diff --git "a/checkpoint-70000/trainer_state.json" "b/checkpoint-70000/trainer_state.json"
--- "a/checkpoint-70000/trainer_state.json"
+++ "b/checkpoint-70000/trainer_state.json"
@@ -1,7 +1,7 @@
 {
-  "best_metric": 3.3461689949035645,
-  "best_model_checkpoint": "/scratch/cl5625/exceptions/models/100M_495/checkpoint-70000",
-  "epoch": 7.534172855451512,
+  "best_metric": 3.347477912902832,
+  "best_model_checkpoint": "/scratch/cl5625/exceptions/models/100M_high_1000_495/checkpoint-70000",
+  "epoch": 7.547169811320755,
   "eval_steps": 1000,
   "global_step": 70000,
   "is_hyper_param_search": false,
@@ -9,10438 +9,10438 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.005381552039608223,
-      "grad_norm": 12.074577331542969,
+      "epoch": 0.005390835579514825,
+      "grad_norm": 2.323624849319458,
       "learning_rate": 0.0003,
-      "loss": 8.6374,
+      "loss": 8.5206,
       "step": 50
     },
     {
-      "epoch": 0.010763104079216447,
-      "grad_norm": 1.3158553838729858,
+      "epoch": 0.01078167115902965,
+      "grad_norm": 2.057921886444092,
       "learning_rate": 0.0006,
-      "loss": 6.8912,
+      "loss": 6.8833,
       "step": 100
     },
     {
-      "epoch": 0.01614465611882467,
-      "grad_norm": 2.628305196762085,
-      "learning_rate": 0.0005996767589699385,
-      "loss": 6.4839,
+      "epoch": 0.016172506738544475,
+      "grad_norm": 1.1483490467071533,
+      "learning_rate": 0.0005996762007555315,
+      "loss": 6.4306,
       "step": 150
     },
     {
-      "epoch": 0.021526208158432893,
-      "grad_norm": 1.0599905252456665,
-      "learning_rate": 0.0005993535179398771,
-      "loss": 6.2264,
+      "epoch": 0.0215633423180593,
+      "grad_norm": 3.967722177505493,
+      "learning_rate": 0.000599352401511063,
+      "loss": 6.2321,
       "step": 200
     },
     {
-      "epoch": 0.026907760198041114,
-      "grad_norm": 1.638780951499939,
-      "learning_rate": 0.0005990302769098158,
-      "loss": 6.0419,
+      "epoch": 0.026954177897574125,
+      "grad_norm": 2.4125149250030518,
+      "learning_rate": 0.0005990286022665946,
+      "loss": 6.0666,
       "step": 250
     },
     {
-      "epoch": 0.03228931223764934,
-      "grad_norm": 1.9191187620162964,
-      "learning_rate": 0.0005987070358797543,
-      "loss": 5.955,
+      "epoch": 0.03234501347708895,
+      "grad_norm": 2.052870750427246,
+      "learning_rate": 0.0005987048030221263,
+      "loss": 5.9672,
       "step": 300
     },
     {
-      "epoch": 0.03767086427725756,
-      "grad_norm": 1.12886643409729,
-      "learning_rate": 0.0005983837948496929,
-      "loss": 5.8538,
+      "epoch": 0.03773584905660377,
+      "grad_norm": 1.2476660013198853,
+      "learning_rate": 0.0005983810037776578,
+      "loss": 5.8655,
       "step": 350
     },
     {
-      "epoch": 0.04305241631686579,
-      "grad_norm": 1.3254728317260742,
-      "learning_rate": 0.0005980605538196314,
-      "loss": 5.7663,
+      "epoch": 0.0431266846361186,
+      "grad_norm": 0.8398758172988892,
+      "learning_rate": 0.0005980572045331894,
+      "loss": 5.8111,
       "step": 400
     },
     {
-      "epoch": 0.048433968356474004,
-      "grad_norm": 2.3004322052001953,
-      "learning_rate": 0.0005977373127895701,
-      "loss": 5.6952,
+      "epoch": 0.04851752021563342,
+      "grad_norm": 1.2200261354446411,
+      "learning_rate": 0.0005977334052887209,
+      "loss": 5.736,
       "step": 450
     },
     {
-      "epoch": 0.05381552039608223,
-      "grad_norm": 1.495474934577942,
-      "learning_rate": 0.0005974140717595086,
-      "loss": 5.6341,
+      "epoch": 0.05390835579514825,
+      "grad_norm": 1.2435953617095947,
+      "learning_rate": 0.0005974096060442526,
+      "loss": 5.6756,
       "step": 500
     },
     {
-      "epoch": 0.05919707243569045,
-      "grad_norm": 0.9526841640472412,
-      "learning_rate": 0.0005970908307294472,
-      "loss": 5.5508,
+      "epoch": 0.05929919137466307,
+      "grad_norm": 2.038189649581909,
+      "learning_rate": 0.0005970858067997841,
+      "loss": 5.5615,
       "step": 550
     },
     {
-      "epoch": 0.06457862447529868,
-      "grad_norm": 1.1846691370010376,
-      "learning_rate": 0.0005967675896993858,
-      "loss": 5.4745,
+      "epoch": 0.0646900269541779,
+      "grad_norm": 1.168102741241455,
+      "learning_rate": 0.0005967620075553157,
+      "loss": 5.529,
       "step": 600
     },
     {
-      "epoch": 0.0699601765149069,
-      "grad_norm": 1.5218234062194824,
-      "learning_rate": 0.0005964443486693243,
-      "loss": 5.4167,
+      "epoch": 0.07008086253369272,
+      "grad_norm": 1.0181015729904175,
+      "learning_rate": 0.0005964382083108472,
+      "loss": 5.4185,
       "step": 650
     },
     {
-      "epoch": 0.07534172855451512,
-      "grad_norm": 1.037400484085083,
-      "learning_rate": 0.000596121107639263,
-      "loss": 5.3214,
+      "epoch": 0.07547169811320754,
+      "grad_norm": 1.2549965381622314,
+      "learning_rate": 0.0005961144090663788,
+      "loss": 5.4023,
       "step": 700
     },
     {
-      "epoch": 0.08072328059412334,
-      "grad_norm": 1.1252976655960083,
-      "learning_rate": 0.0005957978666092015,
-      "loss": 5.3002,
+      "epoch": 0.08086253369272237,
+      "grad_norm": 1.1174157857894897,
+      "learning_rate": 0.0005957906098219104,
+      "loss": 5.3263,
       "step": 750
     },
     {
-      "epoch": 0.08610483263373157,
-      "grad_norm": 0.9828529953956604,
-      "learning_rate": 0.0005954746255791401,
-      "loss": 5.224,
+      "epoch": 0.0862533692722372,
+      "grad_norm": 1.3886544704437256,
+      "learning_rate": 0.0005954668105774419,
+      "loss": 5.2671,
       "step": 800
     },
     {
-      "epoch": 0.09148638467333979,
-      "grad_norm": 1.4241634607315063,
-      "learning_rate": 0.0005951513845490787,
-      "loss": 5.2284,
+      "epoch": 0.09164420485175202,
+      "grad_norm": 0.8726224303245544,
+      "learning_rate": 0.0005951430113329735,
+      "loss": 5.2379,
       "step": 850
     },
     {
-      "epoch": 0.09686793671294801,
-      "grad_norm": 1.0950024127960205,
-      "learning_rate": 0.0005948281435190174,
-      "loss": 5.1559,
+      "epoch": 0.09703504043126684,
+      "grad_norm": 1.1303365230560303,
+      "learning_rate": 0.0005948192120885051,
+      "loss": 5.1541,
       "step": 900
     },
     {
-      "epoch": 0.10224948875255624,
-      "grad_norm": 0.8689674735069275,
-      "learning_rate": 0.0005945049024889559,
-      "loss": 5.1198,
+      "epoch": 0.10242587601078167,
+      "grad_norm": 1.1907507181167603,
+      "learning_rate": 0.0005944954128440366,
+      "loss": 5.1459,
       "step": 950
     },
     {
-      "epoch": 0.10763104079216446,
-      "grad_norm": 0.8567312955856323,
-      "learning_rate": 0.0005941816614588944,
-      "loss": 5.0633,
+      "epoch": 0.1078167115902965,
+      "grad_norm": 1.1421568393707275,
+      "learning_rate": 0.0005941716135995682,
+      "loss": 5.0905,
       "step": 1000
     },
     {
-      "epoch": 0.10763104079216446,
-      "eval_accuracy": 0.22941774764519313,
-      "eval_loss": 4.999941825866699,
-      "eval_runtime": 181.8712,
-      "eval_samples_per_second": 99.032,
-      "eval_steps_per_second": 6.191,
+      "epoch": 0.1078167115902965,
+      "eval_accuracy": 0.2273858299901745,
+      "eval_loss": 5.018712520599365,
+      "eval_runtime": 185.0397,
+      "eval_samples_per_second": 97.336,
+      "eval_steps_per_second": 6.085,
       "step": 1000
     },
     {
-      "epoch": 0.11301259283177269,
-      "grad_norm": 1.0926947593688965,
-      "learning_rate": 0.000593858420428833,
-      "loss": 5.0126,
+      "epoch": 0.11320754716981132,
+      "grad_norm": 1.1553248167037964,
+      "learning_rate": 0.0005938478143550997,
+      "loss": 5.0487,
       "step": 1050
     },
     {
-      "epoch": 0.1183941448713809,
-      "grad_norm": 1.079075813293457,
-      "learning_rate": 0.0005935351793987716,
-      "loss": 4.9581,
+      "epoch": 0.11859838274932614,
+      "grad_norm": 1.4468927383422852,
+      "learning_rate": 0.0005935240151106314,
+      "loss": 5.0104,
       "step": 1100
     },
     {
-      "epoch": 0.12377569691098914,
-      "grad_norm": 1.9798409938812256,
-      "learning_rate": 0.0005932119383687103,
-      "loss": 4.9724,
+      "epoch": 0.12398921832884097,
+      "grad_norm": 0.8825078010559082,
+      "learning_rate": 0.0005932002158661629,
+      "loss": 4.9857,
       "step": 1150
     },
     {
-      "epoch": 0.12915724895059735,
-      "grad_norm": 1.3321768045425415,
-      "learning_rate": 0.0005928886973386488,
-      "loss": 4.9741,
+      "epoch": 0.1293800539083558,
+      "grad_norm": 1.3887170553207397,
+      "learning_rate": 0.0005928764166216945,
+      "loss": 4.9682,
       "step": 1200
     },
     {
-      "epoch": 0.13453880099020557,
-      "grad_norm": 1.1671667098999023,
-      "learning_rate": 0.0005925654563085874,
-      "loss": 4.9051,
+      "epoch": 0.1347708894878706,
+      "grad_norm": 1.168694019317627,
+      "learning_rate": 0.000592552617377226,
+      "loss": 4.9421,
       "step": 1250
     },
     {
-      "epoch": 0.1399203530298138,
-      "grad_norm": 0.8295050263404846,
-      "learning_rate": 0.000592242215278526,
-      "loss": 4.9002,
+      "epoch": 0.14016172506738545,
+      "grad_norm": 0.8403277397155762,
+      "learning_rate": 0.0005922288181327577,
+      "loss": 4.8864,
       "step": 1300
     },
     {
-      "epoch": 0.14530190506942203,
-      "grad_norm": 1.0559158325195312,
-      "learning_rate": 0.0005919189742484645,
-      "loss": 4.8626,
+      "epoch": 0.14555256064690028,
+      "grad_norm": 1.0229427814483643,
+      "learning_rate": 0.0005919050188882893,
+      "loss": 4.8666,
       "step": 1350
     },
     {
-      "epoch": 0.15068345710903025,
-      "grad_norm": 1.2337747812271118,
-      "learning_rate": 0.0005915957332184032,
-      "loss": 4.851,
+      "epoch": 0.1509433962264151,
+      "grad_norm": 1.0546599626541138,
+      "learning_rate": 0.0005915812196438207,
+      "loss": 4.8429,
       "step": 1400
     },
     {
-      "epoch": 0.15606500914863847,
-      "grad_norm": 1.0464680194854736,
-      "learning_rate": 0.0005912724921883417,
-      "loss": 4.8242,
+      "epoch": 0.15633423180592992,
+      "grad_norm": 1.2618271112442017,
+      "learning_rate": 0.0005912574203993524,
+      "loss": 4.8244,
       "step": 1450
     },
     {
-      "epoch": 0.16144656118824668,
-      "grad_norm": 0.8018038868904114,
-      "learning_rate": 0.0005909492511582803,
-      "loss": 4.7894,
+      "epoch": 0.16172506738544473,
+      "grad_norm": 1.0748432874679565,
+      "learning_rate": 0.0005909336211548839,
+      "loss": 4.8066,
       "step": 1500
     },
     {
-      "epoch": 0.1668281132278549,
-      "grad_norm": 1.5508265495300293,
-      "learning_rate": 0.0005906260101282189,
-      "loss": 4.74,
+      "epoch": 0.16711590296495957,
+      "grad_norm": 0.9467751979827881,
+      "learning_rate": 0.0005906098219104155,
+      "loss": 4.7638,
       "step": 1550
     },
     {
-      "epoch": 0.17220966526746315,
-      "grad_norm": 0.9371482133865356,
-      "learning_rate": 0.0005903027690981575,
-      "loss": 4.7316,
+      "epoch": 0.1725067385444744,
+      "grad_norm": 0.9856839776039124,
+      "learning_rate": 0.000590286022665947,
+      "loss": 4.7827,
       "step": 1600
     },
     {
-      "epoch": 0.17759121730707136,
-      "grad_norm": 1.0305496454238892,
-      "learning_rate": 0.000589979528068096,
-      "loss": 4.7221,
+      "epoch": 0.1778975741239892,
+      "grad_norm": 0.9949484467506409,
+      "learning_rate": 0.0005899622234214787,
+      "loss": 4.7595,
       "step": 1650
     },
     {
-      "epoch": 0.18297276934667958,
-      "grad_norm": 0.90619957447052,
-      "learning_rate": 0.0005896562870380347,
-      "loss": 4.7063,
+      "epoch": 0.18328840970350405,
+      "grad_norm": 0.956685483455658,
+      "learning_rate": 0.0005896384241770102,
+      "loss": 4.6836,
       "step": 1700
     },
     {
-      "epoch": 0.1883543213862878,
-      "grad_norm": 1.0820300579071045,
-      "learning_rate": 0.0005893330460079732,
-      "loss": 4.668,
+      "epoch": 0.18867924528301888,
+      "grad_norm": 0.861179530620575,
+      "learning_rate": 0.0005893146249325418,
+      "loss": 4.698,
       "step": 1750
     },
     {
-      "epoch": 0.19373587342589602,
-      "grad_norm": 0.857718825340271,
-      "learning_rate": 0.0005890098049779118,
-      "loss": 4.6453,
+      "epoch": 0.1940700808625337,
+      "grad_norm": 0.8717144727706909,
+      "learning_rate": 0.0005889908256880733,
+      "loss": 4.6851,
       "step": 1800
     },
     {
-      "epoch": 0.19911742546550426,
-      "grad_norm": 0.9784050583839417,
-      "learning_rate": 0.0005886865639478504,
-      "loss": 4.6397,
+      "epoch": 0.19946091644204852,
+      "grad_norm": 1.0138026475906372,
+      "learning_rate": 0.0005886670264436049,
+      "loss": 4.6388,
       "step": 1850
     },
     {
-      "epoch": 0.20449897750511248,
-      "grad_norm": 0.9087166786193848,
-      "learning_rate": 0.0005883633229177889,
-      "loss": 4.6246,
+      "epoch": 0.20485175202156333,
+      "grad_norm": 1.0646897554397583,
+      "learning_rate": 0.0005883432271991365,
+      "loss": 4.6339,
       "step": 1900
     },
     {
-      "epoch": 0.2098805295447207,
-      "grad_norm": 0.9737820029258728,
-      "learning_rate": 0.0005880400818877276,
-      "loss": 4.6047,
+      "epoch": 0.21024258760107817,
+      "grad_norm": 0.9919822812080383,
+      "learning_rate": 0.0005880194279546681,
+      "loss": 4.5957,
       "step": 1950
     },
     {
-      "epoch": 0.2152620815843289,
-      "grad_norm": 0.8255292773246765,
-      "learning_rate": 0.0005877168408576662,
-      "loss": 4.5492,
+      "epoch": 0.215633423180593,
+      "grad_norm": 1.2434839010238647,
+      "learning_rate": 0.0005876956287101996,
+      "loss": 4.5775,
       "step": 2000
     },
     {
-      "epoch": 0.2152620815843289,
-      "eval_accuracy": 0.27245189692187455,
-      "eval_loss": 4.491321563720703,
-      "eval_runtime": 181.6962,
-      "eval_samples_per_second": 99.127,
-      "eval_steps_per_second": 6.197,
+      "epoch": 0.215633423180593,
+      "eval_accuracy": 0.27008391588484576,
+      "eval_loss": 4.515827178955078,
+      "eval_runtime": 184.8516,
+      "eval_samples_per_second": 97.435,
+      "eval_steps_per_second": 6.091,
       "step": 2000
     },
     {
-      "epoch": 0.22064363362393713,
-      "grad_norm": 0.7951588034629822,
-      "learning_rate": 0.0005873935998276048,
-      "loss": 4.5555,
+      "epoch": 0.2210242587601078,
+      "grad_norm": 0.9802135229110718,
+      "learning_rate": 0.0005873718294657312,
+      "loss": 4.5897,
       "step": 2050
     },
     {
-      "epoch": 0.22602518566354537,
-      "grad_norm": 0.8975157141685486,
-      "learning_rate": 0.0005870703587975433,
-      "loss": 4.5375,
+      "epoch": 0.22641509433962265,
+      "grad_norm": 1.04331636428833,
+      "learning_rate": 0.0005870480302212628,
+      "loss": 4.5522,
       "step": 2100
     },
     {
-      "epoch": 0.2314067377031536,
-      "grad_norm": 1.0026023387908936,
-      "learning_rate": 0.0005867471177674818,
-      "loss": 4.5323,
+      "epoch": 0.23180592991913745,
+      "grad_norm": 0.9205772280693054,
+      "learning_rate": 0.0005867242309767943,
+      "loss": 4.4983,
       "step": 2150
     },
     {
-      "epoch": 0.2367882897427618,
-      "grad_norm": 0.803642988204956,
-      "learning_rate": 0.0005864238767374205,
-      "loss": 4.4924,
+      "epoch": 0.2371967654986523,
+      "grad_norm": 1.154678225517273,
+      "learning_rate": 0.0005864004317323259,
+      "loss": 4.5365,
       "step": 2200
     },
     {
-      "epoch": 0.24216984178237003,
-      "grad_norm": 1.0086907148361206,
-      "learning_rate": 0.0005861006357073591,
-      "loss": 4.5072,
+      "epoch": 0.24258760107816713,
+      "grad_norm": 0.763646125793457,
+      "learning_rate": 0.0005860766324878575,
+      "loss": 4.4747,
       "step": 2250
     },
     {
-      "epoch": 0.24755139382197827,
-      "grad_norm": 0.8211397528648376,
-      "learning_rate": 0.0005857773946772977,
-      "loss": 4.4681,
+      "epoch": 0.24797843665768193,
+      "grad_norm": 0.7726097702980042,
+      "learning_rate": 0.000585752833243389,
+      "loss": 4.4536,
       "step": 2300
     },
     {
-      "epoch": 0.2529329458615865,
-      "grad_norm": 0.7603225111961365,
-      "learning_rate": 0.0005854541536472362,
-      "loss": 4.4691,
+      "epoch": 0.25336927223719674,
+      "grad_norm": 0.8371621966362,
+      "learning_rate": 0.0005854290339989206,
+      "loss": 4.4495,
       "step": 2350
     },
     {
-      "epoch": 0.2583144979011947,
-      "grad_norm": 0.8533746004104614,
-      "learning_rate": 0.0005851309126171749,
-      "loss": 4.4473,
+      "epoch": 0.2587601078167116,
+      "grad_norm": 0.9792793989181519,
+      "learning_rate": 0.0005851052347544521,
+      "loss": 4.4429,
       "step": 2400
     },
     {
-      "epoch": 0.2636960499408029,
-      "grad_norm": 1.300952434539795,
-      "learning_rate": 0.0005848076715871134,
-      "loss": 4.4076,
+      "epoch": 0.2641509433962264,
+      "grad_norm": 1.2388581037521362,
+      "learning_rate": 0.0005847814355099838,
+      "loss": 4.4476,
       "step": 2450
     },
     {
-      "epoch": 0.26907760198041114,
-      "grad_norm": 0.7179780006408691,
-      "learning_rate": 0.000584484430557052,
-      "loss": 4.4134,
+      "epoch": 0.2695417789757412,
+      "grad_norm": 0.6825560331344604,
+      "learning_rate": 0.0005844576362655154,
+      "loss": 4.4384,
       "step": 2500
     },
     {
-      "epoch": 0.27445915402001936,
-      "grad_norm": 0.8938673734664917,
-      "learning_rate": 0.0005841611895269906,
-      "loss": 4.4109,
+      "epoch": 0.2749326145552561,
+      "grad_norm": 0.7776582837104797,
+      "learning_rate": 0.0005841338370210469,
+      "loss": 4.4017,
       "step": 2550
     },
     {
-      "epoch": 0.2798407060596276,
-      "grad_norm": 0.8200742602348328,
-      "learning_rate": 0.0005838379484969291,
-      "loss": 4.4013,
+      "epoch": 0.2803234501347709,
+      "grad_norm": 1.007873773574829,
+      "learning_rate": 0.0005838100377765785,
+      "loss": 4.3936,
       "step": 2600
     },
     {
-      "epoch": 0.2852222580992358,
-      "grad_norm": 0.8515391945838928,
-      "learning_rate": 0.0005835147074668678,
-      "loss": 4.3763,
+      "epoch": 0.2857142857142857,
+      "grad_norm": 1.2725998163223267,
+      "learning_rate": 0.0005834862385321101,
+      "loss": 4.3894,
       "step": 2650
     },
     {
-      "epoch": 0.29060381013884407,
-      "grad_norm": 0.8918663859367371,
-      "learning_rate": 0.0005831914664368063,
-      "loss": 4.3774,
+      "epoch": 0.29110512129380056,
+      "grad_norm": 0.8046647310256958,
+      "learning_rate": 0.0005831624392876417,
+      "loss": 4.3826,
       "step": 2700
     },
     {
-      "epoch": 0.2959853621784523,
-      "grad_norm": 0.6826805472373962,
-      "learning_rate": 0.0005828682254067449,
-      "loss": 4.3631,
+      "epoch": 0.29649595687331537,
+      "grad_norm": 0.9733596444129944,
+      "learning_rate": 0.0005828386400431731,
+      "loss": 4.3404,
       "step": 2750
     },
     {
-      "epoch": 0.3013669142180605,
-      "grad_norm": 0.7818616628646851,
-      "learning_rate": 0.0005825449843766835,
-      "loss": 4.3313,
+      "epoch": 0.3018867924528302,
+      "grad_norm": 0.6732513308525085,
+      "learning_rate": 0.0005825148407987048,
+      "loss": 4.3807,
       "step": 2800
     },
     {
-      "epoch": 0.3067484662576687,
