0003870178089584457,
+      "loss": 3.4693,
       "step": 33000
     },
     {
-      "epoch": 3.5518243461414274,
-      "eval_accuracy": 0.3738635043750715,
-      "eval_loss": 3.468912363052368,
-      "eval_runtime": 186.1237,
-      "eval_samples_per_second": 96.769,
-      "eval_steps_per_second": 6.05,
+      "epoch": 3.557951482479784,
+      "eval_accuracy": 0.37294147596907784,
+      "eval_loss": 3.4770760536193848,
+      "eval_runtime": 183.9856,
+      "eval_samples_per_second": 97.894,
+      "eval_steps_per_second": 6.12,
       "step": 33000
     },
     {
-      "epoch": 3.5572058981810355,
-      "grad_norm": 0.6404278874397278,
-      "learning_rate": 0.0003870617390367417,
-      "loss": 3.4471,
+      "epoch": 3.5633423180592994,
+      "grad_norm": 0.6050445437431335,
+      "learning_rate": 0.0003866940097139773,
+      "loss": 3.4711,
       "step": 33050
     },
     {
-      "epoch": 3.5625874502206436,
-      "grad_norm": 0.5913998484611511,
-      "learning_rate": 0.0003867384980066803,
-      "loss": 3.4486,
+      "epoch": 3.568733153638814,
+      "grad_norm": 0.671772837638855,
+      "learning_rate": 0.0003863702104695089,
+      "loss": 3.4574,
       "step": 33100
     },
     {
-      "epoch": 3.5679690022602517,
-      "grad_norm": 0.5721989274024963,
-      "learning_rate": 0.00038641525697661884,
-      "loss": 3.4556,
+      "epoch": 3.5741239892183287,
+      "grad_norm": 0.6668798923492432,
+      "learning_rate": 0.0003860464112250404,
+      "loss": 3.4647,
       "step": 33150
     },
     {
-      "epoch": 3.57335055429986,
-      "grad_norm": 0.6180511713027954,
-      "learning_rate": 0.00038609201594655743,
-      "loss": 3.4585,
+      "epoch": 3.579514824797844,
+      "grad_norm": 0.6181750297546387,
+      "learning_rate": 0.00038572261198057203,
+      "loss": 3.4658,
       "step": 33200
     },
     {
-      "epoch": 3.5787321063394684,
-      "grad_norm": 0.6626840829849243,
-      "learning_rate": 0.000385768774916496,
-      "loss": 3.4542,
+      "epoch": 3.5849056603773586,
+      "grad_norm": 0.6425030827522278,
+      "learning_rate": 0.000385405288720993,
+      "loss": 3.4579,
       "step": 33250
     },
     {
-      "epoch": 3.5841136583790765,
-      "grad_norm": 0.5718008279800415,
-      "learning_rate": 0.0003854455338864346,
-      "loss": 3.4513,
+      "epoch": 3.5902964959568733,
+      "grad_norm": 0.6708866953849792,
+      "learning_rate": 0.00038508148947652453,
+      "loss": 3.4479,
       "step": 33300
     },
     {
-      "epoch": 3.5894952104186846,
-      "grad_norm": 0.6096655130386353,
-      "learning_rate": 0.0003851222928563732,
-      "loss": 3.4395,
+      "epoch": 3.595687331536388,
+      "grad_norm": 0.6270230412483215,
+      "learning_rate": 0.00038475769023205614,
+      "loss": 3.4604,
       "step": 33350
     },
     {
-      "epoch": 3.594876762458293,
-      "grad_norm": 0.6503807306289673,
-      "learning_rate": 0.0003847990518263118,
-      "loss": 3.4605,
+      "epoch": 3.601078167115903,
+      "grad_norm": 0.5887383818626404,
+      "learning_rate": 0.00038443389098758763,
+      "loss": 3.4595,
       "step": 33400
     },
     {
-      "epoch": 3.6002583144979012,
-      "grad_norm": 0.726240873336792,
-      "learning_rate": 0.0003844822756168516,
-      "loss": 3.4619,
+      "epoch": 3.606469002695418,
+      "grad_norm": 0.6195206046104431,
+      "learning_rate": 0.00038411009174311924,
+      "loss": 3.4709,
       "step": 33450
     },
     {
-      "epoch": 3.6056398665375093,
-      "grad_norm": 0.6151783466339111,
-      "learning_rate": 0.00038415903458679016,
-      "loss": 3.4497,
+      "epoch": 3.6118598382749325,
+      "grad_norm": 0.6247096061706543,
+      "learning_rate": 0.0003837862924986508,
+      "loss": 3.4778,
       "step": 33500
     },
     {
-      "epoch": 3.611021418577118,
-      "grad_norm": 0.6405103206634521,
-      "learning_rate": 0.00038383579355672875,
-      "loss": 3.4242,
+      "epoch": 3.617250673854447,
+      "grad_norm": 0.6405837535858154,
+      "learning_rate": 0.00038346249325418234,
+      "loss": 3.4557,
       "step": 33550
     },
     {
-      "epoch": 3.616402970616726,
-      "grad_norm": 0.632905125617981,
-      "learning_rate": 0.00038351255252666735,
-      "loss": 3.462,
+      "epoch": 3.6226415094339623,
+      "grad_norm": 0.6239786148071289,
+      "learning_rate": 0.00038313869400971395,
+      "loss": 3.4593,
       "step": 33600
     },
     {
-      "epoch": 3.621784522656334,
-      "grad_norm": 0.6433292627334595,
-      "learning_rate": 0.00038318931149660594,
-      "loss": 3.4729,
+      "epoch": 3.628032345013477,
+      "grad_norm": 0.6770374178886414,
+      "learning_rate": 0.0003828148947652455,
+      "loss": 3.4566,
       "step": 33650
     },
     {
-      "epoch": 3.627166074695942,
-      "grad_norm": 0.6479583382606506,
-      "learning_rate": 0.00038286607046654454,
-      "loss": 3.4475,
+      "epoch": 3.6334231805929917,
+      "grad_norm": 0.6650347113609314,
+      "learning_rate": 0.0003824910955207771,
+      "loss": 3.4415,
       "step": 33700
     },
     {
-      "epoch": 3.6325476267355503,
-      "grad_norm": 0.6312809586524963,
-      "learning_rate": 0.00038254282943648313,
-      "loss": 3.4518,
+      "epoch": 3.638814016172507,
+      "grad_norm": 0.6508654952049255,
+      "learning_rate": 0.00038216729627630865,
+      "loss": 3.4676,
       "step": 33750
     },
     {
-      "epoch": 3.637929178775159,
-      "grad_norm": 0.5925193428993225,
-      "learning_rate": 0.00038221958840642167,
-      "loss": 3.452,
+      "epoch": 3.6442048517520216,
+      "grad_norm": 0.638340175151825,
+      "learning_rate": 0.00038184349703184026,
+      "loss": 3.447,
       "step": 33800
     },
     {
-      "epoch": 3.643310730814767,
-      "grad_norm": 0.6522447466850281,
-      "learning_rate": 0.00038189634737636027,
-      "loss": 3.4453,
+      "epoch": 3.6495956873315363,
+      "grad_norm": 0.6654194593429565,
+      "learning_rate": 0.0003815196977873718,
+      "loss": 3.478,
       "step": 33850
     },
     {
-      "epoch": 3.648692282854375,
-      "grad_norm": 0.6305121183395386,
-      "learning_rate": 0.0003815731063462988,
-      "loss": 3.4527,
+      "epoch": 3.6549865229110514,
+      "grad_norm": 0.6244774460792542,
+      "learning_rate": 0.0003811958985429034,
+      "loss": 3.4537,
       "step": 33900
     },
     {
-      "epoch": 3.6540738348939836,
-      "grad_norm": 0.5587605834007263,
-      "learning_rate": 0.00038124986531623745,
-      "loss": 3.4464,
+      "epoch": 3.660377358490566,
+      "grad_norm": 0.6253437399864197,
+      "learning_rate": 0.00038087209929843496,
+      "loss": 3.4535,
       "step": 33950
     },
     {
-      "epoch": 3.6594553869335917,
-      "grad_norm": 0.6513285040855408,
-      "learning_rate": 0.00038092662428617605,
-      "loss": 3.4556,
+      "epoch": 3.665768194070081,
+      "grad_norm": 0.6101527214050293,
+      "learning_rate": 0.00038054830005396646,
+      "loss": 3.4643,
       "step": 34000
     },
     {
-      "epoch": 3.6594553869335917,
-      "eval_accuracy": 0.3747203410483765,
-      "eval_loss": 3.457984685897827,
-      "eval_runtime": 186.4737,
-      "eval_samples_per_second": 96.587,
-      "eval_steps_per_second": 6.038,
+      "epoch": 3.665768194070081,
+      "eval_accuracy": 0.3732578731783936,
+      "eval_loss": 3.4692325592041016,
+      "eval_runtime": 184.5179,
+      "eval_samples_per_second": 97.611,
+      "eval_steps_per_second": 6.102,
       "step": 34000
     },
     {
-      "epoch": 3.6648369389732,
-      "grad_norm": 0.7378771305084229,
-      "learning_rate": 0.0003806033832561146,
-      "loss": 3.4306,
+      "epoch": 3.671159029649596,
+      "grad_norm": 0.6090540289878845,
+      "learning_rate": 0.00038022450080949806,
+      "loss": 3.459,
       "step": 34050
     },
     {
-      "epoch": 3.670218491012808,
-      "grad_norm": 0.6052830815315247,
-      "learning_rate": 0.0003802801422260532,
-      "loss": 3.4673,
+      "epoch": 3.6765498652291106,
+      "grad_norm": 0.605751097202301,
+      "learning_rate": 0.0003799007015650296,
+      "loss": 3.4759,
       "step": 34100
     },
     {
-      "epoch": 3.675600043052416,
-      "grad_norm": 0.6993939876556396,
-      "learning_rate": 0.0003799569011959918,
-      "loss": 3.4539,
+      "epoch": 3.6819407008086253,
+      "grad_norm": 0.6620280742645264,
+      "learning_rate": 0.0003795769023205612,
+      "loss": 3.4725,
       "step": 34150
     },
     {
-      "epoch": 3.6809815950920246,
-      "grad_norm": 0.5885092616081238,
-      "learning_rate": 0.0003796336601659303,
-      "loss": 3.4365,
+      "epoch": 3.68733153638814,
+      "grad_norm": 0.6590796113014221,
+      "learning_rate": 0.00037925310307609277,
+      "loss": 3.4524,
       "step": 34200
     },
     {
-      "epoch": 3.6863631471316327,
-      "grad_norm": 0.5841361284255981,
-      "learning_rate": 0.00037931041913586897,
-      "loss": 3.4316,
+      "epoch": 3.6927223719676547,
+      "grad_norm": 0.6148298978805542,
+      "learning_rate": 0.0003789293038316244,
+      "loss": 3.4481,
       "step": 34250
     },
     {
-      "epoch": 3.691744699171241,
-      "grad_norm": 0.65105801820755,
-      "learning_rate": 0.00037898717810580756,
-      "loss": 3.4595,
+      "epoch": 3.69811320754717,
+      "grad_norm": 0.6003066301345825,
+      "learning_rate": 0.0003786055045871559,
+      "loss": 3.4473,
       "step": 34300
     },
     {
-      "epoch": 3.6971262512108494,
-      "grad_norm": 0.6178062558174133,
-      "learning_rate": 0.0003786639370757461,
-      "loss": 3.4596,
+      "epoch": 3.7035040431266846,
+      "grad_norm": 0.6199920177459717,
+      "learning_rate": 0.0003782817053426875,
+      "loss": 3.4526,
       "step": 34350
     },
     {
-      "epoch": 3.7025078032504575,
-      "grad_norm": 0.6483685970306396,
-      "learning_rate": 0.0003783406960456847,
-      "loss": 3.4821,
+      "epoch": 3.7088948787061993,
+      "grad_norm": 0.5896412134170532,
+      "learning_rate": 0.0003779579060982191,
+      "loss": 3.4478,
       "step": 34400
     },
     {
-      "epoch": 3.7078893552900656,
-      "grad_norm": 0.6345661282539368,
-      "learning_rate": 0.00037801745501562324,
-      "loss": 3.4529,
+      "epoch": 3.7142857142857144,
+      "grad_norm": 0.567940354347229,
+      "learning_rate": 0.00037763410685375063,
+      "loss": 3.4676,
       "step": 34450
     },
     {
-      "epoch": 3.713270907329674,
-      "grad_norm": 0.6499034762382507,
-      "learning_rate": 0.0003777006788061631,
-      "loss": 3.4565,
+      "epoch": 3.719676549865229,
+      "grad_norm": 0.6009531617164612,
+      "learning_rate": 0.00037731030760928223,
+      "loss": 3.4522,
       "step": 34500
     },
     {
-      "epoch": 3.7186524593692822,
-      "grad_norm": 0.6253588199615479,
-      "learning_rate": 0.00037737743777610164,
-      "loss": 3.4573,
+      "epoch": 3.725067385444744,
+      "grad_norm": 0.6172782182693481,
+      "learning_rate": 0.0003769865083648138,
+      "loss": 3.4441,
       "step": 34550
     },
     {
-      "epoch": 3.7240340114088903,
-      "grad_norm": 0.6538006067276001,
-      "learning_rate": 0.0003770541967460403,
-      "loss": 3.453,
+      "epoch": 3.730458221024259,
+      "grad_norm": 0.7144559025764465,
+      "learning_rate": 0.0003766627091203454,
+      "loss": 3.4644,
       "step": 34600
     },
     {
-      "epoch": 3.7294155634484984,
-      "grad_norm": 0.571806013584137,
-      "learning_rate": 0.0003767309557159789,
-      "loss": 3.439,
+      "epoch": 3.7358490566037736,
+      "grad_norm": 0.6669462323188782,
+      "learning_rate": 0.00037633890987587694,
+      "loss": 3.4636,
       "step": 34650
     },
     {
-      "epoch": 3.7347971154881066,
-      "grad_norm": 0.6002705097198486,
-      "learning_rate": 0.0003764077146859174,
-      "loss": 3.4536,
+      "epoch": 3.7412398921832883,
+      "grad_norm": 0.6669538617134094,
+      "learning_rate": 0.00037601511063140855,
+      "loss": 3.4732,
       "step": 34700
     },
     {
-      "epoch": 3.740178667527715,
-      "grad_norm": 0.6094973683357239,
-      "learning_rate": 0.000376084473655856,
-      "loss": 3.4596,
+      "epoch": 3.7466307277628035,
+      "grad_norm": 0.5949172377586365,
+      "learning_rate": 0.00037569131138694004,
+      "loss": 3.4686,
       "step": 34750
     },
     {