-      "grad_norm": 0.7908890843391418,
-      "learning_rate": 0.0005822217433466221,
-      "loss": 4.3407,
+      "epoch": 0.30727762803234504,
+      "grad_norm": 0.8311448097229004,
+      "learning_rate": 0.0005821910415542363,
+      "loss": 4.3505,
       "step": 2850
     },
     {
-      "epoch": 0.31213001829727693,
-      "grad_norm": 1.1532859802246094,
-      "learning_rate": 0.0005818985023165607,
-      "loss": 4.3188,
+      "epoch": 0.31266846361185985,
+      "grad_norm": 0.8010222911834717,
+      "learning_rate": 0.0005818672423097679,
+      "loss": 4.3291,
       "step": 2900
     },
     {
-      "epoch": 0.31751157033688515,
-      "grad_norm": 0.7944912910461426,
-      "learning_rate": 0.0005815752612864992,
-      "loss": 4.3062,
+      "epoch": 0.31805929919137466,
+      "grad_norm": 0.7362540364265442,
+      "learning_rate": 0.0005815434430652994,
+      "loss": 4.3354,
       "step": 2950
     },
     {
-      "epoch": 0.32289312237649337,
-      "grad_norm": 0.7332167029380798,
-      "learning_rate": 0.0005812520202564378,
-      "loss": 4.2863,
+      "epoch": 0.32345013477088946,
+      "grad_norm": 0.7173567414283752,
+      "learning_rate": 0.0005812196438208311,
+      "loss": 4.332,
       "step": 3000
     },
     {
-      "epoch": 0.32289312237649337,
-      "eval_accuracy": 0.29881217403454574,
-      "eval_loss": 4.233547687530518,
-      "eval_runtime": 181.7872,
-      "eval_samples_per_second": 99.077,
-      "eval_steps_per_second": 6.194,
+      "epoch": 0.32345013477088946,
+      "eval_accuracy": 0.29800922919359674,
+      "eval_loss": 4.234623908996582,
+      "eval_runtime": 184.9545,
+      "eval_samples_per_second": 97.381,
+      "eval_steps_per_second": 6.088,
       "step": 3000
     },
     {
-      "epoch": 0.3282746744161016,
-      "grad_norm": 0.6582856774330139,
-      "learning_rate": 0.0005809287792263764,
-      "loss": 4.3131,
+      "epoch": 0.3288409703504043,
+      "grad_norm": 0.7485026121139526,
+      "learning_rate": 0.0005808958445763626,
+      "loss": 4.3078,
       "step": 3050
     },
     {
-      "epoch": 0.3336562264557098,
-      "grad_norm": 0.7379726767539978,
-      "learning_rate": 0.0005806055381963151,
-      "loss": 4.2889,
+      "epoch": 0.33423180592991913,
+      "grad_norm": 0.7733100652694702,
+      "learning_rate": 0.0005805720453318942,
+      "loss": 4.3032,
       "step": 3100
     },
     {
-      "epoch": 0.3390377784953181,
-      "grad_norm": 0.7286725640296936,
-      "learning_rate": 0.0005802822971662536,
-      "loss": 4.287,
+      "epoch": 0.33962264150943394,
+      "grad_norm": 0.835577130317688,
+      "learning_rate": 0.0005802482460874257,
+      "loss": 4.2658,
       "step": 3150
     },
     {
-      "epoch": 0.3444193305349263,
-      "grad_norm": 0.7306691408157349,
-      "learning_rate": 0.0005799590561361922,
-      "loss": 4.2596,
+      "epoch": 0.3450134770889488,
+      "grad_norm": 0.8923593163490295,
+      "learning_rate": 0.0005799244468429573,
+      "loss": 4.2757,
       "step": 3200
     },
     {
-      "epoch": 0.3498008825745345,
-      "grad_norm": 0.6126540303230286,
-      "learning_rate": 0.0005796358151061307,
-      "loss": 4.2404,
+      "epoch": 0.3504043126684636,
+      "grad_norm": 0.7083461880683899,
+      "learning_rate": 0.0005796006475984889,
+      "loss": 4.2502,
       "step": 3250
     },
     {
-      "epoch": 0.35518243461414273,
-      "grad_norm": 0.7405567765235901,
-      "learning_rate": 0.0005793125740760694,
-      "loss": 4.253,
+      "epoch": 0.3557951482479784,
+      "grad_norm": 0.8044667840003967,
+      "learning_rate": 0.0005792768483540205,
+      "loss": 4.2714,
       "step": 3300
     },
     {
-      "epoch": 0.36056398665375095,
-      "grad_norm": 0.8209679126739502,
-      "learning_rate": 0.0005789893330460079,
-      "loss": 4.2517,
+      "epoch": 0.3611859838274933,
+      "grad_norm": 0.6673637628555298,
+      "learning_rate": 0.000578953049109552,
+      "loss": 4.2516,
       "step": 3350
     },
     {
-      "epoch": 0.36594553869335916,
-      "grad_norm": 0.7422078847885132,
-      "learning_rate": 0.0005786660920159465,
-      "loss": 4.2362,
+      "epoch": 0.3665768194070081,
+      "grad_norm": 0.8567188382148743,
+      "learning_rate": 0.0005786292498650836,
+      "loss": 4.244,
       "step": 3400
     },
     {
-      "epoch": 0.3713270907329674,
-      "grad_norm": 0.6978297233581543,
-      "learning_rate": 0.0005783428509858851,
-      "loss": 4.2293,
+      "epoch": 0.3719676549865229,
+      "grad_norm": 0.6368674635887146,
+      "learning_rate": 0.0005783054506206152,
+      "loss": 4.2306,
       "step": 3450
     },
     {
-      "epoch": 0.3767086427725756,
-      "grad_norm": 0.7036471962928772,
-      "learning_rate": 0.0005780196099558237,
-      "loss": 4.2389,
+      "epoch": 0.37735849056603776,
+      "grad_norm": 0.7097017765045166,
+      "learning_rate": 0.0005779816513761467,
+      "loss": 4.21,
       "step": 3500
     },
     {
-      "epoch": 0.3820901948121838,
-      "grad_norm": 0.8010238409042358,
-      "learning_rate": 0.0005776963689257623,
-      "loss": 4.2198,
+      "epoch": 0.38274932614555257,
+      "grad_norm": 0.6933664679527283,
+      "learning_rate": 0.0005776578521316782,
+      "loss": 4.2159,
       "step": 3550
     },
     {
-      "epoch": 0.38747174685179203,
-      "grad_norm": 0.6499230265617371,
-      "learning_rate": 0.0005773731278957008,
-      "loss": 4.1941,
+      "epoch": 0.3881401617250674,
+      "grad_norm": 0.7401413917541504,
+      "learning_rate": 0.0005773340528872099,
+      "loss": 4.2171,
       "step": 3600
     },
     {
-      "epoch": 0.3928532988914003,
-      "grad_norm": 0.7018994688987732,
-      "learning_rate": 0.0005770498868656394,
-      "loss": 4.2102,
+      "epoch": 0.3935309973045822,
+      "grad_norm": 0.899066686630249,
+      "learning_rate": 0.0005770102536427414,
+      "loss": 4.2236,
       "step": 3650
     },
     {
-      "epoch": 0.3982348509310085,
-      "grad_norm": 0.6573231816291809,
-      "learning_rate": 0.000576726645835578,
-      "loss": 4.1892,
+      "epoch": 0.39892183288409705,
+      "grad_norm": 0.8368923664093018,
+      "learning_rate": 0.000576686454398273,
+      "loss": 4.1929,
       "step": 3700
     },
     {
-      "epoch": 0.40361640297061674,
-      "grad_norm": 0.6301137804985046,
-      "learning_rate": 0.0005764034048055167,
-      "loss": 4.1782,
+      "epoch": 0.40431266846361186,
+      "grad_norm": 0.7426706552505493,
+      "learning_rate": 0.0005763626551538045,
+      "loss": 4.1921,
       "step": 3750
     },
     {
-      "epoch": 0.40899795501022496,
-      "grad_norm": 0.6420815587043762,
-      "learning_rate": 0.0005760801637754552,
-      "loss": 4.1881,
+      "epoch": 0.40970350404312667,
+      "grad_norm": 0.8189597725868225,
+      "learning_rate": 0.0005760388559093362,
+      "loss": 4.2154,
       "step": 3800
     },
     {
-      "epoch": 0.4143795070498332,
-      "grad_norm": 0.7286609411239624,
-      "learning_rate": 0.0005757569227453937,
-      "loss": 4.1681,
+      "epoch": 0.41509433962264153,
+      "grad_norm": 0.664476752281189,
+      "learning_rate": 0.0005757150566648678,
+      "loss": 4.1881,
       "step": 3850
     },
     {
-      "epoch": 0.4197610590894414,
-      "grad_norm": 0.7999758720397949,
-      "learning_rate": 0.0005754336817153324,
-      "loss": 4.1663,
+      "epoch": 0.42048517520215634,
+      "grad_norm": 0.8473009467124939,
+      "learning_rate": 0.0005753912574203993,
+      "loss": 4.1562,
       "step": 3900
     },
     {
-      "epoch": 0.4251426111290496,
-      "grad_norm": 0.681491494178772,
-      "learning_rate": 0.0005751104406852709,
-      "loss": 4.1526,
+      "epoch": 0.42587601078167114,
+      "grad_norm": 0.7176007628440857,
+      "learning_rate": 0.0005750674581759309,
+      "loss": 4.158,
       "step": 3950
     },
     {
-      "epoch": 0.4305241631686578,
-      "grad_norm": 0.8898637890815735,
-      "learning_rate": 0.0005747871996552096,
-      "loss": 4.1472,
+      "epoch": 0.431266846361186,
+      "grad_norm": 0.7438318729400635,
+      "learning_rate": 0.0005747436589314624,
+      "loss": 4.1601,
       "step": 4000
     },
     {
-      "epoch": 0.4305241631686578,
-      "eval_accuracy": 0.31332352777238437,
-      "eval_loss": 4.088411808013916,
-      "eval_runtime": 182.0558,
-      "eval_samples_per_second": 98.931,
-      "eval_steps_per_second": 6.185,
+      "epoch": 0.431266846361186,
+      "eval_accuracy": 0.31239595806911213,
+      "eval_loss": 4.091654300689697,
+      "eval_runtime": 185.0599,
+      "eval_samples_per_second": 97.325,
+      "eval_steps_per_second": 6.085,
       "step": 4000
     },
     {
-      "epoch": 0.43590571520826604,
-      "grad_norm": 1.0454559326171875,
-      "learning_rate": 0.0005744639586251481,
-      "loss": 4.174,
+      "epoch": 0.4366576819407008,
+      "grad_norm": 0.6847988963127136,
+      "learning_rate": 0.0005744198596869941,
+      "loss": 4.1601,
       "step": 4050
     },
     {
-      "epoch": 0.44128726724787426,
-      "grad_norm": 0.7238677144050598,
-      "learning_rate": 0.0005741407175950867,
-      "loss": 4.1417,
+      "epoch": 0.4420485175202156,
+      "grad_norm": 0.6240386962890625,
+      "learning_rate": 0.0005740960604425255,
+      "loss": 4.1664,
       "step": 4100
     },
     {
-      "epoch": 0.44666881928748253,
-      "grad_norm": 0.6636929512023926,
-      "learning_rate": 0.0005738174765650253,
-      "loss": 4.14,
+      "epoch": 0.4474393530997305,
+      "grad_norm": 0.5988531708717346,
+      "learning_rate": 0.0005737722611980572,
+      "loss": 4.153,
       "step": 4150
     },
     {
-      "epoch": 0.45205037132709075,
-      "grad_norm": 0.553901195526123,
-      "learning_rate": 0.0005734942355349638,
-      "loss": 4.1244,
+      "epoch": 0.4528301886792453,
+      "grad_norm": 0.6246315240859985,
+      "learning_rate": 0.0005734484619535887,
+      "loss": 4.1408,
       "step": 4200
     },
     {
-      "epoch": 0.45743192336669897,
-      "grad_norm": 0.7060267925262451,
-      "learning_rate": 0.0005731709945049025,
-      "loss": 4.1363,
+      "epoch": 0.4582210242587601,
+      "grad_norm": 0.6362719535827637,
+      "learning_rate": 0.0005731246627091203,
+      "loss": 4.1336,
       "step": 4250
     },
     {
-      "epoch": 0.4628134754063072,
-      "grad_norm": 0.6566117405891418,
-      "learning_rate": 0.000572847753474841,
-      "loss": 4.1328,
+      "epoch": 0.4636118598382749,
+      "grad_norm": 0.7302852272987366,
+      "learning_rate": 0.0005728008634646518,
+      "loss": 4.1261,
       "step": 4300
     },
     {
-      "epoch": 0.4681950274459154,
-      "grad_norm": 0.663583517074585,
-      "learning_rate": 0.0005725245124447796,
-      "loss": 4.1281,
+      "epoch": 0.46900269541778977,
+      "grad_norm": 0.7277068495750427,
+      "learning_rate": 0.0005724770642201835,
+      "loss": 4.1332,
       "step": 4350
     },
     {
-      "epoch": 0.4735765794855236,
-      "grad_norm": 0.797214925289154,
-      "learning_rate": 0.0005722012714147182,
-      "loss": 4.1199,
+      "epoch": 0.4743935309973046,
+      "grad_norm": 0.8166965246200562,
+      "learning_rate": 0.000572153264975715,
+      "loss": 4.1182,
       "step": 4400
     },
     {
-      "epoch": 0.47895813152513184,
-      "grad_norm": 0.6216398477554321,
-      "learning_rate": 0.0005718780303846568,
-      "loss": 4.1097,
+      "epoch": 0.4797843665768194,
+      "grad_norm": 0.772174596786499,
+      "learning_rate": 0.0005718294657312466,
+      "loss": 4.128,
       "step": 4450
     },
     {
-      "epoch": 0.48433968356474005,
-      "grad_norm": 0.7291539907455444,
-      "learning_rate": 0.0005715547893545953,
-      "loss": 4.0946,
+      "epoch": 0.48517520215633425,
+      "grad_norm": 0.6778306365013123,
+      "learning_rate": 0.0005715056664867781,
+      "loss": 4.1057,
       "step": 4500
     },
     {
-      "epoch": 0.48972123560434827,
-      "grad_norm": 0.6444728374481201,
-      "learning_rate": 0.000571231548324534,
-      "loss": 4.0871,
+      "epoch": 0.49056603773584906,
+      "grad_norm": 0.5974618792533875,
+      "learning_rate": 0.0005711818672423097,
+      "loss": 4.0954,
       "step": 4550
     },
     {
-      "epoch": 0.49510278764395654,
-      "grad_norm": 0.6845434308052063,
-      "learning_rate": 0.0005709083072944725,
-      "loss": 4.0885,
+      "epoch": 0.49595687331536387,
+      "grad_norm": 0.7339206337928772,
+      "learning_rate": 0.0005708580679978413,
+      "loss": 4.1006,
       "step": 4600
     },
     {
-      "epoch": 0.5004843396835648,
-      "grad_norm": 0.7555699944496155,
-      "learning_rate": 0.0005705850662644111,
-      "loss": 4.0955,
+      "epoch": 0.5013477088948787,
+      "grad_norm": 0.732438862323761,
+      "learning_rate": 0.0005705342687533729,
+      "loss": 4.0836,
       "step": 4650
     },
     {
-      "epoch": 0.505865891723173,
-      "grad_norm": 0.7021520733833313,
-      "learning_rate": 0.0005702618252343497,
-      "loss": 4.0961,
+      "epoch": 0.5067385444743935,
+      "grad_norm": 0.7650737166404724,
+      "learning_rate": 0.0005702104695089044,
+      "loss": 4.0887,
       "step": 4700
     },
     {
-      "epoch": 0.5112474437627812,
-      "grad_norm": 0.5818867087364197,
-      "learning_rate": 0.0005699385842042882,
-      "loss": 4.0799,
+      "epoch": 0.5121293800539084,
+      "grad_norm": 0.5491770505905151,
+      "learning_rate": 0.000569886670264436,
+      "loss": 4.0701,
       "step": 4750
     },
     {
-      "epoch": 0.5166289958023894,
-      "grad_norm": 0.665334939956665,
-      "learning_rate": 0.0005696153431742269,
-      "loss": 4.0524,
+      "epoch": 0.5175202156334232,
+      "grad_norm": 0.590347170829773,
+      "learning_rate": 0.0005695628710199675,
+      "loss": 4.0851,
       "step": 4800
     },
     {
-      "epoch": 0.5220105478419976,
-      "grad_norm": 0.6017621755599976,
-      "learning_rate": 0.0005692921021441655,
-      "loss": 4.0778,
+      "epoch": 0.522911051212938,
+      "grad_norm": 0.6122878789901733,
+      "learning_rate": 0.0005692390717754991,
+      "loss": 4.0779,
       "step": 4850
     },
     {
-      "epoch": 0.5273920998816058,
-      "grad_norm": 0.6869152188301086,
-      "learning_rate": 0.0005689688611141041,
-      "loss": 4.0557,
+      "epoch": 0.5283018867924528,
+      "grad_norm": 0.6329747438430786,
+      "learning_rate": 0.0005689152725310306,
+      "loss": 4.0857,
       "step": 4900
     },
     {
-      "epoch": 0.5327736519212141,
-      "grad_norm": 0.7289915680885315,
-      "learning_rate": 0.0005686456200840426,
-      "loss": 4.0789,
+      "epoch": 0.5336927223719676,
+      "grad_norm": 0.7208688259124756,
+      "learning_rate": 0.0005685914732865623,
+      "loss": 4.0632,
       "step": 4950
     },
     {
-      "epoch": 0.5381552039608223,
-      "grad_norm": 0.6597164869308472,
-      "learning_rate": 0.0005683223790539811,
-      "loss": 4.062,
+      "epoch": 0.5390835579514824,
+      "grad_norm": 0.5675130486488342,
+      "learning_rate": 0.0005682676740420939,
+      "loss": 4.0517,
       "step": 5000
     },
     {
-      "epoch": 0.5381552039608223,
-      "eval_accuracy": 0.32235986249325127,
-      "eval_loss": 3.9826269149780273,
-      "eval_runtime": 181.7213,
-      "eval_samples_per_second": 99.113,
-      "eval_steps_per_second": 6.196,
+      "epoch": 0.5390835579514824,
+      "eval_accuracy": 0.32126485868985694,
+      "eval_loss": 3.9956471920013428,
+      "eval_runtime": 184.8848,
+      "eval_samples_per_second": 97.417,
+      "eval_steps_per_second": 6.09,
       "step": 5000
     },
     {
-      "epoch": 0.5435367560004305,
-      "grad_norm": 0.6261274218559265,
-      "learning_rate": 0.0005679991380239198,
-      "loss": 4.0553,
+      "epoch": 0.5444743935309974,
+      "grad_norm": 0.646916925907135,
+      "learning_rate": 0.0005679438747976254,
+      "loss": 4.0447,
       "step": 5050
     },
     {
-      "epoch": 0.5489183080400387,
-      "grad_norm": 0.649811863899231,
-      "learning_rate": 0.0005676823618144596,
-      "loss": 4.0688,
+      "epoch": 0.5498652291105122,
+      "grad_norm": 0.6901677250862122,
+      "learning_rate": 0.000567620075553157,
+      "loss": 4.0497,
       "step": 5100
     },
     {
-      "epoch": 0.5542998600796469,
-      "grad_norm": 0.6599764823913574,
-      "learning_rate": 0.0005673591207843981,
-      "loss": 4.0458,
+      "epoch": 0.555256064690027,
+      "grad_norm": 0.6170982718467712,
+      "learning_rate": 0.0005672962763086886,
+      "loss": 4.0538,
       "step": 5150
     },
     {
-      "epoch": 0.5596814121192552,
-      "grad_norm": 0.5525273680686951,
-      "learning_rate": 0.0005670358797543368,
-      "loss": 4.032,
+      "epoch": 0.5606469002695418,
+      "grad_norm": 0.6177210807800293,
+      "learning_rate": 0.0005669724770642202,
+      "loss": 4.0316,
       "step": 5200
     },
     {
-      "epoch": 0.5650629641588634,
-      "grad_norm": 0.5733092427253723,
-      "learning_rate": 0.0005667126387242753,
-      "loss": 4.0628,
+      "epoch": 0.5660377358490566,
+      "grad_norm": 0.6208630204200745,
+      "learning_rate": 0.0005666486778197517,
+      "loss": 4.0317,
       "step": 5250
     },
     {
-      "epoch": 0.5704445161984716,
-      "grad_norm": 0.6523923277854919,
-      "learning_rate": 0.000566389397694214,
-      "loss": 4.0207,
+      "epoch": 0.5714285714285714,
+      "grad_norm": 0.6907792091369629,
+      "learning_rate": 0.0005663248785752833,
+      "loss": 4.0154,
       "step": 5300
     },
     {
-      "epoch": 0.5758260682380799,
-      "grad_norm": 0.5884820222854614,
-      "learning_rate": 0.0005660661566641525,
-      "loss": 4.0289,
+      "epoch": 0.5768194070080862,
+      "grad_norm": 0.5993092656135559,
+      "learning_rate": 0.0005660010793308148,
+      "loss": 4.0339,
       "step": 5350
     },
     {
-      "epoch": 0.5812076202776881,
-      "grad_norm": 0.6992023587226868,
-      "learning_rate": 0.0005657429156340911,
-      "loss": 4.0149,
+      "epoch": 0.5822102425876011,
+      "grad_norm": 0.7210185527801514,
+      "learning_rate": 0.0005656772800863465,
+      "loss": 4.024,
       "step": 5400
     },
     {
-      "epoch": 0.5865891723172963,
-      "grad_norm": 0.6001935601234436,
-      "learning_rate": 0.0005654196746040297,
-      "loss": 4.0226,
+      "epoch": 0.5876010781671159,
+      "grad_norm": 0.6388777494430542,
+      "learning_rate": 0.0005653534808418779,
+      "loss": 4.0306,
       "step": 5450
     },
     {
-      "epoch": 0.5919707243569046,
-      "grad_norm": 0.7079454660415649,
-      "learning_rate": 0.0005650964335739684,
-      "loss": 4.0292,
+      "epoch": 0.5929919137466307,
+      "grad_norm": 0.5874868035316467,
+      "learning_rate": 0.0005650296815974096,
+      "loss": 4.0322,
       "step": 5500
     },
     {
-      "epoch": 0.5973522763965128,
-      "grad_norm": 0.6667420864105225,
-      "learning_rate": 0.0005647731925439069,
-      "loss": 3.9974,
+      "epoch": 0.5983827493261455,
+      "grad_norm": 0.6064094305038452,
+      "learning_rate": 0.0005647058823529411,
+      "loss": 4.0365,
       "step": 5550
     },
     {
-      "epoch": 0.602733828436121,
-      "grad_norm": 0.589589536190033,
-      "learning_rate": 0.0005644499515138454,
-      "loss": 4.0047,
+      "epoch": 0.6037735849056604,
+      "grad_norm": 0.6970310211181641,
+      "learning_rate": 0.0005643820831084727,
+      "loss": 4.0156,
       "step": 5600
     },
     {
-      "epoch": 0.6081153804757292,
-      "grad_norm": 0.6412240266799927,