-      "epoch": 3.745560219567323,
-      "grad_norm": 0.5751643180847168,
-      "learning_rate": 0.00037576123262579456,
-      "loss": 3.456,
+      "epoch": 3.752021563342318,
+      "grad_norm": 0.633142352104187,
+      "learning_rate": 0.0003753675121424716,
+      "loss": 3.4499,
       "step": 34800
     },
     {
-      "epoch": 3.7509417716069313,
-      "grad_norm": 0.6586737632751465,
-      "learning_rate": 0.00037543799159573315,
-      "loss": 3.4575,
+      "epoch": 3.757412398921833,
+      "grad_norm": 0.7153587937355042,
+      "learning_rate": 0.0003750437128980032,
+      "loss": 3.4529,
       "step": 34850
     },
     {
-      "epoch": 3.75632332364654,
-      "grad_norm": 0.5949563384056091,
-      "learning_rate": 0.0003751147505656718,
-      "loss": 3.4461,
+      "epoch": 3.7628032345013476,
+      "grad_norm": 0.5956360101699829,
+      "learning_rate": 0.00037471991365353475,
+      "loss": 3.4684,
       "step": 34900
     },
     {
-      "epoch": 3.761704875686148,
-      "grad_norm": 0.612346887588501,
-      "learning_rate": 0.00037479150953561034,
-      "loss": 3.4536,
+      "epoch": 3.7681940700808623,
+      "grad_norm": 0.7301541566848755,
+      "learning_rate": 0.00037439611440906635,
+      "loss": 3.4591,
       "step": 34950
     },
     {
-      "epoch": 3.767086427725756,
-      "grad_norm": 0.6009963154792786,
-      "learning_rate": 0.00037446826850554894,
-      "loss": 3.4589,
+      "epoch": 3.7735849056603774,
+      "grad_norm": 0.6961854696273804,
+      "learning_rate": 0.0003740723151645979,
+      "loss": 3.4664,
       "step": 35000
     },
     {
-      "epoch": 3.767086427725756,
-      "eval_accuracy": 0.37542854057115127,
-      "eval_loss": 3.453930616378784,
-      "eval_runtime": 186.6741,
-      "eval_samples_per_second": 96.484,
-      "eval_steps_per_second": 6.032,
+      "epoch": 3.7735849056603774,
+      "eval_accuracy": 0.3742131493680585,
+      "eval_loss": 3.4640750885009766,
+      "eval_runtime": 184.1649,
+      "eval_samples_per_second": 97.798,
+      "eval_steps_per_second": 6.114,
       "step": 35000
     },
     {
-      "epoch": 3.772467979765364,
-      "grad_norm": 0.689411997795105,
-      "learning_rate": 0.00037414502747548753,
-      "loss": 3.4454,
+      "epoch": 3.778975741239892,
+      "grad_norm": 0.6273418664932251,
+      "learning_rate": 0.0003737485159201295,
+      "loss": 3.4756,
       "step": 35050
     },
     {
-      "epoch": 3.7778495318049723,
-      "grad_norm": 0.5965470671653748,
-      "learning_rate": 0.00037382178644542607,
-      "loss": 3.4649,
+      "epoch": 3.784366576819407,
+      "grad_norm": 0.6409321427345276,
+      "learning_rate": 0.00037342471667566106,
+      "loss": 3.4433,
       "step": 35100
     },
     {
-      "epoch": 3.783231083844581,
-      "grad_norm": 0.645729660987854,
-      "learning_rate": 0.00037349854541536467,
-      "loss": 3.4418,
+      "epoch": 3.789757412398922,
+      "grad_norm": 0.6555987596511841,
+      "learning_rate": 0.00037310091743119266,
+      "loss": 3.4664,
       "step": 35150
     },
     {
-      "epoch": 3.788612635884189,
-      "grad_norm": 0.6074603199958801,
-      "learning_rate": 0.0003731753043853033,
-      "loss": 3.4315,
+      "epoch": 3.7951482479784366,
+      "grad_norm": 0.6194452047348022,
+      "learning_rate": 0.0003727771181867242,
+      "loss": 3.4636,
       "step": 35200
     },
     {
-      "epoch": 3.793994187923797,
-      "grad_norm": 0.6144258975982666,
-      "learning_rate": 0.00037285206335524186,
-      "loss": 3.4453,
+      "epoch": 3.8005390835579513,
+      "grad_norm": 0.6294335126876831,
+      "learning_rate": 0.00037245331894225576,
+      "loss": 3.4449,
       "step": 35250
     },
     {
-      "epoch": 3.7993757399634056,
-      "grad_norm": 0.6099472641944885,
-      "learning_rate": 0.00037252882232518045,
-      "loss": 3.4285,
+      "epoch": 3.8059299191374665,
+      "grad_norm": 0.6391084790229797,
+      "learning_rate": 0.00037212951969778737,
+      "loss": 3.4586,
       "step": 35300
     },
     {
-      "epoch": 3.8047572920030137,
-      "grad_norm": 0.6138466000556946,
-      "learning_rate": 0.000372205581295119,
-      "loss": 3.4505,
+      "epoch": 3.811320754716981,
+      "grad_norm": 0.6358237266540527,
+      "learning_rate": 0.00037180572045331887,
+      "loss": 3.4602,
       "step": 35350
     },
     {
-      "epoch": 3.810138844042622,
-      "grad_norm": 0.6321321725845337,
-      "learning_rate": 0.0003718823402650576,
-      "loss": 3.4402,
+      "epoch": 3.816711590296496,
+      "grad_norm": 0.6337974071502686,
+      "learning_rate": 0.00037148192120885047,
+      "loss": 3.4612,
       "step": 35400
     },
     {
-      "epoch": 3.8155203960822304,
-      "grad_norm": 0.6698060631752014,
-      "learning_rate": 0.00037155909923499624,
-      "loss": 3.4472,
+      "epoch": 3.822102425876011,
+      "grad_norm": 0.6847667694091797,
+      "learning_rate": 0.000371158121964382,
+      "loss": 3.4542,
       "step": 35450
     },
     {
-      "epoch": 3.8209019481218385,
-      "grad_norm": 0.6117549538612366,
-      "learning_rate": 0.0003712358582049348,
-      "loss": 3.4677,
+      "epoch": 3.8274932614555257,
+      "grad_norm": 0.6090772747993469,
+      "learning_rate": 0.0003708343227199136,
+      "loss": 3.4563,
       "step": 35500
     },
     {
-      "epoch": 3.8262835001614466,
-      "grad_norm": 0.6063292026519775,
-      "learning_rate": 0.00037091261717487337,
-      "loss": 3.4539,
+      "epoch": 3.8328840970350404,
+      "grad_norm": 0.7042020559310913,
+      "learning_rate": 0.0003705105234754452,
+      "loss": 3.4699,
       "step": 35550
     },
     {
-      "epoch": 3.8316650522010547,
-      "grad_norm": 0.6454995274543762,
-      "learning_rate": 0.00037058937614481197,
-      "loss": 3.4489,
+      "epoch": 3.838274932614555,
+      "grad_norm": 0.623544454574585,
+      "learning_rate": 0.0003701867242309768,
+      "loss": 3.4755,
       "step": 35600
     },
     {
-      "epoch": 3.837046604240663,
-      "grad_norm": 0.6279877424240112,
-      "learning_rate": 0.0003702661351147505,
-      "loss": 3.4443,
+      "epoch": 3.8436657681940702,
+      "grad_norm": 0.6115444898605347,