-      "learning_rate": 0.000564126710483784,
-      "loss": 3.9956,
+      "epoch": 0.6091644204851752,
+      "grad_norm": 0.5707041025161743,
+      "learning_rate": 0.0005640582838640042,
+      "loss": 4.0067,
       "step": 5650
     },
     {
-      "epoch": 0.6134969325153374,
-      "grad_norm": 0.6452463269233704,
-      "learning_rate": 0.0005638034694537226,
-      "loss": 3.9969,
+      "epoch": 0.6145552560646901,
+      "grad_norm": 0.6243141293525696,
+      "learning_rate": 0.0005637344846195358,
+      "loss": 4.0111,
       "step": 5700
     },
     {
-      "epoch": 0.6188784845549457,
-      "grad_norm": 0.634573221206665,
-      "learning_rate": 0.0005634802284236612,
-      "loss": 3.9971,
+      "epoch": 0.6199460916442049,
+      "grad_norm": 0.6075974106788635,
+      "learning_rate": 0.0005634106853750674,
+      "loss": 4.0213,
       "step": 5750
     },
     {
-      "epoch": 0.6242600365945539,
-      "grad_norm": 0.6738696694374084,
-      "learning_rate": 0.0005631569873935998,
-      "loss": 3.9907,
+      "epoch": 0.6253369272237197,
+      "grad_norm": 0.5607590675354004,
+      "learning_rate": 0.000563086886130599,
+      "loss": 4.0105,
       "step": 5800
     },
     {
-      "epoch": 0.6296415886341621,
-      "grad_norm": 0.6989587545394897,
-      "learning_rate": 0.0005628337463635384,
-      "loss": 4.0001,
+      "epoch": 0.6307277628032345,
+      "grad_norm": 0.6650785803794861,
+      "learning_rate": 0.0005627630868861305,
+      "loss": 4.023,
       "step": 5850
     },
     {
-      "epoch": 0.6350231406737703,
-      "grad_norm": 0.7382339239120483,
-      "learning_rate": 0.0005625105053334769,
-      "loss": 3.9866,
+      "epoch": 0.6361185983827493,
+      "grad_norm": 0.6571787595748901,
+      "learning_rate": 0.0005624392876416621,
+      "loss": 3.9933,
       "step": 5900
     },
     {
-      "epoch": 0.6404046927133785,
-      "grad_norm": 0.6265085339546204,
-      "learning_rate": 0.0005621872643034155,
-      "loss": 3.965,
+      "epoch": 0.6415094339622641,
+      "grad_norm": 0.637128472328186,
+      "learning_rate": 0.0005621154883971937,
+      "loss": 3.9854,
       "step": 5950
     },
     {
-      "epoch": 0.6457862447529867,
-      "grad_norm": 0.675687313079834,
-      "learning_rate": 0.0005618640232733541,
-      "loss": 3.9919,
+      "epoch": 0.6469002695417789,
+      "grad_norm": 0.6387634873390198,
+      "learning_rate": 0.0005617916891527253,
+      "loss": 3.9951,
       "step": 6000
     },
     {
-      "epoch": 0.6457862447529867,
-      "eval_accuracy": 0.3284569192929609,
-      "eval_loss": 3.910811424255371,
-      "eval_runtime": 181.5773,
-      "eval_samples_per_second": 99.192,
-      "eval_steps_per_second": 6.201,
+      "epoch": 0.6469002695417789,
+      "eval_accuracy": 0.32733703397825703,
+      "eval_loss": 3.921706199645996,
+      "eval_runtime": 184.8387,
+      "eval_samples_per_second": 97.442,
+      "eval_steps_per_second": 6.092,
       "step": 6000
     },
     {
-      "epoch": 0.651167796792595,
-      "grad_norm": 0.6309311389923096,
-      "learning_rate": 0.0005615407822432927,
-      "loss": 3.9811,
+      "epoch": 0.6522911051212938,
+      "grad_norm": 0.653874933719635,
+      "learning_rate": 0.0005614678899082568,
+      "loss": 3.9682,
       "step": 6050
     },
     {
-      "epoch": 0.6565493488322032,
-      "grad_norm": 0.5584713220596313,
-      "learning_rate": 0.0005612175412132313,
-      "loss": 3.9619,
+      "epoch": 0.6576819407008087,
+      "grad_norm": 0.7536391615867615,
+      "learning_rate": 0.0005611440906637884,
+      "loss": 4.0014,
       "step": 6100
     },
     {
-      "epoch": 0.6619309008718114,
-      "grad_norm": 0.5973617434501648,
-      "learning_rate": 0.0005608943001831699,
-      "loss": 3.9528,
+      "epoch": 0.6630727762803235,
+      "grad_norm": 0.6264755129814148,
+      "learning_rate": 0.00056082029141932,
+      "loss": 3.9809,
       "step": 6150
     },
     {
-      "epoch": 0.6673124529114196,
-      "grad_norm": 0.6784116625785828,
-      "learning_rate": 0.0005605710591531085,
-      "loss": 3.9584,
+      "epoch": 0.6684636118598383,
+      "grad_norm": 0.6374775767326355,
+      "learning_rate": 0.0005604964921748515,
+      "loss": 3.9595,
       "step": 6200
     },
     {
-      "epoch": 0.6726940049510278,
-      "grad_norm": 0.5735004544258118,
-      "learning_rate": 0.000560247818123047,
-      "loss": 3.9522,
+      "epoch": 0.6738544474393531,
+      "grad_norm": 0.6206533312797546,
+      "learning_rate": 0.000560172692930383,
+      "loss": 3.9782,
       "step": 6250
     },
     {
-      "epoch": 0.6780755569906362,
-      "grad_norm": 0.6804719567298889,
-      "learning_rate": 0.0005599245770929855,
-      "loss": 3.9563,
+      "epoch": 0.6792452830188679,
+      "grad_norm": 0.612671434879303,
+      "learning_rate": 0.0005598488936859147,
+      "loss": 3.9957,
       "step": 6300
     },
     {
-      "epoch": 0.6834571090302444,
-      "grad_norm": 0.6392138004302979,
-      "learning_rate": 0.0005596013360629242,
-      "loss": 3.9444,
+      "epoch": 0.6846361185983828,
+      "grad_norm": 0.6080566048622131,
+      "learning_rate": 0.0005595250944414463,
+      "loss": 3.9573,
       "step": 6350
     },
     {
-      "epoch": 0.6888386610698526,
-      "grad_norm": 0.5223947763442993,
-      "learning_rate": 0.0005592780950328628,
-      "loss": 3.9685,
+      "epoch": 0.6900269541778976,
+      "grad_norm": Infinity,
+      "learning_rate": 0.0005592077711818672,
+      "loss": 3.9618,
       "step": 6400
     },
     {
-      "epoch": 0.6942202131094608,
-      "grad_norm": 0.6110231280326843,
-      "learning_rate": 0.0005589548540028014,
-      "loss": 3.9501,
+      "epoch": 0.6954177897574124,
+      "grad_norm": 0.6666589975357056,
+      "learning_rate": 0.0005588839719373988,
+      "loss": 3.9455,
       "step": 6450
     },
     {
-      "epoch": 0.699601765149069,
-      "grad_norm": 0.6495262980461121,
-      "learning_rate": 0.0005586316129727399,
-      "loss": 3.9733,
+      "epoch": 0.7008086253369272,
+      "grad_norm": 0.6202605962753296,
+      "learning_rate": 0.0005585601726929303,
+      "loss": 3.9596,
       "step": 6500
     },
     {
-      "epoch": 0.7049833171886772,
-      "grad_norm": 0.5392972230911255,
-      "learning_rate": 0.0005583083719426786,
-      "loss": 3.9475,
+      "epoch": 0.706199460916442,
+      "grad_norm": 0.570480465888977,
+      "learning_rate": 0.0005582363734484619,
+      "loss": 3.9847,
       "step": 6550
     },
     {
-      "epoch": 0.7103648692282855,
-      "grad_norm": 0.5373910665512085,
-      "learning_rate": 0.0005579851309126171,
-      "loss": 3.9475,
+      "epoch": 0.7115902964959568,
+      "grad_norm": 0.579180121421814,
+      "learning_rate": 0.0005579125742039935,
+      "loss": 3.9406,
       "step": 6600
     },
     {
-      "epoch": 0.7157464212678937,
-      "grad_norm": 0.5609496235847473,
-      "learning_rate": 0.0005576618898825558,
-      "loss": 3.9292,
+      "epoch": 0.7169811320754716,
+      "grad_norm": 0.5900434255599976,
+      "learning_rate": 0.0005575887749595251,
+      "loss": 3.9683,
       "step": 6650
     },
     {
-      "epoch": 0.7211279733075019,
-      "grad_norm": 0.5721986293792725,
-      "learning_rate": 0.0005573386488524943,
-      "loss": 3.9432,
+      "epoch": 0.7223719676549866,
+      "grad_norm": 0.6138500571250916,
+      "learning_rate": 0.0005572649757150566,
+      "loss": 3.9438,
       "step": 6700
     },
     {
-      "epoch": 0.7265095253471101,
-      "grad_norm": 0.5666886568069458,
-      "learning_rate": 0.0005570154078224328,
-      "loss": 3.9348,
+      "epoch": 0.7277628032345014,
+      "grad_norm": 0.5448847413063049,
+      "learning_rate": 0.0005569411764705882,
+      "loss": 3.9507,
       "step": 6750
     },
     {
-      "epoch": 0.7318910773867183,
-      "grad_norm": 0.5143133997917175,
-      "learning_rate": 0.0005566921667923715,
-      "loss": 3.9322,
+      "epoch": 0.7331536388140162,
+      "grad_norm": 0.6583116054534912,
+      "learning_rate": 0.0005566173772261198,
+      "loss": 3.9383,
       "step": 6800
     },
     {
-      "epoch": 0.7372726294263265,
-      "grad_norm": 0.5764801502227783,
-      "learning_rate": 0.00055636892576231,
-      "loss": 3.9196,
+      "epoch": 0.738544474393531,
+      "grad_norm": 0.6227813959121704,
+      "learning_rate": 0.0005562935779816513,
+      "loss": 3.9187,
       "step": 6850
     },
     {
-      "epoch": 0.7426541814659348,
-      "grad_norm": 0.6438068151473999,
-      "learning_rate": 0.0005560456847322487,
-      "loss": 3.9207,
+      "epoch": 0.7439353099730458,
+      "grad_norm": 0.6529786586761475,
+      "learning_rate": 0.0005559697787371828,
+      "loss": 3.9458,
       "step": 6900
     },
     {
-      "epoch": 0.748035733505543,
-      "grad_norm": 0.547141432762146,
-      "learning_rate": 0.0005557224437021872,
-      "loss": 3.9211,
+      "epoch": 0.7493261455525606,
+      "grad_norm": 0.6059437990188599,
+      "learning_rate": 0.0005556459794927145,
+      "loss": 3.9425,
       "step": 6950
     },
     {
-      "epoch": 0.7534172855451512,
-      "grad_norm": 0.5400163531303406,
-      "learning_rate": 0.0005553992026721258,
-      "loss": 3.912,
+      "epoch": 0.7547169811320755,
+      "grad_norm": 0.6971544623374939,
+      "learning_rate": 0.000555322180248246,
+      "loss": 3.9281,
       "step": 7000
     },
     {
-      "epoch": 0.7534172855451512,
-      "eval_accuracy": 0.3342374702304669,
-      "eval_loss": 3.858619451522827,
-      "eval_runtime": 181.854,
-      "eval_samples_per_second": 99.041,
-      "eval_steps_per_second": 6.192,
+      "epoch": 0.7547169811320755,
+      "eval_accuracy": 0.3333070755521115,
+      "eval_loss": 3.862715482711792,
+      "eval_runtime": 184.7265,
+      "eval_samples_per_second": 97.501,
+      "eval_steps_per_second": 6.095,
       "step": 7000
     },
     {
-      "epoch": 0.7587988375847594,
-      "grad_norm": 0.6468452215194702,
-      "learning_rate": 0.0005550759616420644,
-      "loss": 3.9213,
+      "epoch": 0.7601078167115903,
+      "grad_norm": 0.5811669826507568,
+      "learning_rate": 0.0005549983810037776,
+      "loss": 3.9215,
       "step": 7050
     },
     {
-      "epoch": 0.7641803896243676,
-      "grad_norm": 0.8306963443756104,
-      "learning_rate": 0.000554752720612003,
-      "loss": 3.9261,
+      "epoch": 0.7654986522911051,
+      "grad_norm": 0.5781192183494568,
+      "learning_rate": 0.0005546745817593091,
+      "loss": 3.9333,
       "step": 7100
     },
     {
-      "epoch": 0.7695619416639758,
-      "grad_norm": 0.5487331748008728,
-      "learning_rate": 0.0005544294795819415,
-      "loss": 3.9015,
+      "epoch": 0.77088948787062,
+      "grad_norm": 0.5684405565261841,
+      "learning_rate": 0.0005543507825148408,
+      "loss": 3.9335,
       "step": 7150
     },
     {
-      "epoch": 0.7749434937035841,
-      "grad_norm": 0.5448126196861267,
-      "learning_rate": 0.0005541062385518801,
-      "loss": 3.9063,
+      "epoch": 0.7762803234501348,
+      "grad_norm": 0.6275970339775085,
+      "learning_rate": 0.0005540269832703723,
+      "loss": 3.9334,
       "step": 7200
     },
     {
-      "epoch": 0.7803250457431924,
-      "grad_norm": 0.5569677948951721,
-      "learning_rate": 0.0005537829975218188,
-      "loss": 3.9022,
+      "epoch": 0.7816711590296496,
+      "grad_norm": 0.5931147933006287,
+      "learning_rate": 0.0005537031840259039,
+      "loss": 3.8893,
       "step": 7250
     },
     {
-      "epoch": 0.7857065977828006,
-      "grad_norm": 0.5910586714744568,
-      "learning_rate": 0.0005534597564917573,
-      "loss": 3.9111,
+      "epoch": 0.7870619946091644,
+      "grad_norm": 0.6141475439071655,
+      "learning_rate": 0.0005533793847814354,
+      "loss": 3.9264,
       "step": 7300
     },
     {
-      "epoch": 0.7910881498224088,
-      "grad_norm": 0.75867760181427,
-      "learning_rate": 0.0005531365154616959,
-      "loss": 3.9149,
+      "epoch": 0.7924528301886793,
+      "grad_norm": 0.6149185299873352,
+      "learning_rate": 0.000553055585536967,
+      "loss": 3.9141,
       "step": 7350
     },
     {
-      "epoch": 0.796469701862017,
-      "grad_norm": 0.5806201696395874,
-      "learning_rate": 0.0005528132744316344,
-      "loss": 3.8975,
+      "epoch": 0.7978436657681941,
+      "grad_norm": 0.608295202255249,
+      "learning_rate": 0.0005527317862924987,
+      "loss": 3.9136,
       "step": 7400
     },
     {
-      "epoch": 0.8018512539016253,
-      "grad_norm": 0.5979334115982056,
-      "learning_rate": 0.0005524900334015731,
-      "loss": 3.9194,
+      "epoch": 0.8032345013477089,
+      "grad_norm": 0.5281786918640137,
+      "learning_rate": 0.0005524079870480301,
+      "loss": 3.9039,
       "step": 7450
     },
     {
-      "epoch": 0.8072328059412335,
-      "grad_norm": 0.6362331509590149,
-      "learning_rate": 0.0005521667923715117,
-      "loss": 3.8981,
+      "epoch": 0.8086253369272237,
+      "grad_norm": 0.6112433671951294,
+      "learning_rate": 0.0005520841878035618,
+      "loss": 3.8906,
       "step": 7500
     },
     {
-      "epoch": 0.8126143579808417,
-      "grad_norm": 0.5838714838027954,
-      "learning_rate": 0.0005518435513414502,
-      "loss": 3.8961,
+      "epoch": 0.8140161725067385,
+      "grad_norm": 0.5515127182006836,
+      "learning_rate": 0.0005517603885590933,
+      "loss": 3.9185,
       "step": 7550
     },
     {
-      "epoch": 0.8179959100204499,
-      "grad_norm": 0.7832590937614441,
-      "learning_rate": 0.0005515203103113888,
-      "loss": 3.8931,
+      "epoch": 0.8194070080862533,
+      "grad_norm": 0.5626883506774902,
+      "learning_rate": 0.0005514365893146249,
+      "loss": 3.9207,
       "step": 7600
     },
     {
-      "epoch": 0.8233774620600581,
-      "grad_norm": 0.5615019798278809,
-      "learning_rate": 0.0005511970692813274,
-      "loss": 3.8981,
+      "epoch": 0.8247978436657682,
+      "grad_norm": 0.5652742981910706,
+      "learning_rate": 0.0005511127900701564,
+      "loss": 3.8936,
       "step": 7650
     },
     {
-      "epoch": 0.8287590140996663,
-      "grad_norm": 0.6099148392677307,
-      "learning_rate": 0.000550873828251266,
-      "loss": 3.8941,
+      "epoch": 0.8301886792452831,
+      "grad_norm": 0.5642729997634888,
+      "learning_rate": 0.000550788990825688,
+      "loss": 3.8981,
       "step": 7700
     },
     {
-      "epoch": 0.8341405661392746,
-      "grad_norm": 0.5140008330345154,
-      "learning_rate": 0.0005505505872212045,
-      "loss": 3.8796,
+      "epoch": 0.8355795148247979,
+      "grad_norm": 0.5795193910598755,
+      "learning_rate": 0.0005504651915812196,
+      "loss": 3.8818,
       "step": 7750
     },
     {
-      "epoch": 0.8395221181788828,
-      "grad_norm": 0.6559896469116211,
-      "learning_rate": 0.0005502273461911432,
-      "loss": 3.8723,
+      "epoch": 0.8409703504043127,
+      "grad_norm": 0.609893798828125,
+      "learning_rate": 0.0005501413923367512,
+      "loss": 3.9107,
       "step": 7800
     },
     {
-      "epoch": 0.844903670218491,
-      "grad_norm": 0.557579517364502,
-      "learning_rate": 0.0005499041051610817,
-      "loss": 3.8736,
+      "epoch": 0.8463611859838275,
+      "grad_norm": 0.6346420049667358,
+      "learning_rate": 0.0005498175930922827,
+      "loss": 3.8884,
       "step": 7850
     },
     {
-      "epoch": 0.8502852222580992,
-      "grad_norm": 0.6086620688438416,
-      "learning_rate": 0.0005495808641310204,
-      "loss": 3.9014,
+      "epoch": 0.8517520215633423,
+      "grad_norm": 0.5454068779945374,
+      "learning_rate": 0.0005494937938478143,
+      "loss": 3.8879,
       "step": 7900
     },
     {
-      "epoch": 0.8556667742977074,
-      "grad_norm": 0.6662868857383728,
-      "learning_rate": 0.0005492576231009589,
-      "loss": 3.8747,
+      "epoch": 0.8571428571428571,
+      "grad_norm": 0.5878623723983765,
+      "learning_rate": 0.0005491699946033459,
+      "loss": 3.8863,
       "step": 7950
     },
     {
-      "epoch": 0.8610483263373157,
-      "grad_norm": 1.007802963256836,
-      "learning_rate": 0.0005489343820708974,
-      "loss": 3.862,
+      "epoch": 0.862533692722372,
+      "grad_norm": 0.6634072661399841,
+      "learning_rate": 0.0005488461953588775,
+      "loss": 3.8745,
       "step": 8000
     },
     {
-      "epoch": 0.8610483263373157,
-      "eval_accuracy": 0.33844711771595115,
-      "eval_loss": 3.811176061630249,
-      "eval_runtime": 181.8668,
-      "eval_samples_per_second": 99.034,
-      "eval_steps_per_second": 6.191,
+      "epoch": 0.862533692722372,
+      "eval_accuracy": 0.33734624665661483,
+      "eval_loss": 3.8194918632507324,
+      "eval_runtime": 184.5541,
+      "eval_samples_per_second": 97.592,
+      "eval_steps_per_second": 6.101,
       "step": 8000
     },
     {
-      "epoch": 0.8664298783769239,
-      "grad_norm": 0.5262889266014099,
-      "learning_rate": 0.0005486111410408361,
-      "loss": 3.8806,
+      "epoch": 0.8679245283018868,
+      "grad_norm": 0.5450937747955322,
+      "learning_rate": 0.000548522396114409,
+      "loss": 3.8827,
       "step": 8050
     },
     {
-      "epoch": 0.8718114304165321,
-      "grad_norm": 0.5448309183120728,
-      "learning_rate": 0.0005482879000107746,
-      "loss": 3.8702,
+      "epoch": 0.8733153638814016,
+      "grad_norm": 0.6128168702125549,
+      "learning_rate": 0.0005481985968699406,
+      "loss": 3.8677,
       "step": 8100
     },
     {
-      "epoch": 0.8771929824561403,
-      "grad_norm": 0.5762999057769775,
-      "learning_rate": 0.0005479646589807133,
-      "loss": 3.8735,
+      "epoch": 0.8787061994609164,
+      "grad_norm": 0.5559018850326538,
+      "learning_rate": 0.0005478747976254721,
+      "loss": 3.8809,
       "step": 8150
     },
     {
-      "epoch": 0.8825745344957485,
-      "grad_norm": 1.141692876815796,
-      "learning_rate": 0.0005476414179506518,
-      "loss": 3.8806,
+      "epoch": 0.8840970350404312,
+      "grad_norm": 0.5544788241386414,
+      "learning_rate": 0.0005475509983810037,
+      "loss": 3.8704,
       "step": 8200
     },
     {
-      "epoch": 0.8879560865353568,
-      "grad_norm": 0.5628105998039246,
-      "learning_rate": 0.0005473181769205904,
-      "loss": 3.8622,
+      "epoch": 0.889487870619946,
+      "grad_norm": 0.582495391368866,
+      "learning_rate": 0.0005472271991365352,
+      "loss": 3.8559,
       "step": 8250
     },
     {
-      "epoch": 0.8933376385749651,
-      "grad_norm": 0.5854573249816895,
-      "learning_rate": 0.000546994935890529,
-      "loss": 3.8763,
+      "epoch": 0.894878706199461,
+      "grad_norm": 0.5992187857627869,
+      "learning_rate": 0.0005469033998920669,
+      "loss": 3.8817,
       "step": 8300
     },
     {
-      "epoch": 0.8987191906145733,
-      "grad_norm": 0.5357927083969116,
-      "learning_rate": 0.0005466716948604677,
-      "loss": 3.8505,
+      "epoch": 0.9002695417789758,
+      "grad_norm": 0.6100888252258301,
+      "learning_rate": 0.0005465796006475984,
+      "loss": 3.8633,
       "step": 8350
     },
     {
-      "epoch": 0.9041007426541815,
-      "grad_norm": 0.6378769874572754,
-      "learning_rate": 0.0005463484538304062,
-      "loss": 3.845,
+      "epoch": 0.9056603773584906,
+      "grad_norm": 0.5837623476982117,
+      "learning_rate": 0.00054625580140313,
+      "loss": 3.8758,
       "step": 8400
     },
     {
-      "epoch": 0.9094822946937897,
-      "grad_norm": 0.6226728558540344,
-      "learning_rate": 0.0005460252128003447,
-      "loss": 3.8583,
+      "epoch": 0.9110512129380054,
+      "grad_norm": 0.6226052641868591,