+      "learning_rate": 0.00036986292498650833,
+      "loss": 3.4753,
       "step": 35650
     },
     {
-      "epoch": 3.8424281562802713,
-      "grad_norm": 0.6130351424217224,
-      "learning_rate": 0.0003699428940846891,
-      "loss": 3.4329,
+      "epoch": 3.849056603773585,
+      "grad_norm": 0.5954403281211853,
+      "learning_rate": 0.0003695391257420399,
+      "loss": 3.4746,
       "step": 35700
     },
     {
-      "epoch": 3.8478097083198795,
-      "grad_norm": 0.634748101234436,
-      "learning_rate": 0.00036961965305462775,
-      "loss": 3.4501,
+      "epoch": 3.8544474393530996,
+      "grad_norm": 0.6841748952865601,
+      "learning_rate": 0.0003692153264975715,
+      "loss": 3.4545,
       "step": 35750
     },
     {
-      "epoch": 3.8531912603594876,
-      "grad_norm": 0.5814838409423828,
-      "learning_rate": 0.0003692964120245663,
-      "loss": 3.4351,
+      "epoch": 3.8598382749326143,
+      "grad_norm": 0.6804195046424866,
+      "learning_rate": 0.00036889152725310304,
+      "loss": 3.4631,
       "step": 35800
     },
     {
-      "epoch": 3.858572812399096,
-      "grad_norm": 0.6361930966377258,
-      "learning_rate": 0.0003689731709945049,
-      "loss": 3.4543,
+      "epoch": 3.8652291105121295,
+      "grad_norm": 0.6171534657478333,
+      "learning_rate": 0.00036856772800863464,
+      "loss": 3.4356,
       "step": 35850
     },
     {
-      "epoch": 3.863954364438704,
-      "grad_norm": 0.7173475623130798,
-      "learning_rate": 0.0003686499299644434,
-      "loss": 3.4617,
+      "epoch": 3.870619946091644,
+      "grad_norm": 0.6135944724082947,
+      "learning_rate": 0.0003682439287641662,
+      "loss": 3.4521,
       "step": 35900
     },
     {
-      "epoch": 3.8693359164783123,
-      "grad_norm": 0.6407899856567383,
-      "learning_rate": 0.000368326688934382,
-      "loss": 3.4435,
+      "epoch": 3.876010781671159,
+      "grad_norm": 0.6205589175224304,
+      "learning_rate": 0.0003679201295196978,
+      "loss": 3.4763,
       "step": 35950
     },
     {
-      "epoch": 3.8747174685179204,
-      "grad_norm": 0.6248735189437866,
-      "learning_rate": 0.0003680034479043206,
-      "loss": 3.4463,
+      "epoch": 3.881401617250674,
+      "grad_norm": 0.6165568828582764,
+      "learning_rate": 0.00036759633027522935,
+      "loss": 3.4553,
       "step": 36000
     },
     {
-      "epoch": 3.8747174685179204,
-      "eval_accuracy": 0.37592497561557564,
-      "eval_loss": 3.4494788646698,
-      "eval_runtime": 186.8412,
-      "eval_samples_per_second": 96.397,
-      "eval_steps_per_second": 6.027,
+      "epoch": 3.881401617250674,
+      "eval_accuracy": 0.3748235612918003,
+      "eval_loss": 3.45568585395813,
+      "eval_runtime": 183.9456,
+      "eval_samples_per_second": 97.915,
+      "eval_steps_per_second": 6.121,
       "step": 36000
     },
     {
-      "epoch": 3.8800990205575285,
-      "grad_norm": 0.6424155235290527,
-      "learning_rate": 0.0003676802068742592,
-      "loss": 3.4373,
+      "epoch": 3.8867924528301887,
+      "grad_norm": 0.7112396955490112,
+      "learning_rate": 0.00036727253103076084,
+      "loss": 3.4645,
       "step": 36050
     },
     {
-      "epoch": 3.885480572597137,
-      "grad_norm": 0.6088709831237793,
-      "learning_rate": 0.0003673569658441978,
-      "loss": 3.4311,
+      "epoch": 3.8921832884097034,
+      "grad_norm": 0.6655269861221313,
+      "learning_rate": 0.00036694873178629245,
+      "loss": 3.4592,
       "step": 36100
     },
     {
-      "epoch": 3.890862124636745,
-      "grad_norm": 0.6665557026863098,
-      "learning_rate": 0.0003670337248141364,
-      "loss": 3.4559,
+      "epoch": 3.8975741239892185,
+      "grad_norm": 0.6312037706375122,
+      "learning_rate": 0.000366624932541824,
+      "loss": 3.4656,
       "step": 36150
     },
     {
-      "epoch": 3.8962436766763533,
-      "grad_norm": 0.6170324087142944,
-      "learning_rate": 0.00036671048378407494,
-      "loss": 3.4391,
+      "epoch": 3.9029649595687332,
+      "grad_norm": 0.6530061364173889,
+      "learning_rate": 0.0003663011332973556,
+      "loss": 3.4542,
       "step": 36200
     },
     {
-      "epoch": 3.901625228715962,
-      "grad_norm": 0.6090626120567322,
-      "learning_rate": 0.00036638724275401353,
-      "loss": 3.4581,
+      "epoch": 3.908355795148248,
+      "grad_norm": 0.5869830250740051,
+      "learning_rate": 0.00036597733405288715,
+      "loss": 3.4698,
       "step": 36250
     },
     {
-      "epoch": 3.90700678075557,
-      "grad_norm": 0.597235381603241,
-      "learning_rate": 0.0003660640017239522,
-      "loss": 3.4468,
+      "epoch": 3.913746630727763,
+      "grad_norm": 0.6425397396087646,
+      "learning_rate": 0.00036565353480841876,
+      "loss": 3.4497,
       "step": 36300
     },
     {
-      "epoch": 3.912388332795178,
-      "grad_norm": 0.6091225743293762,
-      "learning_rate": 0.0003657407606938907,
-      "loss": 3.441,
+      "epoch": 3.9191374663072778,
+      "grad_norm": 0.6408144235610962,
+      "learning_rate": 0.0003653297355639503,
+      "loss": 3.4717,
       "step": 36350
     },
     {
-      "epoch": 3.9177698848347866,
-      "grad_norm": 0.6485524773597717,
-      "learning_rate": 0.0003654175196638293,
-      "loss": 3.4551,
+      "epoch": 3.9245283018867925,
+      "grad_norm": 0.6111749410629272,
+      "learning_rate": 0.0003650059363194819,
+      "loss": 3.472,
       "step": 36400
     },
     {
-      "epoch": 3.9231514368743947,
-      "grad_norm": 0.6352561116218567,
-      "learning_rate": 0.00036509427863376786,
-      "loss": 3.4497,
+      "epoch": 3.929919137466307,
+      "grad_norm": 0.6590710282325745,
+      "learning_rate": 0.00036468213707501347,
+      "loss": 3.448,
       "step": 36450
     },
     {
-      "epoch": 3.928532988914003,
-      "grad_norm": 0.6507611870765686,
-      "learning_rate": 0.00036477103760370645,
-      "loss": 3.444,
+      "epoch": 3.935309973045822,
+      "grad_norm": 0.6589022874832153,
+      "learning_rate": 0.000364358337830545,
+      "loss": 3.4568,
       "step": 36500
     },
     {
-      "epoch": 3.933914540953611,
-      "grad_norm": 0.6284761428833008,
-      "learning_rate": 0.00036444779657364505,
-      "loss": 3.4295,
+      "epoch": 3.940700808625337,
+      "grad_norm": 0.6541099548339844,
+      "learning_rate": 0.0003640345385860766,
+      "loss": 3.4581,
       "step": 36550
     },
     {
-      "epoch": 3.939296092993219,
-      "grad_norm": 0.6066423654556274,
-      "learning_rate": 0.00036412455554358364,
-      "loss": 3.4514,
+      "epoch": 3.9460916442048517,
+      "grad_norm": 0.6201278567314148,
+      "learning_rate": 0.00036371073934160817,
+      "loss": 3.4557,
       "step": 36600
     },
     {
-      "epoch": 3.9446776450328276,
-      "grad_norm": 0.6189970374107361,
-      "learning_rate": 0.00036380131451352224,
-      "loss": 3.4345,
+      "epoch": 3.9514824797843664,
+      "grad_norm": 0.677780032157898,
+      "learning_rate": 0.0003633869400971398,
+      "loss": 3.4466,
       "step": 36650
     },
     {
-      "epoch": 3.9500591970724357,
-      "grad_norm": 0.6521716713905334,
-      "learning_rate": 0.00036347807348346083,
-      "loss": 3.4502,
+      "epoch": 3.9568733153638815,
+      "grad_norm": 0.6356298327445984,
+      "learning_rate": 0.00036306314085267127,
+      "loss": 3.456,
       "step": 36700
     },
     {
-      "epoch": 3.955440749112044,
-      "grad_norm": 0.6237977743148804,
-      "learning_rate": 0.00036315483245339937,
-      "loss": 3.4288,
+      "epoch": 3.9622641509433962,
+      "grad_norm": 0.6926275491714478,
+      "learning_rate": 0.00036273934160820293,
+      "loss": 3.449,
       "step": 36750
     },
     {
-      "epoch": 3.9608223011516523,
-      "grad_norm": 0.7352908253669739,
-      "learning_rate": 0.00036283159142333797,
-      "loss": 3.4411,
+      "epoch": 3.967654986522911,
+      "grad_norm": 0.6441013216972351,
+      "learning_rate": 0.00036241554236373443,
+      "loss": 3.4421,
       "step": 36800
     },
     {
-      "epoch": 3.9662038531912605,
-      "grad_norm": 0.6240646243095398,
-      "learning_rate": 0.0003625083503932765,
-      "loss": 3.4445,
+      "epoch": 3.973045822102426,
+      "grad_norm": 0.6412491202354431,
+      "learning_rate": 0.00036209174311926603,
+      "loss": 3.4592,
       "step": 36850
     },
     {
-      "epoch": 3.9715854052308686,
-      "grad_norm": 0.5882095694541931,
-      "learning_rate": 0.00036218510936321516,
-      "loss": 3.4523,
+      "epoch": 3.9784366576819408,
+      "grad_norm": 0.6484386324882507,
+      "learning_rate": 0.0003617679438747976,
+      "loss": 3.4484,
       "step": 36900
     },
     {
-      "epoch": 3.9769669572704767,
-      "grad_norm": 0.6271676421165466,
-      "learning_rate": 0.00036186186833315375,
-      "loss": 3.4528,
+      "epoch": 3.9838274932614555,
+      "grad_norm": 0.6936894059181213,
+      "learning_rate": 0.00036144414463032913,
+      "loss": 3.4586,
       "step": 36950
     },
     {
-      "epoch": 3.9823485093100848,
-      "grad_norm": 0.619851291179657,
-      "learning_rate": 0.0003615386273030923,
-      "loss": 3.46,
+      "epoch": 3.9892183288409706,
+      "grad_norm": 0.6636582612991333,
+      "learning_rate": 0.00036112034538586074,
+      "loss": 3.4537,
       "step": 37000
     },
     {
-      "epoch": 3.9823485093100848,
-      "eval_accuracy": 0.3766869583178186,
-      "eval_loss": 3.4407858848571777,
-      "eval_runtime": 186.9877,
-      "eval_samples_per_second": 96.322,
-      "eval_steps_per_second": 6.022,
+      "epoch": 3.9892183288409706,
+      "eval_accuracy": 0.3756027111503179,
+      "eval_loss": 3.449716329574585,
+      "eval_runtime": 184.186,
+      "eval_samples_per_second": 97.787,
+      "eval_steps_per_second": 6.113,
       "step": 37000
     },
     {
-      "epoch": 3.9877300613496933,
-      "grad_norm": 0.6204333305358887,
-      "learning_rate": 0.0003612153862730309,
-      "loss": 3.4546,
+      "epoch": 3.9946091644204853,
+      "grad_norm": 0.6402984857559204,
+      "learning_rate": 0.0003607965461413923,
+      "loss": 3.4669,
       "step": 37050
     },
     {
-      "epoch": 3.9931116133893014,
-      "grad_norm": 0.6192218661308289,
-      "learning_rate": 0.0003608921452429695,
-      "loss": 3.4466,
+      "epoch": 4.0,
+      "grad_norm": 1.287859320640564,
+      "learning_rate": 0.0003604727468969239,
+      "loss": 3.4387,
       "step": 37100
     },
     {
-      "epoch": 3.9984931654289095,
-      "grad_norm": 0.599573016166687,
-      "learning_rate": 0.000360568904212908,
-      "loss": 3.4448,
+      "epoch": 4.005390835579515,
+      "grad_norm": 0.6282164454460144,
+      "learning_rate": 0.00036014894765245544,
+      "loss": 3.3343,
       "step": 37150
     },
     {
-      "epoch": 4.003874717468518,
-      "grad_norm": 0.6175059676170349,
-      "learning_rate": 0.00036024566318284667,
-      "loss": 3.3896,
+      "epoch": 4.010781671159029,
+      "grad_norm": 0.6469982862472534,
+      "learning_rate": 0.00035982514840798705,
+      "loss": 3.3643,
       "step": 37200
     },
     {
-      "epoch": 4.009256269508126,
-      "grad_norm": 0.644124448299408,
-      "learning_rate": 0.00035992242215278526,
-      "loss": 3.3281,
+      "epoch": 4.0161725067385445,
+      "grad_norm": 0.6892557144165039,
+      "learning_rate": 0.00035950782514840795,
+      "loss": 3.3571,
       "step": 37250
     },
     {
-      "epoch": 4.014637821547734,
-      "grad_norm": 0.5991693139076233,
-      "learning_rate": 0.0003595991811227238,
-      "loss": 3.3428,
+      "epoch": 4.02156334231806,
+      "grad_norm": 0.6430110931396484,
+      "learning_rate": 0.00035918402590393955,
+      "loss": 3.361,
       "step": 37300
     },
     {
-      "epoch": 4.020019373587343,
-      "grad_norm": 0.6538693308830261,
-      "learning_rate": 0.0003592759400926624,
-      "loss": 3.3562,
+      "epoch": 4.026954177897574,
+      "grad_norm": 0.6850172877311707,
+      "learning_rate": 0.00035886022665947105,
+      "loss": 3.3486,
       "step": 37350
     },
     {
-      "epoch": 4.0254009256269505,
-      "grad_norm": 0.6204119324684143,
-      "learning_rate": 0.00035895269906260094,
-      "loss": 3.3339,
+      "epoch": 4.032345013477089,
+      "grad_norm": 0.645811915397644,
+      "learning_rate": 0.0003585364274150027,
+      "loss": 3.3686,
       "step": 37400
     },