+      "learning_rate": 0.0005459320021586615,
+      "loss": 3.886,
       "step": 8450
     },
     {
-      "epoch": 0.9148638467333979,
-      "grad_norm": 0.5731797218322754,
-      "learning_rate": 0.0005457019717702833,
-      "loss": 3.8498,
+      "epoch": 0.9164420485175202,
+      "grad_norm": 0.5699930787086487,
+      "learning_rate": 0.0005456082029141932,
+      "loss": 3.8766,
       "step": 8500
     },
     {
-      "epoch": 0.9202453987730062,
-      "grad_norm": 0.6198984384536743,
-      "learning_rate": 0.0005453787307402219,
-      "loss": 3.8389,
+      "epoch": 0.921832884097035,
+      "grad_norm": 0.5467044711112976,
+      "learning_rate": 0.0005452844036697248,
+      "loss": 3.8585,
       "step": 8550
     },
     {
-      "epoch": 0.9256269508126144,
-      "grad_norm": 0.600885272026062,
-      "learning_rate": 0.0005450554897101605,
-      "loss": 3.8547,
+      "epoch": 0.9272237196765498,
+      "grad_norm": 0.6719435453414917,
+      "learning_rate": 0.0005449670804101457,
+      "loss": 3.8533,
       "step": 8600
     },
     {
-      "epoch": 0.9310085028522226,
-      "grad_norm": 0.5399230122566223,
-      "learning_rate": 0.0005447322486800991,
-      "loss": 3.8578,
+      "epoch": 0.9326145552560647,
+      "grad_norm": 0.5468204617500305,
+      "learning_rate": 0.0005446432811656773,
+      "loss": 3.8461,
       "step": 8650
     },
     {
-      "epoch": 0.9363900548918308,
-      "grad_norm": 0.5042197108268738,
-      "learning_rate": 0.0005444090076500377,
-      "loss": 3.8613,
+      "epoch": 0.9380053908355795,
+      "grad_norm": 0.5554046630859375,
+      "learning_rate": 0.0005443194819212088,
+      "loss": 3.8516,
       "step": 8700
     },
     {
-      "epoch": 0.941771606931439,
-      "grad_norm": 0.5343637466430664,
-      "learning_rate": 0.0005440857666199763,
-      "loss": 3.8458,
+      "epoch": 0.9433962264150944,
+      "grad_norm": 0.5696746706962585,
+      "learning_rate": 0.0005439956826767404,
+      "loss": 3.8491,
       "step": 8750
     },
     {
-      "epoch": 0.9471531589710472,
-      "grad_norm": 0.5491181015968323,
-      "learning_rate": 0.0005437625255899148,
-      "loss": 3.8335,
+      "epoch": 0.9487870619946092,
+      "grad_norm": 0.5562222003936768,
+      "learning_rate": 0.000543671883432272,
+      "loss": 3.8421,
       "step": 8800
     },
     {
-      "epoch": 0.9525347110106555,
-      "grad_norm": 0.5667814612388611,
-      "learning_rate": 0.0005434392845598534,
-      "loss": 3.845,
+      "epoch": 0.954177897574124,
+      "grad_norm": 0.6517609357833862,
+      "learning_rate": 0.0005433480841878035,
+      "loss": 3.8343,
       "step": 8850
     },
     {
-      "epoch": 0.9579162630502637,
-      "grad_norm": 0.5124309659004211,
-      "learning_rate": 0.000543116043529792,
-      "loss": 3.8512,
+      "epoch": 0.9595687331536388,
+      "grad_norm": 0.570240318775177,
+      "learning_rate": 0.000543024284943335,
+      "loss": 3.8533,
       "step": 8900
     },
     {
-      "epoch": 0.9632978150898719,
-      "grad_norm": 0.5880436301231384,
-      "learning_rate": 0.0005427928024997306,
-      "loss": 3.8386,
+      "epoch": 0.9649595687331537,
+      "grad_norm": 0.5864043831825256,
+      "learning_rate": 0.0005427004856988667,
+      "loss": 3.8559,
       "step": 8950
     },
     {
-      "epoch": 0.9686793671294801,
-      "grad_norm": 0.5593850016593933,
-      "learning_rate": 0.0005424695614696692,
-      "loss": 3.8266,
+      "epoch": 0.9703504043126685,
+      "grad_norm": 0.5611904859542847,
+      "learning_rate": 0.0005423766864543982,
+      "loss": 3.8429,
       "step": 9000
     },
     {
-      "epoch": 0.9686793671294801,
-      "eval_accuracy": 0.34156187005092886,
-      "eval_loss": 3.7744014263153076,
-      "eval_runtime": 179.4219,
-      "eval_samples_per_second": 100.384,
-      "eval_steps_per_second": 6.276,
+      "epoch": 0.9703504043126685,
+      "eval_accuracy": 0.3413870475544354,
+      "eval_loss": 3.780839681625366,
+      "eval_runtime": 185.0401,
+      "eval_samples_per_second": 97.336,
+      "eval_steps_per_second": 6.085,
       "step": 9000
     },
     {
-      "epoch": 0.9740609191690883,
-      "grad_norm": 0.6109694838523865,
-      "learning_rate": 0.0005421463204396078,
-      "loss": 3.8173,
+      "epoch": 0.9757412398921833,
+      "grad_norm": 0.578873336315155,
+      "learning_rate": 0.0005420528872099298,
+      "loss": 3.8385,
       "step": 9050
     },
     {
-      "epoch": 0.9794424712086965,
-      "grad_norm": 0.5747610330581665,
-      "learning_rate": 0.0005418295442301476,
-      "loss": 3.8562,
+      "epoch": 0.9811320754716981,
+      "grad_norm": 0.657522439956665,
+      "learning_rate": 0.0005417290879654613,
+      "loss": 3.8524,
       "step": 9100
     },
     {
-      "epoch": 0.9848240232483048,
-      "grad_norm": 0.536498486995697,
-      "learning_rate": 0.0005415063032000861,
-      "loss": 3.8285,
+      "epoch": 0.9865229110512129,
+      "grad_norm": 0.5951496958732605,
+      "learning_rate": 0.000541405288720993,
+      "loss": 3.8276,
       "step": 9150
     },
     {
-      "epoch": 0.9902055752879131,
-      "grad_norm": 0.5872951745986938,
-      "learning_rate": 0.0005411830621700248,
-      "loss": 3.8311,
+      "epoch": 0.9919137466307277,
+      "grad_norm": 0.6185961961746216,
+      "learning_rate": 0.0005410814894765245,
+      "loss": 3.8227,
       "step": 9200
     },
     {
-      "epoch": 0.9955871273275213,
-      "grad_norm": 0.6314157843589783,
-      "learning_rate": 0.0005408598211399633,
-      "loss": 3.8067,
+      "epoch": 0.9973045822102425,
+      "grad_norm": 0.6109092831611633,
+      "learning_rate": 0.0005407576902320561,
+      "loss": 3.835,
       "step": 9250
     },
     {
-      "epoch": 1.0009686793671295,
-      "grad_norm": 0.5174910426139832,
-      "learning_rate": 0.0005405365801099019,
-      "loss": 3.8005,
+      "epoch": 1.0026954177897573,
+      "grad_norm": 0.5460496544837952,
+      "learning_rate": 0.0005404338909875876,
+      "loss": 3.8096,
       "step": 9300
     },
     {
-      "epoch": 1.0063502314067376,
-      "grad_norm": 0.5861645936965942,
-      "learning_rate": 0.0005402133390798405,
-      "loss": 3.7532,
+      "epoch": 1.0080862533692723,
+      "grad_norm": 0.5667276978492737,
+      "learning_rate": 0.0005401100917431192,
+      "loss": 3.7744,
       "step": 9350
     },
     {
-      "epoch": 1.011731783446346,
-      "grad_norm": 0.6189785003662109,
-      "learning_rate": 0.000539890098049779,
-      "loss": 3.7568,
+      "epoch": 1.013477088948787,
+      "grad_norm": 0.5935086607933044,
+      "learning_rate": 0.0005397862924986508,
+      "loss": 3.7692,
       "step": 9400
     },
     {
-      "epoch": 1.017113335485954,
-      "grad_norm": 0.5914862751960754,
-      "learning_rate": 0.0005395668570197177,
-      "loss": 3.7533,
+      "epoch": 1.0188679245283019,
+      "grad_norm": 0.6849828958511353,
+      "learning_rate": 0.0005394624932541824,
+      "loss": 3.7707,
       "step": 9450
     },
     {
-      "epoch": 1.0224948875255624,
-      "grad_norm": 0.618517279624939,
-      "learning_rate": 0.0005392436159896562,
-      "loss": 3.7495,
+      "epoch": 1.0242587601078168,
+      "grad_norm": 0.5856293439865112,
+      "learning_rate": 0.0005391386940097139,
+      "loss": 3.7689,
       "step": 9500
     },
     {
-      "epoch": 1.0278764395651705,
-      "grad_norm": 0.5718392133712769,
-      "learning_rate": 0.0005389203749595948,
-      "loss": 3.7624,
+      "epoch": 1.0296495956873315,
+      "grad_norm": 0.5455779433250427,
+      "learning_rate": 0.0005388148947652455,
+      "loss": 3.7681,
       "step": 9550
     },
     {
-      "epoch": 1.0332579916047788,
-      "grad_norm": 0.5652843713760376,
-      "learning_rate": 0.0005385971339295334,
-      "loss": 3.7637,
+      "epoch": 1.0350404312668464,
+      "grad_norm": 0.6295051574707031,
+      "learning_rate": 0.000538491095520777,
+      "loss": 3.7576,
       "step": 9600
     },
     {
-      "epoch": 1.0386395436443872,
-      "grad_norm": 0.6264984011650085,
-      "learning_rate": 0.000538273892899472,
-      "loss": 3.7699,
+      "epoch": 1.0404312668463611,
+      "grad_norm": 0.6791478395462036,
+      "learning_rate": 0.0005381672962763086,
+      "loss": 3.796,
       "step": 9650
     },
     {
-      "epoch": 1.0440210956839953,
-      "grad_norm": 0.5926131010055542,
-      "learning_rate": 0.0005379506518694106,
-      "loss": 3.7665,
+      "epoch": 1.045822102425876,
+      "grad_norm": 0.6339579820632935,
+      "learning_rate": 0.0005378434970318403,
+      "loss": 3.7826,
       "step": 9700
     },
     {
-      "epoch": 1.0494026477236036,
-      "grad_norm": 0.539849579334259,
-      "learning_rate": 0.0005376274108393491,
-      "loss": 3.7516,
+      "epoch": 1.0512129380053907,
+      "grad_norm": 0.5362462401390076,
+      "learning_rate": 0.0005375196977873718,
+      "loss": 3.7799,
       "step": 9750
     },
     {
-      "epoch": 1.0547841997632117,
-      "grad_norm": 0.626375138759613,
-      "learning_rate": 0.0005373041698092877,
-      "loss": 3.7572,
+      "epoch": 1.0566037735849056,
+      "grad_norm": 0.5385799407958984,
+      "learning_rate": 0.0005371958985429034,
+      "loss": 3.7875,
       "step": 9800
     },
     {
-      "epoch": 1.06016575180282,
-      "grad_norm": 0.5491912364959717,
-      "learning_rate": 0.0005369809287792263,
-      "loss": 3.7701,
+      "epoch": 1.0619946091644206,
+      "grad_norm": 0.5548085570335388,
+      "learning_rate": 0.0005368720992984349,
+      "loss": 3.7837,
       "step": 9850
     },
     {
-      "epoch": 1.0655473038424281,
-      "grad_norm": 0.5011469721794128,
-      "learning_rate": 0.000536657687749165,
-      "loss": 3.7628,
+      "epoch": 1.0673854447439353,
+      "grad_norm": 0.5549419522285461,
+      "learning_rate": 0.0005365483000539665,
+      "loss": 3.7745,
       "step": 9900
     },
     {
-      "epoch": 1.0709288558820365,
-      "grad_norm": 0.6251816153526306,
-      "learning_rate": 0.0005363344467191035,
-      "loss": 3.7487,
+      "epoch": 1.0727762803234502,
+      "grad_norm": 0.5701829195022583,
+      "learning_rate": 0.0005362245008094981,
+      "loss": 3.7694,
       "step": 9950
     },
     {
-      "epoch": 1.0763104079216446,
-      "grad_norm": 0.702565610408783,
-      "learning_rate": 0.000536011205689042,
-      "loss": 3.7541,
+      "epoch": 1.0781671159029649,
+      "grad_norm": 0.5939721465110779,
+      "learning_rate": 0.0005359007015650297,
+      "loss": 3.7625,
       "step": 10000
     },
     {
-      "epoch": 1.0763104079216446,
-      "eval_accuracy": 0.34536037500892314,
-      "eval_loss": 3.741765022277832,
-      "eval_runtime": 180.285,
-      "eval_samples_per_second": 99.903,
-      "eval_steps_per_second": 6.246,
+      "epoch": 1.0781671159029649,
+      "eval_accuracy": 0.34421343512515346,
+      "eval_loss": 3.7506351470947266,
+      "eval_runtime": 184.7523,
+      "eval_samples_per_second": 97.487,
+      "eval_steps_per_second": 6.095,
       "step": 10000
     },
     {
-      "epoch": 1.081691959961253,
-      "grad_norm": 0.5714436769485474,
-      "learning_rate": 0.0005356879646589807,
-      "loss": 3.7614,
+      "epoch": 1.0835579514824798,
+      "grad_norm": 0.5285921692848206,
+      "learning_rate": 0.0005355769023205612,
+      "loss": 3.7661,
       "step": 10050
     },
     {
-      "epoch": 1.087073512000861,
-      "grad_norm": 0.5856513977050781,
-      "learning_rate": 0.0005353647236289192,
-      "loss": 3.7646,
+      "epoch": 1.0889487870619945,
+      "grad_norm": 0.5900604128837585,
+      "learning_rate": 0.0005352531030760928,
+      "loss": 3.7748,
       "step": 10100
     },
     {
-      "epoch": 1.0924550640404693,
-      "grad_norm": 0.5325464606285095,
-      "learning_rate": 0.0005350414825988579,
-      "loss": 3.77,
+      "epoch": 1.0943396226415094,
+      "grad_norm": 0.5793938040733337,
+      "learning_rate": 0.0005349293038316244,
+      "loss": 3.7681,
       "step": 10150
     },
     {
-      "epoch": 1.0978366160800774,
-      "grad_norm": 0.6802165508270264,
-      "learning_rate": 0.0005347182415687964,
-      "loss": 3.745,
+      "epoch": 1.0997304582210243,
+      "grad_norm": 0.5640652179718018,
+      "learning_rate": 0.0005346055045871559,
+      "loss": 3.7681,
       "step": 10200
     },
     {
-      "epoch": 1.1032181681196858,
-      "grad_norm": 0.5991120934486389,
-      "learning_rate": 0.000534395000538735,
-      "loss": 3.7629,
+      "epoch": 1.105121293800539,
+      "grad_norm": 0.5162309408187866,
+      "learning_rate": 0.0005342817053426874,
+      "loss": 3.7755,
       "step": 10250
     },
     {
-      "epoch": 1.1085997201592939,
-      "grad_norm": 0.5548239350318909,
-      "learning_rate": 0.0005340717595086736,
-      "loss": 3.7561,
+      "epoch": 1.110512129380054,
+      "grad_norm": 0.5436460971832275,
+      "learning_rate": 0.0005339579060982191,
+      "loss": 3.763,
       "step": 10300
     },
     {
-      "epoch": 1.1139812721989022,
-      "grad_norm": 0.5725270509719849,
-      "learning_rate": 0.0005337485184786122,
-      "loss": 3.7547,
+      "epoch": 1.1159029649595686,
+      "grad_norm": 0.637514591217041,
+      "learning_rate": 0.0005336341068537506,
+      "loss": 3.75,
       "step": 10350
     },
     {
-      "epoch": 1.1193628242385103,
-      "grad_norm": 0.6069482564926147,
-      "learning_rate": 0.0005334252774485507,
-      "loss": 3.755,
+      "epoch": 1.1212938005390836,
+      "grad_norm": 0.5975980758666992,
+      "learning_rate": 0.0005333103076092822,
+      "loss": 3.7669,
       "step": 10400
     },
     {
-      "epoch": 1.1247443762781186,
-      "grad_norm": 0.5790407657623291,
-      "learning_rate": 0.0005331020364184894,
-      "loss": 3.742,
+      "epoch": 1.1266846361185983,
+      "grad_norm": 0.5235165953636169,
+      "learning_rate": 0.0005329865083648137,
+      "loss": 3.7541,
       "step": 10450
     },
     {
-      "epoch": 1.1301259283177267,
-      "grad_norm": 0.5913881659507751,
-      "learning_rate": 0.0005327787953884279,
-      "loss": 3.7511,
+      "epoch": 1.1320754716981132,
+      "grad_norm": 0.6354934573173523,
+      "learning_rate": 0.0005326627091203454,
+      "loss": 3.7428,
       "step": 10500
     },
     {
-      "epoch": 1.135507480357335,
-      "grad_norm": 0.6094677448272705,
-      "learning_rate": 0.0005324555543583665,
-      "loss": 3.7278,
+      "epoch": 1.137466307277628,
+      "grad_norm": 0.5056789517402649,
+      "learning_rate": 0.0005323389098758769,
+      "loss": 3.7647,
       "step": 10550
     },
     {
-      "epoch": 1.1408890323969434,
-      "grad_norm": 0.5710995197296143,
-      "learning_rate": 0.0005321323133283051,
-      "loss": 3.7383,
+      "epoch": 1.1428571428571428,
+      "grad_norm": 0.5910845398902893,
+      "learning_rate": 0.0005320151106314085,
+      "loss": 3.7758,
       "step": 10600
     },
     {
-      "epoch": 1.1462705844365515,
-      "grad_norm": 0.532470703125,
-      "learning_rate": 0.0005318090722982436,
-      "loss": 3.7373,
+      "epoch": 1.1482479784366577,
+      "grad_norm": 0.6386284232139587,
+      "learning_rate": 0.00053169131138694,
+      "loss": 3.7546,
       "step": 10650
     },
     {
-      "epoch": 1.1516521364761596,
-      "grad_norm": 0.5976846218109131,
-      "learning_rate": 0.0005314858312681823,
-      "loss": 3.7609,
+      "epoch": 1.1536388140161726,
+      "grad_norm": 0.5644304752349854,
+      "learning_rate": 0.0005313675121424716,
+      "loss": 3.751,
       "step": 10700
     },
     {
-      "epoch": 1.157033688515768,
-      "grad_norm": 0.5880893468856812,
-      "learning_rate": 0.0005311625902381209,
-      "loss": 3.7448,
+      "epoch": 1.1590296495956873,
+      "grad_norm": 0.5989267826080322,
+      "learning_rate": 0.0005310437128980032,
+      "loss": 3.7387,
       "step": 10750
     },
     {
-      "epoch": 1.1624152405553763,
-      "grad_norm": 0.6322693228721619,
-      "learning_rate": 0.0005308393492080595,
-      "loss": 3.7477,
+      "epoch": 1.1644204851752022,
+      "grad_norm": 0.545608639717102,
+      "learning_rate": 0.0005307199136535348,
+      "loss": 3.7279,
       "step": 10800
     },
     {
-      "epoch": 1.1677967925949844,
-      "grad_norm": 0.5472283363342285,
-      "learning_rate": 0.000530516108177998,
-      "loss": 3.737,
+      "epoch": 1.169811320754717,
+      "grad_norm": 0.5512583255767822,
+      "learning_rate": 0.0005303961144090663,
+      "loss": 3.7474,
       "step": 10850
     },
     {
-      "epoch": 1.1731783446345927,
-      "grad_norm": 0.5412709712982178,
-      "learning_rate": 0.0005301928671479365,
-      "loss": 3.7412,
+      "epoch": 1.1752021563342319,
+      "grad_norm": 0.5729238390922546,
+      "learning_rate": 0.0005300723151645979,
+      "loss": 3.7492,
       "step": 10900
     },
     {
-      "epoch": 1.1785598966742008,
-      "grad_norm": 0.5746583342552185,
-      "learning_rate": 0.0005298696261178752,
-      "loss": 3.7427,
+      "epoch": 1.1805929919137466,
+      "grad_norm": 0.596947431564331,
+      "learning_rate": 0.0005297485159201295,
+      "loss": 3.7606,
       "step": 10950
     },
     {
-      "epoch": 1.1839414487138091,
-      "grad_norm": 0.543925940990448,
-      "learning_rate": 0.0005295463850878138,
-      "loss": 3.7349,
+      "epoch": 1.1859838274932615,
+      "grad_norm": 0.6165409088134766,
+      "learning_rate": 0.000529424716675661,
+      "loss": 3.7757,
       "step": 11000
     },
     {
-      "epoch": 1.1839414487138091,
-      "eval_accuracy": 0.34741369728284116,
-      "eval_loss": 3.7178292274475098,
-      "eval_runtime": 179.9682,
-      "eval_samples_per_second": 100.079,
-      "eval_steps_per_second": 6.257,
+      "epoch": 1.1859838274932615,
+      "eval_accuracy": 0.3462492642841334,
+      "eval_loss": 3.726261854171753,
+      "eval_runtime": 184.6494,
+      "eval_samples_per_second": 97.542,
+      "eval_steps_per_second": 6.098,
       "step": 11000
     },
     {
-      "epoch": 1.1893230007534172,
-      "grad_norm": 0.6048632860183716,
-      "learning_rate": 0.0005292231440577524,
-      "loss": 3.7374,
+      "epoch": 1.1913746630727764,
+      "grad_norm": 0.5490753054618835,
+      "learning_rate": 0.0005291009174311926,
+      "loss": 3.7638,
       "step": 11050
     },
     {
-      "epoch": 1.1947045527930256,
-      "grad_norm": 0.5460487008094788,
-      "learning_rate": 0.0005288999030276909,
-      "loss": 3.7464,
+      "epoch": 1.196765498652291,
+      "grad_norm": 0.5522347092628479,
+      "learning_rate": 0.0005287771181867242,
+      "loss": 3.7567,
       "step": 11100
     },
     {
-      "epoch": 1.2000861048326337,
-      "grad_norm": 0.6168481111526489,
-      "learning_rate": 0.0005285766619976295,
-      "loss": 3.7319,
+      "epoch": 1.202156334231806,
+      "grad_norm": 0.5174296498298645,
+      "learning_rate": 0.0005284597949271452,
+      "loss": 3.7552,
       "step": 11150
     },
     {
-      "epoch": 1.205467656872242,
-      "grad_norm": 0.5460136532783508,
-      "learning_rate": 0.0005282534209675681,
-      "loss": 3.7478,
+      "epoch": 1.2075471698113207,
+      "grad_norm": 0.5238677263259888,
+      "learning_rate": 0.0005281359956826767,
+      "loss": 3.743,
       "step": 11200
     },
     {
-      "epoch": 1.21084920891185,
-      "grad_norm": 0.5822840332984924,
-      "learning_rate": 0.0005279301799375066,
-      "loss": 3.7287,
+      "epoch": 1.2129380053908356,
+      "grad_norm": 0.5870748162269592,
+      "learning_rate": 0.0005278121964382083,
+      "loss": 3.7358,
       "step": 11250
     },
     {
-      "epoch": 1.2162307609514584,