     {
-      "epoch": 4.030782477666559,
-      "grad_norm": 0.6643719673156738,
-      "learning_rate": 0.0003586359228531408,
-      "loss": 3.3504,
+      "epoch": 4.037735849056604,
+      "grad_norm": 0.6480479836463928,
+      "learning_rate": 0.0003582126281705342,
+      "loss": 3.3651,
       "step": 37450
     },
     {
-      "epoch": 4.036164029706168,
-      "grad_norm": 0.6423208117485046,
-      "learning_rate": 0.00035831268182307934,
-      "loss": 3.3688,
+      "epoch": 4.0431266846361185,
+      "grad_norm": 0.6276705265045166,
+      "learning_rate": 0.0003578888289260658,
+      "loss": 3.3544,
       "step": 37500
     },
     {
-      "epoch": 4.041545581745775,
-      "grad_norm": 0.615216076374054,
-      "learning_rate": 0.000357989440793018,
-      "loss": 3.3521,
+      "epoch": 4.048517520215634,
+      "grad_norm": 0.6628215909004211,
+      "learning_rate": 0.00035756502968159736,
+      "loss": 3.3734,
       "step": 37550
     },
     {
-      "epoch": 4.046927133785384,
-      "grad_norm": 0.6236596703529358,
-      "learning_rate": 0.0003576661997629566,
-      "loss": 3.3533,
+      "epoch": 4.053908355795148,
+      "grad_norm": 0.6510850787162781,
+      "learning_rate": 0.00035724123043712896,
+      "loss": 3.3634,
       "step": 37600
     },
     {
-      "epoch": 4.0523086858249915,
-      "grad_norm": 0.6723390221595764,
-      "learning_rate": 0.0003573429587328951,
-      "loss": 3.3728,
+      "epoch": 4.059299191374663,
+      "grad_norm": 0.663616955280304,
+      "learning_rate": 0.0003569174311926605,
+      "loss": 3.3782,
       "step": 37650
     },
     {
-      "epoch": 4.0576902378646,
-      "grad_norm": 0.5689955949783325,
-      "learning_rate": 0.0003570197177028337,
-      "loss": 3.3683,
+      "epoch": 4.064690026954178,
+      "grad_norm": 0.6863150000572205,
+      "learning_rate": 0.00035659363194819206,
+      "loss": 3.3634,
       "step": 37700
     },
     {
-      "epoch": 4.063071789904209,
-      "grad_norm": 0.5888593196868896,
-      "learning_rate": 0.00035669647667277226,
-      "loss": 3.3812,
+      "epoch": 4.070080862533692,
+      "grad_norm": 0.6542084813117981,
+      "learning_rate": 0.00035626983270372367,
+      "loss": 3.3843,
       "step": 37750
     },
     {
-      "epoch": 4.068453341943816,
-      "grad_norm": 0.6122689247131348,
-      "learning_rate": 0.00035637323564271085,
-      "loss": 3.356,
+      "epoch": 4.0754716981132075,
+      "grad_norm": 0.7337487936019897,
+      "learning_rate": 0.0003559460334592552,
+      "loss": 3.3681,
       "step": 37800
     },
     {
-      "epoch": 4.073834893983425,
-      "grad_norm": 0.6232444047927856,
-      "learning_rate": 0.0003560499946126495,
-      "loss": 3.3544,
+      "epoch": 4.080862533692723,
+      "grad_norm": 0.6522841453552246,
+      "learning_rate": 0.0003556222342147868,
+      "loss": 3.3665,
       "step": 37850
     },
     {
-      "epoch": 4.079216446023033,
-      "grad_norm": 0.6149317622184753,
-      "learning_rate": 0.00035572675358258804,
-      "loss": 3.3774,
+      "epoch": 4.086253369272237,
+      "grad_norm": 0.671728253364563,
+      "learning_rate": 0.0003552984349703184,
+      "loss": 3.3662,
       "step": 37900
     },
     {
-      "epoch": 4.084597998062641,
-      "grad_norm": 0.6281867623329163,
-      "learning_rate": 0.00035540351255252664,
-      "loss": 3.3497,
+      "epoch": 4.091644204851752,
+      "grad_norm": 0.638214647769928,
+      "learning_rate": 0.00035497463572585,
+      "loss": 3.3772,
       "step": 37950
     },
     {
-      "epoch": 4.08997955010225,
-      "grad_norm": 0.6363036036491394,
-      "learning_rate": 0.00035508027152246523,
-      "loss": 3.3706,
+      "epoch": 4.097035040431267,
+      "grad_norm": 0.6835560202598572,
+      "learning_rate": 0.00035465083648138153,
+      "loss": 3.3703,
       "step": 38000
     },
     {
-      "epoch": 4.08997955010225,
-      "eval_accuracy": 0.3769571780498132,
-      "eval_loss": 3.445598840713501,
-      "eval_runtime": 186.5,
-      "eval_samples_per_second": 96.574,
-      "eval_steps_per_second": 6.038,
+      "epoch": 4.097035040431267,
+      "eval_accuracy": 0.37589248840211914,
+      "eval_loss": 3.45267915725708,
+      "eval_runtime": 184.0592,
+      "eval_samples_per_second": 97.854,
+      "eval_steps_per_second": 6.118,
       "step": 38000
     },
     {
-      "epoch": 4.095361102141858,
-      "grad_norm": 0.6738185882568359,
-      "learning_rate": 0.0003547570304924038,
-      "loss": 3.3815,
+      "epoch": 4.1024258760107815,
+      "grad_norm": 0.6934317350387573,
+      "learning_rate": 0.00035432703723691314,
+      "loss": 3.3731,
       "step": 38050
     },
     {
-      "epoch": 4.100742654181466,
-      "grad_norm": 0.6790494918823242,
-      "learning_rate": 0.0003544337894623424,
-      "loss": 3.389,
+      "epoch": 4.107816711590297,
+      "grad_norm": 0.6698592305183411,
+      "learning_rate": 0.00035400323799244463,
+      "loss": 3.3791,
       "step": 38100
     },
     {
-      "epoch": 4.106124206221074,
-      "grad_norm": 0.6565952301025391,
-      "learning_rate": 0.000354110548432281,
-      "loss": 3.3778,
+      "epoch": 4.113207547169812,
+      "grad_norm": 0.6645461916923523,
+      "learning_rate": 0.0003536794387479762,
+      "loss": 3.3871,
       "step": 38150
     },
     {
-      "epoch": 4.111505758260682,
-      "grad_norm": 0.6302433609962463,
-      "learning_rate": 0.00035378730740221956,
-      "loss": 3.3882,
+      "epoch": 4.118598382749326,
+      "grad_norm": 0.6031655669212341,
+      "learning_rate": 0.0003533556395035078,
+      "loss": 3.3665,
       "step": 38200
     },
     {
-      "epoch": 4.1168873103002905,
-      "grad_norm": 0.6297251582145691,
-      "learning_rate": 0.00035346406637215815,
-      "loss": 3.3703,
+      "epoch": 4.123989218328841,
+      "grad_norm": 0.6706673502922058,
+      "learning_rate": 0.00035303184025903934,
+      "loss": 3.3709,
       "step": 38250
     },
     {
-      "epoch": 4.122268862339899,
-      "grad_norm": 0.6215413212776184,
-      "learning_rate": 0.0003531408253420967,
-      "loss": 3.3705,
+      "epoch": 4.129380053908355,
+      "grad_norm": 0.6412116289138794,
+      "learning_rate": 0.00035270804101457094,
+      "loss": 3.3824,
       "step": 38300
     },
     {
-      "epoch": 4.127650414379507,
-      "grad_norm": 0.6195633411407471,
-      "learning_rate": 0.0003528175843120353,
-      "loss": 3.3602,
+      "epoch": 4.1347708894878705,
+      "grad_norm": 0.6715707182884216,
+      "learning_rate": 0.0003523842417701025,
+      "loss": 3.3617,
       "step": 38350
     },
     {
-      "epoch": 4.133031966419115,
-      "grad_norm": 0.7500778436660767,
-      "learning_rate": 0.00035249434328197394,
-      "loss": 3.3651,
+      "epoch": 4.140161725067386,
+      "grad_norm": 0.6099421977996826,
+      "learning_rate": 0.0003520604425256341,
+      "loss": 3.3858,
       "step": 38400
     },
     {
-      "epoch": 4.138413518458724,
-      "grad_norm": 0.6209246516227722,
-      "learning_rate": 0.0003521711022519125,
-      "loss": 3.3767,
+      "epoch": 4.1455525606469,
+      "grad_norm": 0.6689791083335876,
+      "learning_rate": 0.00035173664328116565,
+      "loss": 3.369,
       "step": 38450
     },
     {
-      "epoch": 4.1437950704983315,
-      "grad_norm": 0.6420706510543823,
-      "learning_rate": 0.00035184786122185107,
-      "loss": 3.3709,
+      "epoch": 4.150943396226415,
+      "grad_norm": 0.624198317527771,
+      "learning_rate": 0.00035141284403669725,
+      "loss": 3.3789,
       "step": 38500
     },
     {
-      "epoch": 4.14917662253794,
-      "grad_norm": 0.6342754364013672,
-      "learning_rate": 0.00035152462019178967,
-      "loss": 3.3707,
+      "epoch": 4.15633423180593,
+      "grad_norm": 0.6686607003211975,
+      "learning_rate": 0.0003510890447922288,
+      "loss": 3.4051,
       "step": 38550
     },
     {
-      "epoch": 4.154558174577549,
-      "grad_norm": 0.6302380561828613,
-      "learning_rate": 0.0003512013791617282,
-      "loss": 3.3651,
+      "epoch": 4.1617250673854445,
+      "grad_norm": 0.6285902261734009,
+      "learning_rate": 0.00035076524554776035,
+      "loss": 3.3777,
       "step": 38600
     },
     {
-      "epoch": 4.159939726617156,
-      "grad_norm": 0.6601458787918091,
-      "learning_rate": 0.0003508781381316668,
-      "loss": 3.3795,
+      "epoch": 4.16711590296496,
+      "grad_norm": 0.6666203141212463,
+      "learning_rate": 0.00035044144630329196,
+      "loss": 3.3915,
       "step": 38650
     },
     {
-      "epoch": 4.165321278656765,
-      "grad_norm": 0.6050419211387634,
-      "learning_rate": 0.00035055489710160545,
-      "loss": 3.3741,
+      "epoch": 4.172506738544475,
+      "grad_norm": 0.7009741067886353,
+      "learning_rate": 0.00035011764705882346,
+      "loss": 3.3581,
       "step": 38700
     },
     {
-      "epoch": 4.1707028306963725,
-      "grad_norm": 0.6658854484558105,
-      "learning_rate": 0.000350231656071544,
-      "loss": 3.3694,
+      "epoch": 4.177897574123989,
+      "grad_norm": 0.6324627995491028,
+      "learning_rate": 0.0003497938478143551,
+      "loss": 3.382,
       "step": 38750
     },
     {
-      "epoch": 4.176084382735981,
-      "grad_norm": 0.6647228598594666,
-      "learning_rate": 0.0003499084150414826,
-      "loss": 3.3825,
+      "epoch": 4.183288409703504,
+      "grad_norm": 0.6125489473342896,
+      "learning_rate": 0.0003494700485698866,
+      "loss": 3.3864,
       "step": 38800
     },
     {
-      "epoch": 4.18146593477559,
-      "grad_norm": 0.654331386089325,
-      "learning_rate": 0.0003495851740114211,
-      "loss": 3.3722,
+      "epoch": 4.188679245283019,
+      "grad_norm": 0.676408588886261,
+      "learning_rate": 0.0003491462493254182,
+      "loss": 3.3898,
       "step": 38850
     },
     {
-      "epoch": 4.186847486815197,
-      "grad_norm": 0.6393372416496277,
-      "learning_rate": 0.0003492619329813597,
-      "loss": 3.3771,
+      "epoch": 4.1940700808625335,
+      "grad_norm": 0.6478334665298462,
+      "learning_rate": 0.00034882245008094977,
+      "loss": 3.3949,
       "step": 38900
     },
     {
-      "epoch": 4.192229038854806,
-      "grad_norm": 0.6229539513587952,
-      "learning_rate": 0.0003489386919512983,
-      "loss": 3.3805,
+      "epoch": 4.199460916442049,
+      "grad_norm": 0.6765270829200745,
+      "learning_rate": 0.0003484986508364813,
+      "loss": 3.3898,
       "step": 38950
     },
     {
-      "epoch": 4.197610590894414,
-      "grad_norm": 0.6732252836227417,
-      "learning_rate": 0.0003486154509212369,
-      "loss": 3.3851,
+      "epoch": 4.204851752021563,
+      "grad_norm": 0.6771070957183838,
+      "learning_rate": 0.0003481748515920129,
+      "loss": 3.37,
       "step": 39000
     },
     {
-      "epoch": 4.197610590894414,
-      "eval_accuracy": 0.3772557561855274,
-      "eval_loss": 3.442148208618164,
-      "eval_runtime": 186.5144,
-      "eval_samples_per_second": 96.566,
-      "eval_steps_per_second": 6.037,
+      "epoch": 4.204851752021563,
+      "eval_accuracy": 0.37679169970167176,
+      "eval_loss": 3.4477601051330566,
+      "eval_runtime": 185.2031,
+      "eval_samples_per_second": 97.25,
+      "eval_steps_per_second": 6.08,
       "step": 39000
     },
     {
-      "epoch": 4.202992142934022,
-      "grad_norm": 0.6215004324913025,
-      "learning_rate": 0.0003482922098911755,
-      "loss": 3.373,
+      "epoch": 4.210242587601078,
+      "grad_norm": 0.67899090051651,
+      "learning_rate": 0.00034785105234754447,
+      "loss": 3.3981,
       "step": 39050
     },
     {
-      "epoch": 4.208373694973631,
-      "grad_norm": 0.6126363277435303,
-      "learning_rate": 0.0003479689688611141,
-      "loss": 3.3703,
+      "epoch": 4.215633423180593,
+      "grad_norm": 0.6708850860595703,
+      "learning_rate": 0.0003475272531030761,
+      "loss": 3.389,
       "step": 39100
     },
     {
-      "epoch": 4.213755247013238,
-      "grad_norm": 0.654690146446228,
-      "learning_rate": 0.00034764572783105264,
-      "loss": 3.3639,
+      "epoch": 4.2210242587601075,
+      "grad_norm": 0.6668167114257812,
+      "learning_rate": 0.00034720345385860763,
+      "loss": 3.401,
       "step": 39150
     },
     {
-      "epoch": 4.219136799052847,
-      "grad_norm": 0.6122671365737915,
-      "learning_rate": 0.00034732248680099123,