-      "grad_norm": 0.528600811958313,
-      "learning_rate": 0.0005276069389074453,
-      "loss": 3.7461,
+      "epoch": 1.2183288409703503,
+      "grad_norm": 0.6238572597503662,
+      "learning_rate": 0.0005274883971937398,
+      "loss": 3.7525,
       "step": 11300
     },
     {
-      "epoch": 1.2216123129910665,
-      "grad_norm": 0.5433237552642822,
-      "learning_rate": 0.0005272836978773838,
-      "loss": 3.738,
+      "epoch": 1.2237196765498652,
+      "grad_norm": 0.6007615923881531,
+      "learning_rate": 0.0005271645979492714,
+      "loss": 3.7364,
       "step": 11350
     },
     {
-      "epoch": 1.2269938650306749,
-      "grad_norm": 0.5144934058189392,
-      "learning_rate": 0.0005269604568473225,
-      "loss": 3.6953,
+      "epoch": 1.2291105121293802,
+      "grad_norm": 0.5366285443305969,
+      "learning_rate": 0.000526840798704803,
+      "loss": 3.7556,
       "step": 11400
     },
     {
-      "epoch": 1.232375417070283,
-      "grad_norm": 0.5954382419586182,
-      "learning_rate": 0.000526637215817261,
-      "loss": 3.7262,
+      "epoch": 1.2345013477088949,
+      "grad_norm": 0.5180173516273499,
+      "learning_rate": 0.0005265169994603346,
+      "loss": 3.7442,
       "step": 11450
     },
     {
-      "epoch": 1.2377569691098913,
-      "grad_norm": 0.6082680225372314,
-      "learning_rate": 0.0005263139747871996,
-      "loss": 3.7328,
+      "epoch": 1.2398921832884098,
+      "grad_norm": 0.5628926157951355,
+      "learning_rate": 0.0005261932002158661,
+      "loss": 3.7362,
       "step": 11500
     },
     {
-      "epoch": 1.2431385211494996,
-      "grad_norm": 0.510694682598114,
-      "learning_rate": 0.0005259907337571381,
-      "loss": 3.7532,
+      "epoch": 1.2452830188679245,
+      "grad_norm": 0.5104094743728638,
+      "learning_rate": 0.0005258694009713977,
+      "loss": 3.7472,
       "step": 11550
     },
     {
-      "epoch": 1.2485200731891077,
-      "grad_norm": 0.5158873796463013,
-      "learning_rate": 0.0005256674927270768,
-      "loss": 3.7257,
+      "epoch": 1.2506738544474394,
+      "grad_norm": 0.5553884506225586,
+      "learning_rate": 0.0005255456017269292,
+      "loss": 3.7208,
       "step": 11600
     },
     {
-      "epoch": 1.2539016252287158,
-      "grad_norm": 0.559540331363678,
-      "learning_rate": 0.0005253442516970154,
-      "loss": 3.7084,
+      "epoch": 1.256064690026954,
+      "grad_norm": 0.5858808755874634,
+      "learning_rate": 0.0005252218024824608,
+      "loss": 3.7538,
       "step": 11650
     },
     {
-      "epoch": 1.2592831772683242,
-      "grad_norm": 0.6382985711097717,
-      "learning_rate": 0.000525021010666954,
-      "loss": 3.7259,
+      "epoch": 1.261455525606469,
+      "grad_norm": 0.694400429725647,
+      "learning_rate": 0.0005248980032379924,
+      "loss": 3.7502,
       "step": 11700
     },
     {
-      "epoch": 1.2646647293079325,
-      "grad_norm": 0.6098189353942871,
-      "learning_rate": 0.0005246977696368925,
-      "loss": 3.7354,
+      "epoch": 1.266846361185984,
+      "grad_norm": 0.552725613117218,
+      "learning_rate": 0.000524574203993524,
+      "loss": 3.7423,
       "step": 11750
     },
     {
-      "epoch": 1.2700462813475406,
-      "grad_norm": 0.5487571358680725,
-      "learning_rate": 0.0005243745286068311,
-      "loss": 3.7322,
+      "epoch": 1.2722371967654986,
+      "grad_norm": 0.5437906980514526,
+      "learning_rate": 0.0005242504047490555,
+      "loss": 3.7346,
       "step": 11800
     },
     {
-      "epoch": 1.275427833387149,
-      "grad_norm": 0.5637610554695129,
-      "learning_rate": 0.0005240577523973709,
-      "loss": 3.7314,
+      "epoch": 1.2776280323450135,
+      "grad_norm": 0.6244624257087708,
+      "learning_rate": 0.0005239266055045871,
+      "loss": 3.7441,
       "step": 11850
     },
     {
-      "epoch": 1.280809385426757,
-      "grad_norm": 0.5707816481590271,
-      "learning_rate": 0.0005237345113673095,
-      "loss": 3.7303,
+      "epoch": 1.2830188679245282,
+      "grad_norm": 0.5783806443214417,
+      "learning_rate": 0.0005236028062601186,
+      "loss": 3.7325,
       "step": 11900
     },
     {
-      "epoch": 1.2861909374663654,
-      "grad_norm": 0.5442221164703369,
-      "learning_rate": 0.0005234112703372481,
-      "loss": 3.7523,
+      "epoch": 1.2884097035040432,
+      "grad_norm": 0.5712251663208008,
+      "learning_rate": 0.0005232790070156503,
+      "loss": 3.7446,
       "step": 11950
     },
     {
-      "epoch": 1.2915724895059735,
-      "grad_norm": 0.7692139148712158,
-      "learning_rate": 0.0005230880293071867,
-      "loss": 3.7312,
+      "epoch": 1.2938005390835579,
+      "grad_norm": 0.637886643409729,
+      "learning_rate": 0.0005229552077711818,
+      "loss": 3.7387,
       "step": 12000
     },
     {
-      "epoch": 1.2915724895059735,
-      "eval_accuracy": 0.3501756536910853,
-      "eval_loss": 3.6950345039367676,
-      "eval_runtime": 182.9419,
-      "eval_samples_per_second": 98.452,
-      "eval_steps_per_second": 6.155,
+      "epoch": 1.2938005390835579,
+      "eval_accuracy": 0.3485640054061331,
+      "eval_loss": 3.701038122177124,
+      "eval_runtime": 184.7994,
+      "eval_samples_per_second": 97.462,
+      "eval_steps_per_second": 6.093,
       "step": 12000
     },
     {
-      "epoch": 1.2969540415455818,
-      "grad_norm": 0.5720347762107849,
-      "learning_rate": 0.0005227647882771253,
-      "loss": 3.7174,
+      "epoch": 1.2991913746630728,
+      "grad_norm": 0.539336621761322,
+      "learning_rate": 0.0005226314085267134,
+      "loss": 3.7356,
       "step": 12050
     },
     {
-      "epoch": 1.30233559358519,
-      "grad_norm": 0.5530905723571777,
-      "learning_rate": 0.0005224415472470639,
-      "loss": 3.7105,
+      "epoch": 1.3045822102425877,
+      "grad_norm": 0.5483951568603516,
+      "learning_rate": 0.000522307609282245,
+      "loss": 3.7429,
       "step": 12100
     },
     {
-      "epoch": 1.3077171456247982,
-      "grad_norm": 0.5424105525016785,
-      "learning_rate": 0.0005221183062170024,
-      "loss": 3.7247,
+      "epoch": 1.3099730458221024,
+      "grad_norm": 0.5724830627441406,
+      "learning_rate": 0.0005219838100377766,
+      "loss": 3.7593,
       "step": 12150
     },
     {
-      "epoch": 1.3130986976644063,
-      "grad_norm": 0.5885405540466309,
-      "learning_rate": 0.0005217950651869409,
-      "loss": 3.7127,
+      "epoch": 1.3153638814016173,
+      "grad_norm": 0.5383995771408081,
+      "learning_rate": 0.000521660010793308,
+      "loss": 3.7294,
       "step": 12200
     },
     {
-      "epoch": 1.3184802497040147,
-      "grad_norm": 0.5528365969657898,
-      "learning_rate": 0.0005214718241568796,
-      "loss": 3.7347,
+      "epoch": 1.320754716981132,
+      "grad_norm": 0.546576738357544,
+      "learning_rate": 0.0005213362115488396,
+      "loss": 3.736,
       "step": 12250
     },
     {
-      "epoch": 1.3238618017436228,
-      "grad_norm": 0.5199754238128662,
-      "learning_rate": 0.0005211485831268182,
-      "loss": 3.7388,
+      "epoch": 1.326145552560647,
+      "grad_norm": 0.570923388004303,
+      "learning_rate": 0.0005210124123043713,
+      "loss": 3.7464,
       "step": 12300
     },
     {
-      "epoch": 1.329243353783231,
-      "grad_norm": 0.581535279750824,
-      "learning_rate": 0.0005208253420967568,
-      "loss": 3.7209,
+      "epoch": 1.3315363881401616,
+      "grad_norm": 0.579967200756073,
+      "learning_rate": 0.0005206886130599028,
+      "loss": 3.7331,
       "step": 12350
     },
     {
-      "epoch": 1.3346249058228392,
-      "grad_norm": 0.614345371723175,
-      "learning_rate": 0.0005205021010666953,
-      "loss": 3.7146,
+      "epoch": 1.3369272237196765,
+      "grad_norm": 0.6479465961456299,
+      "learning_rate": 0.0005203648138154344,
+      "loss": 3.722,
       "step": 12400
     },
     {
-      "epoch": 1.3400064578624475,
-      "grad_norm": 0.6020660400390625,
-      "learning_rate": 0.0005201788600366339,
-      "loss": 3.7209,
+      "epoch": 1.3423180592991915,
+      "grad_norm": 0.552336573600769,
+      "learning_rate": 0.0005200410145709659,
+      "loss": 3.7365,
       "step": 12450
     },
     {
-      "epoch": 1.3453880099020559,
-      "grad_norm": 0.5743170976638794,
-      "learning_rate": 0.0005198556190065725,
-      "loss": 3.7123,
+      "epoch": 1.3477088948787062,
+      "grad_norm": 0.7343564629554749,
+      "learning_rate": 0.0005197172153264976,
+      "loss": 3.7177,
       "step": 12500
     },
     {
-      "epoch": 1.350769561941664,
-      "grad_norm": 0.5332421660423279,
-      "learning_rate": 0.0005195323779765112,
-      "loss": 3.7184,
+      "epoch": 1.353099730458221,
+      "grad_norm": 0.5686933398246765,
+      "learning_rate": 0.0005193934160820291,
+      "loss": 3.7304,
       "step": 12550
     },
     {
-      "epoch": 1.356151113981272,
-      "grad_norm": 0.5720219612121582,
-      "learning_rate": 0.0005192091369464497,
-      "loss": 3.7093,
+      "epoch": 1.3584905660377358,
+      "grad_norm": 0.6135874390602112,
+      "learning_rate": 0.0005190696168375607,
+      "loss": 3.7254,
       "step": 12600
     },
     {
-      "epoch": 1.3615326660208804,
-      "grad_norm": 0.5706057548522949,
-      "learning_rate": 0.0005188858959163882,
-      "loss": 3.7122,
+      "epoch": 1.3638814016172507,
+      "grad_norm": 0.5901519060134888,
+      "learning_rate": 0.0005187458175930922,
+      "loss": 3.7145,
       "step": 12650
     },
     {
-      "epoch": 1.3669142180604887,
-      "grad_norm": 0.5560119152069092,
-      "learning_rate": 0.0005185626548863269,
-      "loss": 3.7149,
+      "epoch": 1.3692722371967654,
+      "grad_norm": 0.5513310432434082,
+      "learning_rate": 0.0005184220183486238,
+      "loss": 3.7122,
       "step": 12700
     },
     {
-      "epoch": 1.3722957701000968,
-      "grad_norm": 0.6042450666427612,
-      "learning_rate": 0.0005182394138562654,
-      "loss": 3.7158,
+      "epoch": 1.3746630727762803,
+      "grad_norm": 0.6472775936126709,
+      "learning_rate": 0.0005180982191041554,
+      "loss": 3.7296,
       "step": 12750
     },
     {
-      "epoch": 1.3776773221397052,
-      "grad_norm": 0.6797989010810852,
-      "learning_rate": 0.0005179161728262041,
-      "loss": 3.7213,
+      "epoch": 1.3800539083557952,
+      "grad_norm": 0.5848823189735413,
+      "learning_rate": 0.000517774419859687,
+      "loss": 3.7099,
       "step": 12800
     },
     {
-      "epoch": 1.3830588741793133,
-      "grad_norm": 0.6140666007995605,
-      "learning_rate": 0.0005175929317961426,
-      "loss": 3.7106,
+      "epoch": 1.38544474393531,
+      "grad_norm": 0.6691808104515076,
+      "learning_rate": 0.0005174506206152185,
+      "loss": 3.7238,
       "step": 12850
     },
     {
-      "epoch": 1.3884404262189216,
-      "grad_norm": 0.5921536684036255,
-      "learning_rate": 0.0005172696907660812,
-      "loss": 3.7036,
+      "epoch": 1.3908355795148248,
+      "grad_norm": 0.5118029117584229,
+      "learning_rate": 0.0005171268213707501,
+      "loss": 3.7223,
       "step": 12900
     },
     {
-      "epoch": 1.3938219782585297,
-      "grad_norm": 0.547536313533783,
-      "learning_rate": 0.0005169464497360198,
-      "loss": 3.7023,
+      "epoch": 1.3962264150943398,
+      "grad_norm": 0.5811614394187927,
+      "learning_rate": 0.0005168030221262816,
+      "loss": 3.7143,
       "step": 12950
     },
     {
-      "epoch": 1.399203530298138,
-      "grad_norm": 0.5648983716964722,
-      "learning_rate": 0.0005166232087059583,
-      "loss": 3.7127,
+      "epoch": 1.4016172506738545,
+      "grad_norm": 0.5549213886260986,
+      "learning_rate": 0.0005164792228818132,
+      "loss": 3.7098,
       "step": 13000
     },
     {
-      "epoch": 1.399203530298138,
-      "eval_accuracy": 0.35228004282227615,
-      "eval_loss": 3.673454761505127,
-      "eval_runtime": 182.9372,
-      "eval_samples_per_second": 98.455,
-      "eval_steps_per_second": 6.155,
+      "epoch": 1.4016172506738545,
+      "eval_accuracy": 0.35118536497754527,
+      "eval_loss": 3.678708076477051,
+      "eval_runtime": 184.9082,
+      "eval_samples_per_second": 97.405,
+      "eval_steps_per_second": 6.09,
       "step": 13000
     },
     {
-      "epoch": 1.4045850823377461,
-      "grad_norm": 0.5507751107215881,
-      "learning_rate": 0.0005162999676758969,
-      "loss": 3.7025,
+      "epoch": 1.4070080862533692,
+      "grad_norm": 0.5535794496536255,
+      "learning_rate": 0.0005161554236373448,
+      "loss": 3.6872,
       "step": 13050
     },
     {
-      "epoch": 1.4099666343773545,
-      "grad_norm": 0.5451176166534424,
-      "learning_rate": 0.0005159767266458355,
-      "loss": 3.7267,
+      "epoch": 1.412398921832884,
+      "grad_norm": 0.5400480031967163,
+      "learning_rate": 0.0005158316243928764,
+      "loss": 3.7089,
       "step": 13100
     },
     {
-      "epoch": 1.4153481864169626,
-      "grad_norm": 0.6242395043373108,
-      "learning_rate": 0.0005156534856157741,
-      "loss": 3.6888,
+      "epoch": 1.417789757412399,
+      "grad_norm": 0.5891815423965454,
+      "learning_rate": 0.0005155078251484079,
+      "loss": 3.7138,
       "step": 13150
     },
     {
-      "epoch": 1.420729738456571,
-      "grad_norm": 0.5988582372665405,
-      "learning_rate": 0.0005153302445857127,
-      "loss": 3.7165,
+      "epoch": 1.4231805929919137,
+      "grad_norm": 0.5472397208213806,
+      "learning_rate": 0.0005151840259039395,
+      "loss": 3.6966,
       "step": 13200
     },
     {
-      "epoch": 1.426111290496179,
-      "grad_norm": 0.5528572201728821,
-      "learning_rate": 0.0005150070035556513,
-      "loss": 3.6995,
+      "epoch": 1.4285714285714286,
+      "grad_norm": 0.5532459020614624,
+      "learning_rate": 0.000514860226659471,
+      "loss": 3.7152,
       "step": 13250
     },
     {
-      "epoch": 1.4314928425357873,
-      "grad_norm": 0.581158459186554,
-      "learning_rate": 0.0005146837625255898,
-      "loss": 3.7006,
+      "epoch": 1.4339622641509435,
+      "grad_norm": 0.5864548683166504,
+      "learning_rate": 0.0005145364274150027,
+      "loss": 3.7092,
       "step": 13300
     },
     {
-      "epoch": 1.4368743945753955,
-      "grad_norm": 0.5531030297279358,
-      "learning_rate": 0.0005143605214955285,
-      "loss": 3.6895,
+      "epoch": 1.4393530997304582,
+      "grad_norm": 0.5949729681015015,
+      "learning_rate": 0.0005142126281705343,
+      "loss": 3.7166,
       "step": 13350
     },
     {
-      "epoch": 1.4422559466150038,
-      "grad_norm": 0.5417513847351074,
-      "learning_rate": 0.0005140372804654671,
-      "loss": 3.6885,
+      "epoch": 1.444743935309973,
+      "grad_norm": 0.5387771129608154,
+      "learning_rate": 0.0005138888289260658,
+      "loss": 3.7041,
       "step": 13400
     },
     {
-      "epoch": 1.447637498654612,
-      "grad_norm": 0.5422365665435791,
-      "learning_rate": 0.0005137140394354056,
-      "loss": 3.7015,
+      "epoch": 1.4501347708894878,
+      "grad_norm": 0.5066041350364685,
+      "learning_rate": 0.0005135650296815974,
+      "loss": 3.7209,
       "step": 13450
     },
     {
-      "epoch": 1.4530190506942202,
-      "grad_norm": 0.5661604404449463,
-      "learning_rate": 0.0005133907984053442,
-      "loss": 3.6965,
+      "epoch": 1.4555256064690028,
+      "grad_norm": 0.6268380284309387,
+      "learning_rate": 0.0005132412304371289,
+      "loss": 3.7135,
       "step": 13500
     },
     {
-      "epoch": 1.4584006027338283,
-      "grad_norm": 0.581283688545227,
-      "learning_rate": 0.0005130675573752827,
-      "loss": 3.7015,
+      "epoch": 1.4609164420485174,
+      "grad_norm": 0.5601277351379395,
+      "learning_rate": 0.0005129174311926605,
+      "loss": 3.6909,
       "step": 13550
     },
     {
-      "epoch": 1.4637821547734367,
-      "grad_norm": 0.5587770938873291,
-      "learning_rate": 0.0005127443163452214,
-      "loss": 3.7007,
+      "epoch": 1.4663072776280324,
+      "grad_norm": 0.552325963973999,
+      "learning_rate": 0.0005126001079330814,
+      "loss": 3.7188,
       "step": 13600
     },
     {
-      "epoch": 1.469163706813045,
-      "grad_norm": 0.5291188955307007,
-      "learning_rate": 0.00051242107531516,
-      "loss": 3.6855,
+      "epoch": 1.4716981132075473,
+      "grad_norm": 0.48801907896995544,
+      "learning_rate": 0.000512276308688613,
+      "loss": 3.7012,
       "step": 13650
     },
     {
-      "epoch": 1.474545258852653,
-      "grad_norm": 0.5866516828536987,
-      "learning_rate": 0.0005120978342850986,
-      "loss": 3.6982,
+      "epoch": 1.477088948787062,
+      "grad_norm": 0.5625745058059692,
+      "learning_rate": 0.0005119525094441446,
+      "loss": 3.7185,
       "step": 13700
     },
     {
-      "epoch": 1.4799268108922612,
-      "grad_norm": 0.6090840697288513,
-      "learning_rate": 0.0005117745932550371,
-      "loss": 3.7063,
+      "epoch": 1.482479784366577,
+      "grad_norm": 0.5748574733734131,
+      "learning_rate": 0.0005116287101996762,
+      "loss": 3.6912,
       "step": 13750
     },
     {
-      "epoch": 1.4853083629318695,
-      "grad_norm": 0.5599469542503357,
-      "learning_rate": 0.0005114513522249758,
-      "loss": 3.6826,
+      "epoch": 1.4878706199460916,
+      "grad_norm": 0.590937614440918,
+      "learning_rate": 0.0005113049109552077,
+      "loss": 3.7156,
       "step": 13800
     },
     {
-      "epoch": 1.4906899149714778,
-      "grad_norm": 0.5345983505249023,
-      "learning_rate": 0.0005111281111949143,
-      "loss": 3.7033,
+      "epoch": 1.4932614555256065,
+      "grad_norm": 0.5972180962562561,
+      "learning_rate": 0.0005109811117107393,
+      "loss": 3.7,
       "step": 13850
     },
     {
-      "epoch": 1.496071467011086,
-      "grad_norm": 0.582324206829071,
-      "learning_rate": 0.0005108048701648528,
-      "loss": 3.706,
+      "epoch": 1.4986522911051212,
+      "grad_norm": 0.5326563715934753,
+      "learning_rate": 0.0005106573124662708,
+      "loss": 3.7097,
       "step": 13900
     },
     {
-      "epoch": 1.501453019050694,
-      "grad_norm": 0.5937202572822571,
-      "learning_rate": 0.0005104880939553926,
-      "loss": 3.6788,
+      "epoch": 1.5040431266846361,
+      "grad_norm": 0.5362458825111389,
+      "learning_rate": 0.0005103335132218025,
+      "loss": 3.699,
       "step": 13950
     },
     {
-      "epoch": 1.5068345710903024,
-      "grad_norm": 0.5526854395866394,
-      "learning_rate": 0.0005101648529253313,
-      "loss": 3.6974,
+      "epoch": 1.509433962264151,
+      "grad_norm": 0.5459834337234497,
+      "learning_rate": 0.000510009713977334,
+      "loss": 3.6989,
       "step": 14000
     },
     {
-      "epoch": 1.5068345710903024,
-      "eval_accuracy": 0.3537209974204718,
-      "eval_loss": 3.6525895595550537,
-      "eval_runtime": 183.0004,
-      "eval_samples_per_second": 98.421,
-      "eval_steps_per_second": 6.153,
+      "epoch": 1.509433962264151,
+      "eval_accuracy": 0.353066689730053,
+      "eval_loss": 3.6584391593933105,
+      "eval_runtime": 184.8467,
+      "eval_samples_per_second": 97.437,
+      "eval_steps_per_second": 6.092,
       "step": 14000
     },
     {
-      "epoch": 1.5122161231299107,
-      "grad_norm": 0.530371904373169,
-      "learning_rate": 0.0005098416118952699,
-      "loss": 3.7083,
+      "epoch": 1.5148247978436657,
+      "grad_norm": 0.5656163096427917,
+      "learning_rate": 0.0005096859147328656,
+      "loss": 3.6969,
       "step": 14050
     },
     {
-      "epoch": 1.5175976751695188,
-      "grad_norm": 0.5713452100753784,
-      "learning_rate": 0.0005095183708652085,
-      "loss": 3.681,
+      "epoch": 1.5202156334231804,
+      "grad_norm": 0.5515555739402771,
+      "learning_rate": 0.0005093621154883971,
+      "loss": 3.6973,
       "step": 14100
     },
     {
-      "epoch": 1.5229792272091272,
-      "grad_norm": 0.5218368768692017,
-      "learning_rate": 0.000509195129835147,
-      "loss": 3.6881,
+      "epoch": 1.5256064690026954,
+      "grad_norm": 0.5614264607429504,
+      "learning_rate": 0.0005090383162439288,
+      "loss": 3.6884,
       "step": 14150
     },
     {
-      "epoch": 1.5283607792487355,
-      "grad_norm": 0.5329176783561707,
-      "learning_rate": 0.0005088718888050856,
-      "loss": 3.6926,
+      "epoch": 1.5309973045822103,
+      "grad_norm": 0.6352477669715881,
+      "learning_rate": 0.0005087145169994602,
+      "loss": 3.6887,
       "step": 14200
     },
     {
-      "epoch": 1.5337423312883436,
-      "grad_norm": 0.5380780100822449,
-      "learning_rate": 0.0005085486477750242,
-      "loss": 3.705,
+      "epoch": 1.536388140161725,
+      "grad_norm": 0.542805552482605,
+      "learning_rate": 0.0005083907177549918,
+      "loss": 3.7049,
       "step": 14250
     },
     {
-      "epoch": 1.5391238833279517,
-      "grad_norm": 0.5449959635734558,
-      "learning_rate": 0.0005082254067449629,
-      "loss": 3.6826,
+      "epoch": 1.54177897574124,
+      "grad_norm": 0.6061757206916809,
+      "learning_rate": 0.0005080669185105234,
+      "loss": 3.7037,
       "step": 14300
     },
     {
-      "epoch": 1.54450543536756,
-      "grad_norm": 0.6050571203231812,
-      "learning_rate": 0.0005079021657149014,
-      "loss": 3.6679,
+      "epoch": 1.5471698113207548,
+      "grad_norm": 0.616715669631958,
+      "learning_rate": 0.000507743119266055,
+      "loss": 3.6958,
       "step": 14350
     },
     {
-      "epoch": 1.5498869874071683,
-      "grad_norm": 0.5612210631370544,
-      "learning_rate": 0.0005075789246848399,
-      "loss": 3.6984,
+      "epoch": 1.5525606469002695,
+      "grad_norm": 0.571155846118927,
+      "learning_rate": 0.0005074193200215865,
+      "loss": 3.7111,
       "step": 14400
     },
     {
-      "epoch": 1.5552685394467765,
-      "grad_norm": 0.5346029996871948,
-      "learning_rate": 0.0005072556836547785,
-      "loss": 3.6831,
+      "epoch": 1.5579514824797842,
+      "grad_norm": 0.586685299873352,
+      "learning_rate": 0.0005070955207771181,
+      "loss": 3.7035,
       "step": 14450
     },
     {
-      "epoch": 1.5606500914863846,
-      "grad_norm": 0.5634347200393677,
-      "learning_rate": 0.0005069324426247171,
-      "loss": 3.6772,
+      "epoch": 1.5633423180592994,
+      "grad_norm": 0.6216734051704407,
+      "learning_rate": 0.0005067717215326498,
+      "loss": 3.6841,
       "step": 14500
     },
     {
-      "epoch": 1.566031643525993,
-      "grad_norm": 0.5833031535148621,
-      "learning_rate": 0.0005066092015946557,
-      "loss": 3.6852,
+      "epoch": 1.568733153638814,
+      "grad_norm": 0.5405490398406982,
+      "learning_rate": 0.0005064479222881813,
+      "loss": 3.6997,
       "step": 14550
     },
     {
-      "epoch": 1.5714131955656012,
-      "grad_norm": 0.5364494919776917,
-      "learning_rate": 0.0005062859605645943,
-      "loss": 3.6904,
+      "epoch": 1.5741239892183287,
+      "grad_norm": 0.5977158546447754,
+      "learning_rate": 0.0005061241230437129,
+      "loss": 3.6862,
       "step": 14600
     },
     {
-      "epoch": 1.5767947476052093,
-      "grad_norm": 0.592128336429596,
-      "learning_rate": 0.0005059627195345329,
-      "loss": 3.6875,
+      "epoch": 1.5795148247978437,
+      "grad_norm": 0.5780313014984131,
+      "learning_rate": 0.0005058003237992444,
+      "loss": 3.6827,
       "step": 14650
     },
     {
-      "epoch": 1.5821762996448174,
-      "grad_norm": 0.5206913352012634,
-      "learning_rate": 0.0005056394785044715,
-      "loss": 3.6847,
+      "epoch": 1.5849056603773586,
+      "grad_norm": 0.5581985116004944,
+      "learning_rate": 0.000505476524554776,
+      "loss": 3.6985,
       "step": 14700
     },
     {
-      "epoch": 1.5875578516844258,
-      "grad_norm": 0.5297030806541443,
-      "learning_rate": 0.00050531623747441,
-      "loss": 3.677,
+      "epoch": 1.5902964959568733,
+      "grad_norm": 0.5698649883270264,
+      "learning_rate": 0.0005051527253103076,
+      "loss": 3.6979,
       "step": 14750
     },
     {
-      "epoch": 1.592939403724034,
-      "grad_norm": 0.6701345443725586,
-      "learning_rate": 0.0005049929964443486,
-      "loss": 3.6901,
+      "epoch": 1.595687331536388,
+      "grad_norm": 0.5590009093284607,
+      "learning_rate": 0.0005048289260658392,
+      "loss": 3.6692,
       "step": 14800
     },
     {
-      "epoch": 1.5983209557636422,
-      "grad_norm": 0.5313422083854675,
-      "learning_rate": 0.0005046697554142871,
-      "loss": 3.6859,
+      "epoch": 1.6010781671159031,
+      "grad_norm": 0.5426591038703918,
+      "learning_rate": 0.0005045051268213707,
+      "loss": 3.6912,
       "step": 14850
     },
     {
-      "epoch": 1.6037025078032503,
-      "grad_norm": 0.5316382646560669,
-      "learning_rate": 0.0005043465143842258,
-      "loss": 3.6802,
+      "epoch": 1.6064690026954178,
+      "grad_norm": 0.6016737818717957,
+      "learning_rate": 0.0005041813275769023,
+      "loss": 3.698,
       "step": 14900
     },
     {
-      "epoch": 1.6090840598428586,
-      "grad_norm": 0.6596531271934509,
-      "learning_rate": 0.0005040232733541644,
-      "loss": 3.6445,
+      "epoch": 1.6118598382749325,
+      "grad_norm": 0.5573039650917053,
+      "learning_rate": 0.0005038575283324338,
+      "loss": 3.6874,
       "step": 14950
     },
     {
-      "epoch": 1.614465611882467,
-      "grad_norm": 0.6033098101615906,
-      "learning_rate": 0.000503700032324103,
-      "loss": 3.6788,
+      "epoch": 1.6172506738544474,
+      "grad_norm": 0.5686517953872681,
+      "learning_rate": 0.0005035337290879654,
+      "loss": 3.6871,
       "step": 15000
     },
     {
-      "epoch": 1.614465611882467,
-      "eval_accuracy": 0.35575758714966643,
-      "eval_loss": 3.6347877979278564,
-      "eval_runtime": 183.0426,
-      "eval_samples_per_second": 98.398,
-      "eval_steps_per_second": 6.152,
+      "epoch": 1.6172506738544474,
+      "eval_accuracy": 0.35506025291567306,
+      "eval_loss": 3.6408228874206543,
+      "eval_runtime": 184.8073,
+      "eval_samples_per_second": 97.458,
+      "eval_steps_per_second": 6.093,
       "step": 15000
     },
     {
-      "epoch": 1.619847163922075,
-      "grad_norm": 0.5396302342414856,
-      "learning_rate": 0.0005033767912940415,
-      "loss": 3.663,
+      "epoch": 1.6226415094339623,
+      "grad_norm": 0.5335454344749451,
+      "learning_rate": 0.000503209929843497,
+      "loss": 3.678,
       "step": 15050
     },
     {
-      "epoch": 1.6252287159616834,
-      "grad_norm": 0.6209707856178284,
-      "learning_rate": 0.0005030535502639802,
-      "loss": 3.6696,
+      "epoch": 1.628032345013477,
+      "grad_norm": 0.5367480516433716,
+      "learning_rate": 0.0005028861305990286,
+      "loss": 3.6716,
       "step": 15100
     },
     {
-      "epoch": 1.6306102680012917,
-      "grad_norm": 0.5924604535102844,
-      "learning_rate": 0.0005027303092339187,
-      "loss": 3.6585,
+      "epoch": 1.633423180592992,
+      "grad_norm": 0.5422275066375732,
+      "learning_rate": 0.0005025623313545601,
+      "loss": 3.6707,
       "step": 15150
     },
     {
-      "epoch": 1.6359918200408998,
-      "grad_norm": 0.5263779163360596,
-      "learning_rate": 0.0005024070682038573,
-      "loss": 3.6734,
+      "epoch": 1.6388140161725069,
+      "grad_norm": 0.5667644739151001,
+      "learning_rate": 0.0005022385321100917,
+      "loss": 3.6751,
       "step": 15200
     },
     {
-      "epoch": 1.641373372080508,
-      "grad_norm": 0.5559638142585754,
-      "learning_rate": 0.0005020838271737959,
-      "loss": 3.6798,
+      "epoch": 1.6442048517520216,
+      "grad_norm": 0.6019392609596252,
+      "learning_rate": 0.0005019147328656232,
+      "loss": 3.6686,
       "step": 15250
     },
     {
-      "epoch": 1.6467549241201163,
-      "grad_norm": 0.5919946432113647,
-      "learning_rate": 0.0005017605861437344,
-      "loss": 3.6702,
+      "epoch": 1.6495956873315363,
+      "grad_norm": 0.566340982913971,
+      "learning_rate": 0.0005015909336211549,
+      "loss": 3.674,
       "step": 15300
     },
     {
-      "epoch": 1.6521364761597246,
-      "grad_norm": 0.4996275007724762,
-      "learning_rate": 0.0005014373451136731,
-      "loss": 3.673,
+      "epoch": 1.6549865229110512,
+      "grad_norm": 0.5398353338241577,
+      "learning_rate": 0.0005012671343766864,
+      "loss": 3.6976,
       "step": 15350
     },
     {
-      "epoch": 1.6575180281993327,
-      "grad_norm": 0.5767028331756592,
-      "learning_rate": 0.0005011141040836116,
-      "loss": 3.6692,
+      "epoch": 1.6603773584905661,
+      "grad_norm": 0.5944126844406128,
+      "learning_rate": 0.000500943335132218,
+      "loss": 3.6878,
       "step": 15400
     },
     {
-      "epoch": 1.6628995802389408,
-      "grad_norm": 0.5547932386398315,
-      "learning_rate": 0.0005007908630535503,
-      "loss": 3.665,
+      "epoch": 1.6657681940700808,
+      "grad_norm": 0.544411301612854,
+      "learning_rate": 0.0005006195358877495,
+      "loss": 3.6535,
       "step": 15450
     },
     {
-      "epoch": 1.6682811322785491,
-      "grad_norm": 0.5808051228523254,
-      "learning_rate": 0.0005004676220234888,
-      "loss": 3.6697,
+      "epoch": 1.6711590296495957,
+      "grad_norm": 0.6025227904319763,
+      "learning_rate": 0.0005002957366432812,
+      "loss": 3.6718,
       "step": 15500
     },
     {
-      "epoch": 1.6736626843181575,
-      "grad_norm": 0.6720907688140869,
-      "learning_rate": 0.0005001443809934273,
-      "loss": 3.666,
+      "epoch": 1.6765498652291106,
+      "grad_norm": 0.5519585013389587,
+      "learning_rate": 0.0004999719373988127,
+      "loss": 3.6706,
       "step": 15550
     },
     {
-      "epoch": 1.6790442363577656,
-      "grad_norm": 0.5560029149055481,
-      "learning_rate": 0.000499821139963366,
-      "loss": 3.66,
+      "epoch": 1.6819407008086253,
+      "grad_norm": 0.5432153344154358,
+      "learning_rate": 0.0004996481381543442,
+      "loss": 3.6583,
       "step": 15600
     },
     {
-      "epoch": 1.6844257883973737,
-      "grad_norm": 0.5708025693893433,
-      "learning_rate": 0.0004994978989333045,
-      "loss": 3.6729,
+      "epoch": 1.68733153638814,
+      "grad_norm": 0.569072961807251,
+      "learning_rate": 0.0004993243389098758,
+      "loss": 3.6736,
       "step": 15650
     },
     {
-      "epoch": 1.689807340436982,
-      "grad_norm": 0.6271914839744568,
-      "learning_rate": 0.0004991746579032431,
-      "loss": 3.6725,
+      "epoch": 1.692722371967655,
+      "grad_norm": 0.5824419260025024,
+      "learning_rate": 0.0004990070156502968,
+      "loss": 3.6625,
       "step": 15700
     },
     {
-      "epoch": 1.6951888924765903,
-      "grad_norm": 0.5649659037590027,
-      "learning_rate": 0.0004988514168731817,
-      "loss": 3.6506,
+      "epoch": 1.6981132075471699,
+      "grad_norm": 0.5825706720352173,
+      "learning_rate": 0.0004986832164058284,
+      "loss": 3.6896,
       "step": 15750
     },
     {
-      "epoch": 1.7005704445161984,
-      "grad_norm": 0.5502942204475403,
-      "learning_rate": 0.0004985281758431204,
-      "loss": 3.68,
+      "epoch": 1.7035040431266846,
+      "grad_norm": 0.5170381665229797,
+      "learning_rate": 0.0004983594171613599,
+      "loss": 3.6834,
       "step": 15800
     },
     {
-      "epoch": 1.7059519965558065,
-      "grad_norm": 0.6230396628379822,
-      "learning_rate": 0.0004982049348130589,
-      "loss": 3.6764,
+      "epoch": 1.7088948787061995,
+      "grad_norm": 0.5656709671020508,
+      "learning_rate": 0.0004980356179168915,
+      "loss": 3.6781,
       "step": 15850
     },
     {
-      "epoch": 1.7113335485954149,
-      "grad_norm": 0.5738170742988586,
-      "learning_rate": 0.0004978816937829975,
-      "loss": 3.6683,
+      "epoch": 1.7142857142857144,
+      "grad_norm": 0.5774093270301819,
+      "learning_rate": 0.000497711818672423,
+      "loss": 3.679,
       "step": 15900
     },
     {
-      "epoch": 1.7167151006350232,
-      "grad_norm": 0.5674835443496704,
-      "learning_rate": 0.000497558452752936,
-      "loss": 3.6641,
+      "epoch": 1.719676549865229,
+      "grad_norm": 0.5535004734992981,
+      "learning_rate": 0.0004973880194279547,
+      "loss": 3.664,
       "step": 15950
     },
     {
-      "epoch": 1.7220966526746313,
-      "grad_norm": 0.5687111616134644,
-      "learning_rate": 0.0004972416765434759,
-      "loss": 3.653,
+      "epoch": 1.7250673854447438,
+      "grad_norm": 0.6128283143043518,
+      "learning_rate": 0.0004970642201834862,
+      "loss": 3.6625,
       "step": 16000
     },
     {
-      "epoch": 1.7220966526746313,
-      "eval_accuracy": 0.35766868279343533,
-      "eval_loss": 3.6169068813323975,
-      "eval_runtime": 182.8519,
-      "eval_samples_per_second": 98.5,
-      "eval_steps_per_second": 6.158,
+      "epoch": 1.7250673854447438,
+      "eval_accuracy": 0.3567382881150799,
+      "eval_loss": 3.622256278991699,
+      "eval_runtime": 185.2729,
+      "eval_samples_per_second": 97.213,
+      "eval_steps_per_second": 6.078,
       "step": 16000
     },
     {
-      "epoch": 1.7274782047142396,
-      "grad_norm": 0.5726053714752197,
-      "learning_rate": 0.0004969184355134145,
-      "loss": 3.6704,
+      "epoch": 1.7304582210242587,
+      "grad_norm": 0.5429639220237732,
+      "learning_rate": 0.0004967404209390178,
+      "loss": 3.6633,
       "step": 16050
     },
     {
-      "epoch": 1.732859756753848,
-      "grad_norm": 0.5218805074691772,
-      "learning_rate": 0.0004965951944833531,
-      "loss": 3.6687,
+      "epoch": 1.7358490566037736,
+      "grad_norm": 0.5845425128936768,
+      "learning_rate": 0.0004964166216945493,
+      "loss": 3.6505,
       "step": 16100
     },
     {
-      "epoch": 1.738241308793456,
-      "grad_norm": 0.5829153060913086,
-      "learning_rate": 0.0004962719534532916,
-      "loss": 3.6701,
+      "epoch": 1.7412398921832883,
+      "grad_norm": 0.519659698009491,
+      "learning_rate": 0.000496092822450081,
+      "loss": 3.6709,
       "step": 16150
     },
     {
-      "epoch": 1.7436228608330642,
-      "grad_norm": 0.6053098440170288,
-      "learning_rate": 0.0004959487124232302,
-      "loss": 3.6489,
+      "epoch": 1.7466307277628033,
+      "grad_norm": 0.5584535002708435,
+      "learning_rate": 0.0004957690232056125,
+      "loss": 3.6862,
       "step": 16200
     },
     {
-      "epoch": 1.7490044128726725,
-      "grad_norm": 0.5329374670982361,
-      "learning_rate": 0.0004956254713931688,
-      "loss": 3.6494,
+      "epoch": 1.7520215633423182,
+      "grad_norm": 0.5469822883605957,
+      "learning_rate": 0.0004954452239611441,
+      "loss": 3.6593,
       "step": 16250
     },
     {
-      "epoch": 1.7543859649122808,
-      "grad_norm": 0.5787749886512756,
-      "learning_rate": 0.0004953022303631074,
-      "loss": 3.6523,
+      "epoch": 1.7574123989218329,
+      "grad_norm": 0.5633559823036194,
+      "learning_rate": 0.0004951214247166756,
+      "loss": 3.6798,
       "step": 16300
     },
     {
-      "epoch": 1.759767516951889,
-      "grad_norm": 0.5446376800537109,
-      "learning_rate": 0.0004949789893330459,
-      "loss": 3.6551,
+      "epoch": 1.7628032345013476,
+      "grad_norm": 0.593555748462677,
+      "learning_rate": 0.0004947976254722072,
+      "loss": 3.6635,
       "step": 16350
     },
     {
-      "epoch": 1.765149068991497,
-      "grad_norm": 0.6261696815490723,
-      "learning_rate": 0.0004946557483029846,
-      "loss": 3.6529,
+      "epoch": 1.7681940700808625,
+      "grad_norm": 0.5291847586631775,
+      "learning_rate": 0.0004944738262277387,
+      "loss": 3.6539,
       "step": 16400
     },
     {
-      "epoch": 1.7705306210311054,
-      "grad_norm": 0.5923592448234558,
-      "learning_rate": 0.0004943325072729231,
-      "loss": 3.6742,
+      "epoch": 1.7735849056603774,
+      "grad_norm": 0.5443708300590515,
+      "learning_rate": 0.0004941500269832703,
+      "loss": 3.66,
       "step": 16450
     },
     {
-      "epoch": 1.7759121730707137,
-      "grad_norm": 0.5514686703681946,
-      "learning_rate": 0.0004940092662428617,
-      "loss": 3.6655,
+      "epoch": 1.778975741239892,
+      "grad_norm": 0.5486942529678345,
+      "learning_rate": 0.0004938262277388019,
+      "loss": 3.6489,
       "step": 16500
     },
     {
-      "epoch": 1.7812937251103218,
-      "grad_norm": 0.5481582283973694,
-      "learning_rate": 0.0004936860252128003,
-      "loss": 3.6613,
+      "epoch": 1.784366576819407,
+      "grad_norm": 0.6374784708023071,
+      "learning_rate": 0.0004935024284943335,
+      "loss": 3.676,
       "step": 16550
     },
     {
-      "epoch": 1.78667527714993,
-      "grad_norm": 0.5708692073822021,
-      "learning_rate": 0.0004933627841827388,
-      "loss": 3.6644,
+      "epoch": 1.789757412398922,
+      "grad_norm": 0.6883034706115723,
+      "learning_rate": 0.000493178629249865,
+      "loss": 3.6612,
       "step": 16600
     },
     {
-      "epoch": 1.7920568291895382,
-      "grad_norm": 0.5547276139259338,
-      "learning_rate": 0.0004930395431526775,
-      "loss": 3.6443,
+      "epoch": 1.7951482479784366,
+      "grad_norm": 0.5932786464691162,
+      "learning_rate": 0.0004928548300053966,
+      "loss": 3.6656,
       "step": 16650
     },
     {
-      "epoch": 1.7974383812291466,
-      "grad_norm": 0.60475754737854,
-      "learning_rate": 0.0004927163021226161,
-      "loss": 3.666,
+      "epoch": 1.8005390835579513,
+      "grad_norm": 0.5369443297386169,
+      "learning_rate": 0.0004925310307609282,
+      "loss": 3.6662,
       "step": 16700
     },
     {
-      "epoch": 1.8028199332687547,
-      "grad_norm": 0.568871796131134,
-      "learning_rate": 0.0004923930610925547,
-      "loss": 3.6573,
+      "epoch": 1.8059299191374663,
+      "grad_norm": 0.5238076448440552,
+      "learning_rate": 0.0004922072315164598,
+      "loss": 3.6535,
       "step": 16750
     },
     {
-      "epoch": 1.8082014853083628,
-      "grad_norm": 0.5564674139022827,
-      "learning_rate": 0.0004920698200624932,
-      "loss": 3.6515,
+      "epoch": 1.8113207547169812,
+      "grad_norm": 0.500116765499115,
+      "learning_rate": 0.0004918834322719913,
+      "loss": 3.6552,
       "step": 16800
     },
     {
-      "epoch": 1.813583037347971,
-      "grad_norm": 0.5341878533363342,
-      "learning_rate": 0.0004917465790324317,
-      "loss": 3.6652,
+      "epoch": 1.8167115902964959,
+      "grad_norm": 0.580346941947937,
+      "learning_rate": 0.0004915596330275229,
+      "loss": 3.6617,
       "step": 16850
     },
     {
-      "epoch": 1.8189645893875794,
-      "grad_norm": 0.58695387840271,
-      "learning_rate": 0.0004914233380023704,
-      "loss": 3.6546,
+      "epoch": 1.8221024258760108,
+      "grad_norm": 0.5278408527374268,
+      "learning_rate": 0.0004912423097679439,
+      "loss": 3.6465,
       "step": 16900
     },
     {
-      "epoch": 1.8243461414271875,
-      "grad_norm": 0.5584468245506287,
-      "learning_rate": 0.0004911000969723089,
-      "loss": 3.631,
+      "epoch": 1.8274932614555257,
+      "grad_norm": 0.5278293490409851,
+      "learning_rate": 0.0004909185105234754,
+      "loss": 3.6773,
       "step": 16950
     },
     {
-      "epoch": 1.8297276934667959,