-      "loss": 3.3673,
+      "epoch": 4.226415094339623,
+      "grad_norm": 0.6457695364952087,
+      "learning_rate": 0.00034687965461413923,
+      "loss": 3.3812,
       "step": 39200
     },
     {
-      "epoch": 4.224518351092455,
-      "grad_norm": 0.6818833947181702,
-      "learning_rate": 0.0003469992457709299,
-      "loss": 3.3897,
+      "epoch": 4.231805929919138,
+      "grad_norm": 0.6423382759094238,
+      "learning_rate": 0.0003465558553696708,
+      "loss": 3.3772,
       "step": 39250
     },
     {
-      "epoch": 4.229899903132063,
-      "grad_norm": 0.6373806595802307,
-      "learning_rate": 0.0003466760047408684,
-      "loss": 3.3832,
+      "epoch": 4.237196765498652,
+      "grad_norm": 0.6771441698074341,
+      "learning_rate": 0.00034623853211009173,
+      "loss": 3.3916,
       "step": 39300
     },
     {
-      "epoch": 4.2352814551716715,
-      "grad_norm": 0.6495124101638794,
-      "learning_rate": 0.000346352763710807,
-      "loss": 3.3596,
+      "epoch": 4.242587601078167,
+      "grad_norm": 0.6296360492706299,
+      "learning_rate": 0.0003459212088505127,
+      "loss": 3.3779,
       "step": 39350
     },
     {
-      "epoch": 4.24066300721128,
-      "grad_norm": 0.648654580116272,
-      "learning_rate": 0.00034602952268074556,
-      "loss": 3.3731,
+      "epoch": 4.247978436657682,
+      "grad_norm": 0.6631340384483337,
+      "learning_rate": 0.0003455974096060442,
+      "loss": 3.3983,
       "step": 39400
     },
     {
-      "epoch": 4.246044559250888,
-      "grad_norm": 0.6400829553604126,
-      "learning_rate": 0.00034570628165068415,
-      "loss": 3.3751,
+      "epoch": 4.2533692722371965,
+      "grad_norm": 0.666734516620636,
+      "learning_rate": 0.0003452736103615758,
+      "loss": 3.3861,
       "step": 39450
     },
     {
-      "epoch": 4.251426111290496,
-      "grad_norm": 0.6451199650764465,
-      "learning_rate": 0.0003453830406206227,
-      "loss": 3.375,
+      "epoch": 4.258760107816712,
+      "grad_norm": 0.6757256388664246,
+      "learning_rate": 0.00034494981111710734,
+      "loss": 3.392,
       "step": 39500
     },
     {
-      "epoch": 4.256807663330104,
-      "grad_norm": 0.6402846574783325,
-      "learning_rate": 0.00034505979959056134,
-      "loss": 3.3836,
+      "epoch": 4.264150943396227,
+      "grad_norm": 0.645785927772522,
+      "learning_rate": 0.00034462601187263894,
+      "loss": 3.3956,
       "step": 39550
     },
     {
-      "epoch": 4.2621892153697125,
-      "grad_norm": 0.6590396165847778,
-      "learning_rate": 0.00034473655856049994,
-      "loss": 3.3797,
+      "epoch": 4.269541778975741,
+      "grad_norm": 0.6643311977386475,
+      "learning_rate": 0.0003443022126281705,
+      "loss": 3.4036,
       "step": 39600
     },
     {
-      "epoch": 4.267570767409321,
-      "grad_norm": 0.6491893529891968,
-      "learning_rate": 0.0003444133175304385,
-      "loss": 3.3775,
+      "epoch": 4.274932614555256,
+      "grad_norm": 0.6284821033477783,
+      "learning_rate": 0.00034397841338370204,
+      "loss": 3.3999,
       "step": 39650
     },
     {
-      "epoch": 4.272952319448929,
-      "grad_norm": 0.6648213863372803,
-      "learning_rate": 0.00034409007650037707,
-      "loss": 3.3774,
+      "epoch": 4.280323450134771,
+      "grad_norm": 0.6818395256996155,
+      "learning_rate": 0.00034365461413923365,
+      "loss": 3.3802,
       "step": 39700
     },
     {
-      "epoch": 4.278333871488537,
-      "grad_norm": 0.6316584944725037,
-      "learning_rate": 0.00034376683547031567,
-      "loss": 3.3912,
+      "epoch": 4.285714285714286,
+      "grad_norm": 0.6951792240142822,
+      "learning_rate": 0.0003433308148947652,
+      "loss": 3.3885,
       "step": 39750
     },
     {
-      "epoch": 4.283715423528146,
-      "grad_norm": 0.6376301646232605,
-      "learning_rate": 0.0003434435944402542,
-      "loss": 3.3706,
+      "epoch": 4.291105121293801,
+      "grad_norm": 0.7116639018058777,
+      "learning_rate": 0.0003430070156502968,
+      "loss": 3.3879,
       "step": 39800
     },
     {
-      "epoch": 4.2890969755677535,
-      "grad_norm": 0.6191912889480591,
-      "learning_rate": 0.00034312035341019286,
-      "loss": 3.3857,
+      "epoch": 4.296495956873315,
+      "grad_norm": 0.6331270933151245,
+      "learning_rate": 0.00034268321640582835,
+      "loss": 3.3888,
       "step": 39850
     },
     {
-      "epoch": 4.294478527607362,
-      "grad_norm": 0.6177436113357544,
-      "learning_rate": 0.00034279711238013145,
-      "loss": 3.3843,
+      "epoch": 4.30188679245283,
+      "grad_norm": 0.7609684467315674,
+      "learning_rate": 0.00034235941716135996,
+      "loss": 3.3956,
       "step": 39900
     },
     {
-      "epoch": 4.299860079646971,
-      "grad_norm": 0.6867303252220154,
-      "learning_rate": 0.00034247387135007,
-      "loss": 3.3935,
+      "epoch": 4.307277628032345,
+      "grad_norm": 0.6793885827064514,
+      "learning_rate": 0.0003420356179168915,
+      "loss": 3.3843,
       "step": 39950
     },
     {
-      "epoch": 4.305241631686578,
-      "grad_norm": 0.6567016839981079,
-      "learning_rate": 0.0003421506303200086,
-      "loss": 3.389,
+      "epoch": 4.3126684636118595,
+      "grad_norm": 0.6133043766021729,
+      "learning_rate": 0.0003417118186724231,
+      "loss": 3.3935,
       "step": 40000
     },
     {
-      "epoch": 4.305241631686578,
-      "eval_accuracy": 0.37798764203784574,
-      "eval_loss": 3.4361512660980225,
-      "eval_runtime": 186.582,
-      "eval_samples_per_second": 96.531,
-      "eval_steps_per_second": 6.035,
+      "epoch": 4.3126684636118595,
+      "eval_accuracy": 0.37689013921803166,
+      "eval_loss": 3.4450483322143555,
+      "eval_runtime": 187.7716,
+      "eval_samples_per_second": 95.92,
+      "eval_steps_per_second": 5.997,
       "step": 40000
     }
   ],
   "logging_steps": 50,
-  "max_steps": 92910,
+  "max_steps": 92750,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 10,
   "save_steps": 10000,
@@ -5986,7 +5986,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 3.3443812343808e+17,
+  "total_flos": 3.344276717568e+17,
   "train_batch_size": 32,
   "trial_name": null,
   "trial_params": null