-      "grad_norm": 0.5749080777168274,
-      "learning_rate": 0.0004907768559422476,
-      "loss": 3.6484,
+      "epoch": 1.8328840970350404,
+      "grad_norm": 0.5422157645225525,
+      "learning_rate": 0.000490594711279007,
+      "loss": 3.6619,
       "step": 17000
     },
     {
-      "epoch": 1.8297276934667959,
-      "eval_accuracy": 0.35881909956961505,
-      "eval_loss": 3.601506233215332,
-      "eval_runtime": 182.9699,
-      "eval_samples_per_second": 98.437,
-      "eval_steps_per_second": 6.154,
+      "epoch": 1.8328840970350404,
+      "eval_accuracy": 0.35810253377447854,
+      "eval_loss": 3.610804796218872,
+      "eval_runtime": 184.8314,
+      "eval_samples_per_second": 97.446,
+      "eval_steps_per_second": 6.092,
       "step": 17000
     },
     {
-      "epoch": 1.8351092455064042,
-      "grad_norm": 0.5957604050636292,
-      "learning_rate": 0.0004904536149121861,
-      "loss": 3.6388,
+      "epoch": 1.838274932614555,
+      "grad_norm": 0.531470775604248,
+      "learning_rate": 0.0004902709120345385,
+      "loss": 3.6441,
       "step": 17050
     },
     {
-      "epoch": 1.8404907975460123,
-      "grad_norm": 0.5906072854995728,
-      "learning_rate": 0.0004901303738821248,
-      "loss": 3.6523,
+      "epoch": 1.8436657681940702,
+      "grad_norm": 0.5433018803596497,
+      "learning_rate": 0.00048994711279007,
+      "loss": 3.6581,
       "step": 17100
     },
     {
-      "epoch": 1.8458723495856204,
-      "grad_norm": 0.6116198301315308,
-      "learning_rate": 0.0004898071328520633,
-      "loss": 3.6424,
+      "epoch": 1.849056603773585,
+      "grad_norm": 0.530763566493988,
+      "learning_rate": 0.0004896233135456017,
+      "loss": 3.6545,
       "step": 17150
     },
     {
-      "epoch": 1.8512539016252287,
-      "grad_norm": 0.5727099776268005,
-      "learning_rate": 0.0004894838918220019,
-      "loss": 3.6625,
+      "epoch": 1.8544474393530996,
+      "grad_norm": 0.5608590245246887,
+      "learning_rate": 0.0004892995143011333,
+      "loss": 3.6389,
       "step": 17200
     },
     {
-      "epoch": 1.856635453664837,
-      "grad_norm": 0.5920504331588745,
-      "learning_rate": 0.0004891606507919405,
-      "loss": 3.6366,
+      "epoch": 1.8598382749326146,
+      "grad_norm": 0.6097438335418701,
+      "learning_rate": 0.0004889757150566648,
+      "loss": 3.6605,
       "step": 17250
     },
     {
-      "epoch": 1.8620170057044452,
-      "grad_norm": 0.6015959978103638,
-      "learning_rate": 0.000488837409761879,
-      "loss": 3.6506,
+      "epoch": 1.8652291105121295,
+      "grad_norm": 0.6334707736968994,
+      "learning_rate": 0.0004886519158121964,
+      "loss": 3.6499,
       "step": 17300
     },
     {
-      "epoch": 1.8673985577440533,
-      "grad_norm": 0.5476948022842407,
-      "learning_rate": 0.0004885141687318177,
-      "loss": 3.654,
+      "epoch": 1.8706199460916442,
+      "grad_norm": 0.6429612040519714,
+      "learning_rate": 0.000488328116567728,
+      "loss": 3.6542,
       "step": 17350
     },
     {
-      "epoch": 1.8727801097836616,
-      "grad_norm": 0.5550304651260376,
-      "learning_rate": 0.00048819092770175623,
-      "loss": 3.6382,
+      "epoch": 1.8760107816711589,
+      "grad_norm": 0.6209458708763123,
+      "learning_rate": 0.0004880043173232595,
+      "loss": 3.6473,
       "step": 17400
     },
     {
-      "epoch": 1.87816166182327,
-      "grad_norm": 0.5779985189437866,
-      "learning_rate": 0.0004878676866716948,
-      "loss": 3.6562,
+      "epoch": 1.881401617250674,
+      "grad_norm": 0.5378085374832153,
+      "learning_rate": 0.0004876805180787911,
+      "loss": 3.6667,
       "step": 17450
     },
     {
-      "epoch": 1.883543213862878,
-      "grad_norm": 0.5878141522407532,
-      "learning_rate": 0.00048754444564163337,
-      "loss": 3.6424,
+      "epoch": 1.8867924528301887,
+      "grad_norm": 0.5677717924118042,
+      "learning_rate": 0.0004873567188343227,
+      "loss": 3.6589,
       "step": 17500
     },
     {
-      "epoch": 1.8889247659024861,
-      "grad_norm": 0.5703890323638916,
-      "learning_rate": 0.000487221204611572,
-      "loss": 3.6491,
+      "epoch": 1.8921832884097034,
+      "grad_norm": 0.5877004861831665,
+      "learning_rate": 0.0004870329195898542,
+      "loss": 3.6427,
       "step": 17550
     },
     {
-      "epoch": 1.8943063179420945,
-      "grad_norm": 0.5712037086486816,
-      "learning_rate": 0.00048689796358151056,
-      "loss": 3.6276,
+      "epoch": 1.8975741239892183,
+      "grad_norm": 0.5844641923904419,
+      "learning_rate": 0.00048670912034538583,
+      "loss": 3.6495,
       "step": 17600
     },
     {
-      "epoch": 1.8996878699817028,
-      "grad_norm": 0.5311342477798462,
-      "learning_rate": 0.00048657472255144915,
-      "loss": 3.6438,
+      "epoch": 1.9029649595687332,
+      "grad_norm": 0.6069989204406738,
+      "learning_rate": 0.0004863853211009174,
+      "loss": 3.6266,
       "step": 17650
     },
     {
-      "epoch": 1.905069422021311,
-      "grad_norm": 0.5727416276931763,
-      "learning_rate": 0.00048625148152138775,
-      "loss": 3.6264,
+      "epoch": 1.908355795148248,
+      "grad_norm": 0.5461699962615967,
+      "learning_rate": 0.000486061521856449,
+      "loss": 3.6356,
       "step": 17700
     },
     {
-      "epoch": 1.910450974060919,
-      "grad_norm": 0.6044231653213501,
-      "learning_rate": 0.0004859282404913263,
-      "loss": 3.6242,
+      "epoch": 1.9137466307277629,
+      "grad_norm": 0.6169970035552979,
+      "learning_rate": 0.00048573772261198054,
+      "loss": 3.6391,
       "step": 17750
     },
     {
-      "epoch": 1.9158325261005273,
-      "grad_norm": 0.5727190971374512,
-      "learning_rate": 0.0004856049994612649,
-      "loss": 3.6462,
+      "epoch": 1.9191374663072778,
+      "grad_norm": 0.5463764071464539,
+      "learning_rate": 0.00048541392336751214,
+      "loss": 3.658,
       "step": 17800
     },
     {
-      "epoch": 1.9212140781401357,
-      "grad_norm": 0.5734631419181824,
-      "learning_rate": 0.00048528175843120353,
-      "loss": 3.6472,
+      "epoch": 1.9245283018867925,
+      "grad_norm": 0.5543783903121948,
+      "learning_rate": 0.0004850901241230437,
+      "loss": 3.6572,
       "step": 17850
     },
     {
-      "epoch": 1.9265956301797438,
-      "grad_norm": 0.5388041734695435,
-      "learning_rate": 0.0004849585174011421,
-      "loss": 3.6489,
+      "epoch": 1.9299191374663072,
+      "grad_norm": 0.5729919672012329,
+      "learning_rate": 0.0004847663248785753,
+      "loss": 3.6619,
       "step": 17900
     },
     {
-      "epoch": 1.931977182219352,
-      "grad_norm": 0.5606278777122498,
-      "learning_rate": 0.00048463527637108067,
-      "loss": 3.6294,
+      "epoch": 1.935309973045822,
+      "grad_norm": 0.5639930963516235,
+      "learning_rate": 0.0004844425256341068,
+      "loss": 3.6524,
       "step": 17950
     },
     {
-      "epoch": 1.9373587342589604,
-      "grad_norm": 0.5878045558929443,
-      "learning_rate": 0.0004843120353410192,
-      "loss": 3.6472,
+      "epoch": 1.940700808625337,
+      "grad_norm": 0.5668043494224548,
+      "learning_rate": 0.00048411872638963834,
+      "loss": 3.6364,
       "step": 18000
     },
     {
-      "epoch": 1.9373587342589604,
-      "eval_accuracy": 0.36079995036736084,
-      "eval_loss": 3.5877466201782227,
-      "eval_runtime": 182.9553,
-      "eval_samples_per_second": 98.445,
-      "eval_steps_per_second": 6.155,
+      "epoch": 1.940700808625337,
+      "eval_accuracy": 0.3596066157004944,
+      "eval_loss": 3.5932109355926514,
+      "eval_runtime": 185.1478,
+      "eval_samples_per_second": 97.279,
+      "eval_steps_per_second": 6.082,
       "step": 18000
     },
     {
-      "epoch": 1.9427402862985685,
-      "grad_norm": 0.5944440960884094,
-      "learning_rate": 0.00048399525913155907,
-      "loss": 3.6349,
+      "epoch": 1.9460916442048517,
+      "grad_norm": 0.6137953996658325,
+      "learning_rate": 0.00048379492714516995,
+      "loss": 3.6263,
       "step": 18050
     },
     {
-      "epoch": 1.9481218383381766,
-      "grad_norm": 0.6114388704299927,
-      "learning_rate": 0.0004836720181014976,
-      "loss": 3.6319,
+      "epoch": 1.9514824797843666,
+      "grad_norm": 0.603476881980896,
+      "learning_rate": 0.0004834711279007015,
+      "loss": 3.6348,
       "step": 18100
     },
     {
-      "epoch": 1.953503390377785,
-      "grad_norm": 0.5699650645256042,
-      "learning_rate": 0.0004833487770714362,
-      "loss": 3.6309,
+      "epoch": 1.9568733153638815,
+      "grad_norm": 0.5930576920509338,
+      "learning_rate": 0.0004831473286562331,
+      "loss": 3.6418,
       "step": 18150
     },
     {
-      "epoch": 1.9588849424173933,
-      "grad_norm": 0.5740841031074524,
-      "learning_rate": 0.00048302553604137485,
-      "loss": 3.6278,
+      "epoch": 1.9622641509433962,
+      "grad_norm": 0.5806794762611389,
+      "learning_rate": 0.00048282352941176465,
+      "loss": 3.6427,
       "step": 18200
     },
     {
-      "epoch": 1.9642664944570014,
-      "grad_norm": 0.6211947798728943,
-      "learning_rate": 0.0004827022950113134,
-      "loss": 3.6346,
+      "epoch": 1.967654986522911,
+      "grad_norm": 0.5229198932647705,
+      "learning_rate": 0.00048249973016729626,
+      "loss": 3.6336,
       "step": 18250
     },
     {
-      "epoch": 1.9696480464966095,
-      "grad_norm": 0.5401277542114258,
-      "learning_rate": 0.000482379053981252,
-      "loss": 3.6603,
+      "epoch": 1.9730458221024259,
+      "grad_norm": 0.5625728964805603,
+      "learning_rate": 0.0004821759309228278,
+      "loss": 3.6622,
       "step": 18300
     },
     {
-      "epoch": 1.9750295985362178,
-      "grad_norm": 0.5557017922401428,
-      "learning_rate": 0.0004820558129511906,
-      "loss": 3.629,
+      "epoch": 1.9784366576819408,
+      "grad_norm": 0.577113151550293,
+      "learning_rate": 0.00048185213167835936,
+      "loss": 3.6386,
       "step": 18350
     },
     {
-      "epoch": 1.9804111505758262,
-      "grad_norm": 0.592074453830719,
-      "learning_rate": 0.0004817325719211291,
-      "loss": 3.6235,
+      "epoch": 1.9838274932614555,
+      "grad_norm": 0.5825108289718628,
+      "learning_rate": 0.00048152833243389096,
+      "loss": 3.6421,
       "step": 18400
     },
     {
-      "epoch": 1.9857927026154343,
-      "grad_norm": 0.600532591342926,
-      "learning_rate": 0.0004814093308910677,
-      "loss": 3.6253,
+      "epoch": 1.9892183288409704,
+      "grad_norm": 0.6418988108634949,
+      "learning_rate": 0.0004812045331894225,
+      "loss": 3.6495,
       "step": 18450
     },
     {
-      "epoch": 1.9911742546550424,
-      "grad_norm": 0.4943503141403198,
-      "learning_rate": 0.00048108608986100637,
-      "loss": 3.6225,
+      "epoch": 1.9946091644204853,
+      "grad_norm": 0.6308207511901855,
+      "learning_rate": 0.0004808807339449541,
+      "loss": 3.6288,
       "step": 18500
     },
     {
-      "epoch": 1.9965558066946507,
-      "grad_norm": 0.5403441190719604,
-      "learning_rate": 0.0004807628488309449,
-      "loss": 3.6333,
+      "epoch": 2.0,
+      "grad_norm": 1.1038312911987305,
+      "learning_rate": 0.0004805569347004856,
+      "loss": 3.6314,
       "step": 18550
     },
     {
-      "epoch": 2.001937358734259,
-      "grad_norm": 0.5593751668930054,
-      "learning_rate": 0.0004804396078008835,
-      "loss": 3.5902,
+      "epoch": 2.0053908355795147,
+      "grad_norm": 0.5954450368881226,
+      "learning_rate": 0.0004802331354560173,
+      "loss": 3.5285,
       "step": 18600
     },
     {
-      "epoch": 2.007318910773867,
-      "grad_norm": 0.5596261024475098,
-      "learning_rate": 0.00048011636677082204,
-      "loss": 3.5438,
+      "epoch": 2.01078167115903,
+      "grad_norm": 0.5570014119148254,
+      "learning_rate": 0.00047990933621154877,
+      "loss": 3.5394,
       "step": 18650
     },
     {
-      "epoch": 2.0127004628134753,
-      "grad_norm": 0.5620816946029663,
-      "learning_rate": 0.00047979312574076064,
-      "loss": 3.5367,
+      "epoch": 2.0161725067385445,
+      "grad_norm": 0.6168981194496155,
+      "learning_rate": 0.0004795855369670804,
+      "loss": 3.5386,
       "step": 18700
     },
     {
-      "epoch": 2.018082014853084,
-      "grad_norm": 0.6553834080696106,
-      "learning_rate": 0.0004794698847106992,
-      "loss": 3.5336,
+      "epoch": 2.0215633423180592,
+      "grad_norm": 0.5821147561073303,
+      "learning_rate": 0.0004792617377226119,
+      "loss": 3.5445,
       "step": 18750
     },
     {
-      "epoch": 2.023463566892692,
-      "grad_norm": 0.5942727327346802,
-      "learning_rate": 0.0004791466436806378,
-      "loss": 3.5532,
+      "epoch": 2.026954177897574,
+      "grad_norm": 0.6035953164100647,
+      "learning_rate": 0.0004789379384781435,
+      "loss": 3.566,
       "step": 18800
     },
     {
-      "epoch": 2.0288451189323,
-      "grad_norm": 0.563507616519928,
-      "learning_rate": 0.0004788234026505764,
-      "loss": 3.565,
+      "epoch": 2.032345013477089,
+      "grad_norm": 0.547264039516449,
+      "learning_rate": 0.0004786141392336751,
+      "loss": 3.563,
       "step": 18850
     },
     {
-      "epoch": 2.034226670971908,
-      "grad_norm": 0.5451712608337402,
-      "learning_rate": 0.00047850016162051496,
-      "loss": 3.55,
+      "epoch": 2.0377358490566038,
+      "grad_norm": 0.5701025724411011,
+      "learning_rate": 0.00047829033998920663,
+      "loss": 3.5472,
       "step": 18900
     },
     {
-      "epoch": 2.0396082230115167,
-      "grad_norm": 0.6260777711868286,
-      "learning_rate": 0.00047817692059045356,
-      "loss": 3.5395,
+      "epoch": 2.0431266846361185,
+      "grad_norm": 0.5930619239807129,
+      "learning_rate": 0.00047796654074473824,
+      "loss": 3.5514,
       "step": 18950
     },
     {
-      "epoch": 2.044989775051125,
-      "grad_norm": 0.6237614750862122,
-      "learning_rate": 0.00047785367956039215,
-      "loss": 3.5601,
+      "epoch": 2.0485175202156336,
+      "grad_norm": 0.5958006978034973,
+      "learning_rate": 0.0004776427415002698,
+      "loss": 3.5822,
       "step": 19000
     },
     {
-      "epoch": 2.044989775051125,
-      "eval_accuracy": 0.36222699739591624,
-      "eval_loss": 3.5780997276306152,
-      "eval_runtime": 182.9631,
-      "eval_samples_per_second": 98.441,
-      "eval_steps_per_second": 6.154,
+      "epoch": 2.0485175202156336,
+      "eval_accuracy": 0.3610058475897693,
+      "eval_loss": 3.584773540496826,
+      "eval_runtime": 184.7103,
+      "eval_samples_per_second": 97.509,
+      "eval_steps_per_second": 6.096,
       "step": 19000
     },
     {
-      "epoch": 2.050371327090733,
-      "grad_norm": 0.5645211935043335,
-      "learning_rate": 0.00047753043853033075,
-      "loss": 3.5712,
+      "epoch": 2.0539083557951483,
+      "grad_norm": 0.5540789365768433,
+      "learning_rate": 0.0004773189422558014,
+      "loss": 3.5571,
       "step": 19050
     },
     {
-      "epoch": 2.055752879130341,
-      "grad_norm": 0.550080418586731,
-      "learning_rate": 0.00047720719750026934,
+      "epoch": 2.059299191374663,
+      "grad_norm": 0.6005356907844543,
+      "learning_rate": 0.00047699514301133294,
       "loss": 3.5519,
       "step": 19100
     },
     {
-      "epoch": 2.0611344311699495,
-      "grad_norm": 0.5775645971298218,
-      "learning_rate": 0.00047688395647020793,
-      "loss": 3.5345,
+      "epoch": 2.0646900269541777,
+      "grad_norm": 0.6260417103767395,
+      "learning_rate": 0.00047667134376686455,
+      "loss": 3.5664,
       "step": 19150
     },
     {
-      "epoch": 2.0665159832095576,
-      "grad_norm": 0.6265087127685547,
-      "learning_rate": 0.0004765607154401465,
-      "loss": 3.5604,
+      "epoch": 2.070080862533693,
+      "grad_norm": 0.6222960948944092,
+      "learning_rate": 0.0004763475445223961,
+      "loss": 3.5777,
       "step": 19200
     },
     {
-      "epoch": 2.0718975352491658,
-      "grad_norm": 0.645406186580658,
-      "learning_rate": 0.00047623747441008507,
-      "loss": 3.5486,
+      "epoch": 2.0754716981132075,
+      "grad_norm": 0.5388205647468567,
+      "learning_rate": 0.0004760237452779276,
+      "loss": 3.5605,
       "step": 19250
     },
     {
-      "epoch": 2.0772790872887743,
-      "grad_norm": 0.5847898721694946,
-      "learning_rate": 0.0004759142333800236,
-      "loss": 3.5457,
+      "epoch": 2.0808625336927222,
+      "grad_norm": 0.6028093099594116,
+      "learning_rate": 0.0004756999460334592,
+      "loss": 3.5498,
       "step": 19300
     },
     {
-      "epoch": 2.0826606393283824,
-      "grad_norm": 0.5650346875190735,
-      "learning_rate": 0.00047559099234996226,
-      "loss": 3.5555,
+      "epoch": 2.0862533692722374,
+      "grad_norm": 0.646430492401123,
+      "learning_rate": 0.00047537614678899075,
+      "loss": 3.5661,
       "step": 19350
     },
     {
-      "epoch": 2.0880421913679905,
-      "grad_norm": 0.599783182144165,
-      "learning_rate": 0.00047526775131990085,
-      "loss": 3.5459,
+      "epoch": 2.091644204851752,
+      "grad_norm": 0.5180575251579285,
+      "learning_rate": 0.00047505234754452235,
+      "loss": 3.5445,
       "step": 19400
     },
     {
-      "epoch": 2.0934237434075986,
-      "grad_norm": 0.5775169730186462,
-      "learning_rate": 0.0004749445102898394,
-      "loss": 3.5612,
+      "epoch": 2.0970350404312668,
+      "grad_norm": 0.5926779508590698,
+      "learning_rate": 0.0004747285483000539,
+      "loss": 3.5611,
       "step": 19450
     },
     {
-      "epoch": 2.098805295447207,
-      "grad_norm": 0.6146634817123413,
-      "learning_rate": 0.000474621269259778,
-      "loss": 3.5666,
+      "epoch": 2.1024258760107815,
+      "grad_norm": 0.5802668929100037,
+      "learning_rate": 0.0004744047490555855,
+      "loss": 3.5625,
       "step": 19500
     },
     {
-      "epoch": 2.1041868474868153,
-      "grad_norm": 0.6270310878753662,
-      "learning_rate": 0.0004742980282297166,
-      "loss": 3.5408,
+      "epoch": 2.1078167115902966,
+      "grad_norm": 0.6308825016021729,
+      "learning_rate": 0.00047408094981111706,
+      "loss": 3.5505,
       "step": 19550
     },
     {
-      "epoch": 2.1095683995264234,
-      "grad_norm": 0.517265796661377,
-      "learning_rate": 0.0004739747871996551,
-      "loss": 3.5576,
+      "epoch": 2.1132075471698113,
+      "grad_norm": 0.5490583181381226,
+      "learning_rate": 0.00047375715056664866,
+      "loss": 3.5526,
       "step": 19600
     },
     {
-      "epoch": 2.1149499515660315,
-      "grad_norm": 0.6480297446250916,
-      "learning_rate": 0.00047365154616959377,
-      "loss": 3.542,
+      "epoch": 2.118598382749326,
+      "grad_norm": 0.5867042541503906,
+      "learning_rate": 0.0004734333513221802,
+      "loss": 3.5601,
       "step": 19650
     },
     {
-      "epoch": 2.12033150360564,
-      "grad_norm": 0.5682874321937561,
-      "learning_rate": 0.00047332830513953237,
-      "loss": 3.5669,
+      "epoch": 2.123989218328841,
+      "grad_norm": 0.5740089416503906,
+      "learning_rate": 0.00047310955207771177,
+      "loss": 3.5641,
       "step": 19700
     },
     {
-      "epoch": 2.125713055645248,
-      "grad_norm": 0.6224113702774048,
-      "learning_rate": 0.0004730050641094709,
-      "loss": 3.5485,
+      "epoch": 2.129380053908356,
+      "grad_norm": 0.5596190690994263,
+      "learning_rate": 0.00047278575283324337,
+      "loss": 3.5679,
       "step": 19750
     },
     {
-      "epoch": 2.1310946076848563,
-      "grad_norm": 0.554499089717865,
-      "learning_rate": 0.0004726818230794095,
-      "loss": 3.5338,
+      "epoch": 2.1347708894878705,
+      "grad_norm": 0.6093752980232239,
+      "learning_rate": 0.0004724619535887749,
+      "loss": 3.5522,
       "step": 19800
     },
     {
-      "epoch": 2.1364761597244644,
-      "grad_norm": 0.6027452349662781,
-      "learning_rate": 0.0004723650468699493,
-      "loss": 3.5481,
+      "epoch": 2.1401617250673857,
+      "grad_norm": 0.581826388835907,
+      "learning_rate": 0.0004721381543443065,
+      "loss": 3.5798,
       "step": 19850
     },
     {
-      "epoch": 2.141857711764073,
-      "grad_norm": 0.6047434210777283,
-      "learning_rate": 0.0004720418058398879,
-      "loss": 3.5465,
+      "epoch": 2.1455525606469004,
+      "grad_norm": 0.5723430514335632,
+      "learning_rate": 0.000471814355099838,
+      "loss": 3.5818,
       "step": 19900
     },
     {
-      "epoch": 2.147239263803681,
-      "grad_norm": 0.5787104964256287,
-      "learning_rate": 0.00047171856480982644,
-      "loss": 3.5562,
+      "epoch": 2.150943396226415,
+      "grad_norm": 0.6087849736213684,
+      "learning_rate": 0.0004714905558553697,
+      "loss": 3.5759,
       "step": 19950
     },
     {
-      "epoch": 2.152620815843289,
-      "grad_norm": 0.5642232298851013,
-      "learning_rate": 0.0004713953237797651,
-      "loss": 3.5694,
+      "epoch": 2.1563342318059298,
+      "grad_norm": 0.5303805470466614,
+      "learning_rate": 0.0004711667566109012,
+      "loss": 3.5592,
       "step": 20000
     },
     {
-      "epoch": 2.152620815843289,
-      "eval_accuracy": 0.36314848253747084,
-      "eval_loss": 3.5687355995178223,
-      "eval_runtime": 179.7649,
-      "eval_samples_per_second": 100.192,
-      "eval_steps_per_second": 6.264,
+      "epoch": 2.1563342318059298,
+      "eval_accuracy": 0.3620970485420901,
+      "eval_loss": 3.578047037124634,
+      "eval_runtime": 185.1715,
+      "eval_samples_per_second": 97.267,
+      "eval_steps_per_second": 6.081,
       "step": 20000
     },
     {
-      "epoch": 2.1580023678828972,
-      "grad_norm": 0.6349261999130249,
-      "learning_rate": 0.0004710720827497037,
-      "loss": 3.5747,
+      "epoch": 2.161725067385445,
+      "grad_norm": 0.6045820116996765,
+      "learning_rate": 0.00047084295736643273,
+      "loss": 3.5599,
       "step": 20050
     },
     {
-      "epoch": 2.163383919922506,
-      "grad_norm": 0.5958331823348999,
-      "learning_rate": 0.00047074884171964223,
-      "loss": 3.5547,
+      "epoch": 2.1671159029649596,
+      "grad_norm": 0.6460957527160645,
+      "learning_rate": 0.00047051915812196433,
+      "loss": 3.5676,
       "step": 20100
     },
     {
-      "epoch": 2.168765471962114,
-      "grad_norm": 0.6347177028656006,
-      "learning_rate": 0.0004704256006895808,
-      "loss": 3.561,
+      "epoch": 2.1725067385444743,
+      "grad_norm": 0.6051047444343567,
+      "learning_rate": 0.0004701953588774959,
+      "loss": 3.5624,
       "step": 20150
     },
     {
-      "epoch": 2.174147024001722,
-      "grad_norm": 0.6787843108177185,
-      "learning_rate": 0.00047010235965951936,
-      "loss": 3.5584,
+      "epoch": 2.177897574123989,
+      "grad_norm": 0.5932462215423584,
+      "learning_rate": 0.0004698715596330275,
+      "loss": 3.5778,
       "step": 20200
     },
     {
-      "epoch": 2.1795285760413305,
-      "grad_norm": 0.5568081140518188,
-      "learning_rate": 0.00046977911862945796,
-      "loss": 3.5667,
+      "epoch": 2.183288409703504,
+      "grad_norm": 0.5538517236709595,
+      "learning_rate": 0.00046954776038855904,
+      "loss": 3.5628,
       "step": 20250
     },
     {
-      "epoch": 2.1849101280809387,
-      "grad_norm": 0.6003414988517761,
-      "learning_rate": 0.0004694558775993966,
-      "loss": 3.5515,
+      "epoch": 2.188679245283019,
+      "grad_norm": 0.5807755589485168,
+      "learning_rate": 0.00046922396114409064,
+      "loss": 3.5691,
       "step": 20300
     },
     {
-      "epoch": 2.1902916801205468,
-      "grad_norm": 0.5988909602165222,
-      "learning_rate": 0.00046913263656933515,
-      "loss": 3.5563,
+      "epoch": 2.1940700808625335,
+      "grad_norm": 0.5965390801429749,
+      "learning_rate": 0.0004689001618996222,
+      "loss": 3.55,
       "step": 20350
     },
     {
-      "epoch": 2.195673232160155,
-      "grad_norm": 0.6274018287658691,
-      "learning_rate": 0.00046880939553927374,
-      "loss": 3.55,
+      "epoch": 2.1994609164420487,
+      "grad_norm": 0.5739687085151672,
+      "learning_rate": 0.0004685763626551538,
+      "loss": 3.5578,
       "step": 20400
     },
     {
-      "epoch": 2.2010547841997634,
-      "grad_norm": 0.5922993421554565,
-      "learning_rate": 0.00046848615450921234,
-      "loss": 3.5385,
+      "epoch": 2.2048517520215634,
+      "grad_norm": 0.5837062001228333,
+      "learning_rate": 0.00046825256341068535,
+      "loss": 3.5594,
       "step": 20450
     },
     {
-      "epoch": 2.2064363362393715,
-      "grad_norm": 0.6406957507133484,
-      "learning_rate": 0.0004681629134791509,
-      "loss": 3.5292,
+      "epoch": 2.210242587601078,
+      "grad_norm": 0.5664048790931702,
+      "learning_rate": 0.0004679287641662169,
+      "loss": 3.5722,
       "step": 20500
     },
     {
-      "epoch": 2.2118178882789796,
-      "grad_norm": 0.5569520592689514,
-      "learning_rate": 0.00046783967244908947,
-      "loss": 3.5393,
+      "epoch": 2.215633423180593,
+      "grad_norm": 0.5342917442321777,
+      "learning_rate": 0.0004676049649217485,
+      "loss": 3.5671,
       "step": 20550
     },
     {
-      "epoch": 2.2171994403185877,
-      "grad_norm": 0.5793364644050598,
-      "learning_rate": 0.0004675164314190281,
-      "loss": 3.5537,
+      "epoch": 2.221024258760108,
+      "grad_norm": 0.5454385876655579,
+      "learning_rate": 0.00046728116567728,
+      "loss": 3.56,
       "step": 20600
     },
     {
-      "epoch": 2.2225809923581963,
-      "grad_norm": 0.5888959765434265,
-      "learning_rate": 0.00046719319038896666,
-      "loss": 3.5403,
+      "epoch": 2.2264150943396226,
+      "grad_norm": 0.5746397972106934,
+      "learning_rate": 0.0004669573664328116,
+      "loss": 3.5615,
       "step": 20650
     },
     {
-      "epoch": 2.2279625443978044,
-      "grad_norm": 0.5922018885612488,
-      "learning_rate": 0.00046686994935890526,
-      "loss": 3.547,
+      "epoch": 2.2318059299191373,
+      "grad_norm": 0.6558805108070374,
+      "learning_rate": 0.00046663356718834316,
+      "loss": 3.5595,
       "step": 20700
     },
     {
-      "epoch": 2.2333440964374125,
-      "grad_norm": 0.6159331798553467,
-      "learning_rate": 0.0004665467083288438,
-      "loss": 3.5509,
+      "epoch": 2.2371967654986524,
+      "grad_norm": 0.5640301704406738,
+      "learning_rate": 0.00046630976794387476,
+      "loss": 3.5584,
       "step": 20750
     },
     {
-      "epoch": 2.2387256484770206,
-      "grad_norm": 0.6433561444282532,
-      "learning_rate": 0.0004662234672987824,
-      "loss": 3.5546,
+      "epoch": 2.242587601078167,
+      "grad_norm": 0.6599795818328857,
+      "learning_rate": 0.0004659859686994063,
+      "loss": 3.5644,
       "step": 20800
     },
     {
-      "epoch": 2.244107200516629,
-      "grad_norm": 0.5780196189880371,
-      "learning_rate": 0.00046590022626872104,
-      "loss": 3.5505,
+      "epoch": 2.247978436657682,
+      "grad_norm": 0.5787530541419983,
+      "learning_rate": 0.0004656621694549379,
+      "loss": 3.5593,
       "step": 20850
     },
     {
-      "epoch": 2.2494887525562373,
-      "grad_norm": 0.6302474141120911,
-      "learning_rate": 0.0004655769852386596,
-      "loss": 3.5528,
+      "epoch": 2.2533692722371965,
+      "grad_norm": 0.5842300653457642,
+      "learning_rate": 0.00046533837021046947,
+      "loss": 3.5535,
       "step": 20900
     },
     {
-      "epoch": 2.2548703045958454,
-      "grad_norm": 0.6163661479949951,
-      "learning_rate": 0.0004652537442085982,
-      "loss": 3.5506,
+      "epoch": 2.2587601078167117,
+      "grad_norm": 0.5889685750007629,
+      "learning_rate": 0.0004650210469508904,
+      "loss": 3.5653,
       "step": 20950
     },
     {
-      "epoch": 2.2602518566354535,
-      "grad_norm": 0.5856783390045166,
-      "learning_rate": 0.00046493050317853677,
-      "loss": 3.5485,
+      "epoch": 2.2641509433962264,
+      "grad_norm": 0.5827546715736389,
+      "learning_rate": 0.00046469724770642197,
+      "loss": 3.5472,
       "step": 21000
     },
     {
-      "epoch": 2.2602518566354535,
-      "eval_accuracy": 0.364305092528256,
-      "eval_loss": 3.5584065914154053,
-      "eval_runtime": 179.7572,
-      "eval_samples_per_second": 100.196,
-      "eval_steps_per_second": 6.264,
+      "epoch": 2.2641509433962264,
+      "eval_accuracy": 0.3631977022956508,
+      "eval_loss": 3.566005229949951,
+      "eval_runtime": 184.999,
+      "eval_samples_per_second": 97.357,
+      "eval_steps_per_second": 6.087,
       "step": 21000
     },
     {
-      "epoch": 2.265633408675062,
-      "grad_norm": 0.5490982532501221,
-      "learning_rate": 0.0004646072621484753,
-      "loss": 3.5765,
+      "epoch": 2.269541778975741,
+      "grad_norm": 0.5650562047958374,
+      "learning_rate": 0.0004643734484619536,
+      "loss": 3.5607,
       "step": 21050
     },
     {
-      "epoch": 2.27101496071467,
-      "grad_norm": 0.6259151697158813,
-      "learning_rate": 0.0004642840211184139,
-      "loss": 3.5557,
+      "epoch": 2.274932614555256,
+      "grad_norm": 0.5835627913475037,
+      "learning_rate": 0.0004640496492174851,
+      "loss": 3.5659,
       "step": 21100
     },
     {
-      "epoch": 2.2763965127542782,
-      "grad_norm": 0.6125593185424805,
-      "learning_rate": 0.00046396078008835255,
-      "loss": 3.562,
+      "epoch": 2.280323450134771,
+      "grad_norm": 0.621948778629303,
+      "learning_rate": 0.00046372584997301673,
+      "loss": 3.5788,
       "step": 21150
     },
     {
-      "epoch": 2.281778064793887,
-      "grad_norm": 0.596523642539978,
-      "learning_rate": 0.0004636375390582911,
-      "loss": 3.544,
+      "epoch": 2.2857142857142856,
+      "grad_norm": 0.5712414383888245,
+      "learning_rate": 0.0004634020507285483,
+      "loss": 3.5315,
       "step": 21200
     },
     {
-      "epoch": 2.287159616833495,
-      "grad_norm": 0.6259929537773132,
-      "learning_rate": 0.0004633142980282297,
-      "loss": 3.5504,
+      "epoch": 2.2911051212938007,
+      "grad_norm": 0.5780783295631409,
+      "learning_rate": 0.0004630782514840798,
+      "loss": 3.5586,
       "step": 21250
     },
     {
-      "epoch": 2.292541168873103,
-      "grad_norm": 0.5609055161476135,
-      "learning_rate": 0.00046299105699816823,
-      "loss": 3.5574,
+      "epoch": 2.2964959568733154,
+      "grad_norm": 0.5562280416488647,
+      "learning_rate": 0.0004627544522396114,
+      "loss": 3.559,
       "step": 21300
     },
     {
-      "epoch": 2.297922720912711,
-      "grad_norm": 0.6183776259422302,
-      "learning_rate": 0.0004626678159681068,
-      "loss": 3.5536,
+      "epoch": 2.30188679245283,
+      "grad_norm": 0.5436495542526245,
+      "learning_rate": 0.00046243065299514293,
+      "loss": 3.5524,
       "step": 21350
     },
     {
-      "epoch": 2.303304272952319,
-      "grad_norm": 0.531791090965271,
-      "learning_rate": 0.0004623445749380454,
-      "loss": 3.5414,
+      "epoch": 2.3072776280323453,
+      "grad_norm": 0.5707684755325317,
+      "learning_rate": 0.00046210685375067454,
+      "loss": 3.5663,
       "step": 21400
     },
     {
-      "epoch": 2.3086858249919278,
-      "grad_norm": 0.558052122592926,
-      "learning_rate": 0.000462021333907984,
-      "loss": 3.538,
+      "epoch": 2.31266846361186,
+      "grad_norm": 0.5568186044692993,
+      "learning_rate": 0.0004617830545062061,
+      "loss": 3.5645,
       "step": 21450
     },
     {
-      "epoch": 2.314067377031536,
-      "grad_norm": 0.5366657972335815,
-      "learning_rate": 0.0004616980928779226,
-      "loss": 3.5508,
+      "epoch": 2.3180592991913747,
+      "grad_norm": 0.5847433805465698,
+      "learning_rate": 0.0004614592552617377,
+      "loss": 3.5663,
       "step": 21500
     },
     {
-      "epoch": 2.319448929071144,
-      "grad_norm": 0.5605040192604065,
-      "learning_rate": 0.0004613748518478612,
-      "loss": 3.5527,
+      "epoch": 2.3234501347708894,
+      "grad_norm": 0.551858127117157,
+      "learning_rate": 0.00046113545601726924,
+      "loss": 3.5486,
       "step": 21550
     },
     {
-      "epoch": 2.3248304811107525,
-      "grad_norm": 0.5950835347175598,
-      "learning_rate": 0.00046105161081779974,
-      "loss": 3.553,
+      "epoch": 2.3288409703504045,
+      "grad_norm": 0.5552904605865479,
+      "learning_rate": 0.00046081165677280085,
+      "loss": 3.5698,
       "step": 21600
     },
     {
-      "epoch": 2.3302120331503606,
-      "grad_norm": 0.5673235058784485,
-      "learning_rate": 0.00046072836978773834,
-      "loss": 3.5484,
+      "epoch": 2.334231805929919,
+      "grad_norm": 0.5644253492355347,
+      "learning_rate": 0.0004604878575283324,
+      "loss": 3.564,
       "step": 21650
     },
     {
-      "epoch": 2.3355935851899687,
-      "grad_norm": 0.5962451100349426,
-      "learning_rate": 0.000460405128757677,
-      "loss": 3.5665,
+      "epoch": 2.339622641509434,
+      "grad_norm": 0.5640168786048889,
+      "learning_rate": 0.00046016405828386395,
+      "loss": 3.5443,
       "step": 21700
     },
     {
-      "epoch": 2.340975137229577,
-      "grad_norm": 0.5817105770111084,
-      "learning_rate": 0.0004600818877276155,
-      "loss": 3.5456,
+      "epoch": 2.3450134770889486,
+      "grad_norm": 0.5596201419830322,
+      "learning_rate": 0.00045984025903939555,
+      "loss": 3.5594,
       "step": 21750
     },
     {
-      "epoch": 2.3463566892691854,
-      "grad_norm": 0.5566921234130859,
-      "learning_rate": 0.0004597586466975541,
-      "loss": 3.5419,
+      "epoch": 2.3504043126684637,
+      "grad_norm": 0.6457816958427429,
+      "learning_rate": 0.0004595164597949271,
+      "loss": 3.5811,
       "step": 21800
     },
     {
-      "epoch": 2.3517382413087935,
-      "grad_norm": 0.5889357924461365,
-      "learning_rate": 0.00045943540566749266,
-      "loss": 3.553,
+      "epoch": 2.3557951482479784,
+      "grad_norm": 0.610068142414093,
+      "learning_rate": 0.0004591926605504587,
+      "loss": 3.5691,
       "step": 21850
     },
     {
-      "epoch": 2.3571197933484016,
-      "grad_norm": 0.5869264006614685,
-      "learning_rate": 0.00045911216463743126,
-      "loss": 3.549,
+      "epoch": 2.361185983827493,
+      "grad_norm": 0.6661171317100525,
+      "learning_rate": 0.0004588688613059902,
+      "loss": 3.5453,
       "step": 21900
     },
     {
-      "epoch": 2.3625013453880097,
-      "grad_norm": 0.5697734951972961,
-      "learning_rate": 0.00045878892360736985,
-      "loss": 3.5583,
+      "epoch": 2.3665768194070083,
+      "grad_norm": 0.5725359916687012,
+      "learning_rate": 0.00045854506206152186,
+      "loss": 3.5598,
       "step": 21950
     },
     {
-      "epoch": 2.3678828974276183,
-      "grad_norm": 0.6061686873435974,
-      "learning_rate": 0.00045846568257730845,
-      "loss": 3.5613,
+      "epoch": 2.371967654986523,
+      "grad_norm": 0.5913756489753723,
+      "learning_rate": 0.00045822126281705336,
+      "loss": 3.5526,
       "step": 22000
     },
     {
-      "epoch": 2.3678828974276183,
-      "eval_accuracy": 0.36498678074640406,
-      "eval_loss": 3.547973871231079,
-      "eval_runtime": 180.0761,
-      "eval_samples_per_second": 100.019,
-      "eval_steps_per_second": 6.253,
+      "epoch": 2.371967654986523,
+      "eval_accuracy": 0.36443677982828715,
+      "eval_loss": 3.5550854206085205,
+      "eval_runtime": 184.9317,
+      "eval_samples_per_second": 97.393,
+      "eval_steps_per_second": 6.089,
       "step": 22000
     },
     {
-      "epoch": 2.3732644494672264,
-      "grad_norm": 0.5746018290519714,
-      "learning_rate": 0.00045814244154724704,
-      "loss": 3.5496,
+      "epoch": 2.3773584905660377,
+      "grad_norm": 0.6312258839607239,
+      "learning_rate": 0.00045789746357258497,
+      "loss": 3.5527,
       "step": 22050
     },
     {
-      "epoch": 2.3786460015068345,
-      "grad_norm": 0.6272563338279724,
-      "learning_rate": 0.00045781920051718563,
-      "loss": 3.5636,
+      "epoch": 2.382749326145553,
+      "grad_norm": 0.5353261828422546,
+      "learning_rate": 0.0004575736643281165,
+      "loss": 3.5654,
       "step": 22100
     },
     {
-      "epoch": 2.384027553546443,
-      "grad_norm": 0.5796659588813782,
-      "learning_rate": 0.0004574959594871242,
-      "loss": 3.5412,
+      "epoch": 2.3881401617250675,
+      "grad_norm": 0.5589489340782166,
+      "learning_rate": 0.00045724986508364807,
+      "loss": 3.5687,
       "step": 22150
     },
     {
-      "epoch": 2.389409105586051,
-      "grad_norm": 0.5960429906845093,
-      "learning_rate": 0.00045717271845706277,
-      "loss": 3.5478,
+      "epoch": 2.393530997304582,
+      "grad_norm": 0.6058165431022644,
+      "learning_rate": 0.00045692606583917967,
+      "loss": 3.5535,
       "step": 22200
     },
     {
-      "epoch": 2.3947906576256592,
-      "grad_norm": 0.6166843175888062,
-      "learning_rate": 0.0004568494774270013,
-      "loss": 3.5584,
+      "epoch": 2.398921832884097,
+      "grad_norm": 0.5829679369926453,
+      "learning_rate": 0.0004566087425796006,
+      "loss": 3.5606,
       "step": 22250
     },
     {
-      "epoch": 2.4001722096652673,
-      "grad_norm": 0.5762814879417419,
-      "learning_rate": 0.00045652623639693996,
-      "loss": 3.537,
+      "epoch": 2.404312668463612,
+      "grad_norm": 0.6008812189102173,
+      "learning_rate": 0.0004562849433351322,
+      "loss": 3.5468,
       "step": 22300
     },
     {
-      "epoch": 2.4055537617048754,
-      "grad_norm": 0.5935600399971008,
-      "learning_rate": 0.00045620299536687855,
-      "loss": 3.5506,
+      "epoch": 2.4097035040431267,
+      "grad_norm": 0.5908893346786499,
+      "learning_rate": 0.0004559611440906638,
+      "loss": 3.5721,
       "step": 22350
     },
     {
-      "epoch": 2.410935313744484,
-      "grad_norm": 0.5715270042419434,
-      "learning_rate": 0.0004558797543368171,
-      "loss": 3.5349,
+      "epoch": 2.4150943396226414,
+      "grad_norm": 0.5803592205047607,
+      "learning_rate": 0.00045563734484619533,
+      "loss": 3.5554,
       "step": 22400
     },
     {
-      "epoch": 2.416316865784092,
-      "grad_norm": 0.5746564269065857,
-      "learning_rate": 0.0004555565133067557,
-      "loss": 3.5393,
+      "epoch": 2.420485175202156,
+      "grad_norm": 0.5852487087249756,
+      "learning_rate": 0.0004553135456017269,
+      "loss": 3.5566,
       "step": 22450
     },
     {
-      "epoch": 2.4216984178237,
-      "grad_norm": 0.6002961993217468,
-      "learning_rate": 0.0004552332722766943,
-      "loss": 3.5703,
+      "epoch": 2.4258760107816713,
+      "grad_norm": 0.618068516254425,
+      "learning_rate": 0.0004549897463572585,
+      "loss": 3.566,
       "step": 22500
     },
     {
-      "epoch": 2.4270799698633088,
-      "grad_norm": 0.6364290118217468,
-      "learning_rate": 0.0004549100312466328,
-      "loss": 3.535,
+      "epoch": 2.431266846361186,
+      "grad_norm": 0.5683373808860779,
+      "learning_rate": 0.00045466594711279,
+      "loss": 3.5557,
       "step": 22550
     },
     {
-      "epoch": 2.432461521902917,
-      "grad_norm": 0.5654012560844421,
-      "learning_rate": 0.0004545867902165715,
-      "loss": 3.5494,
+      "epoch": 2.4366576819407006,
+      "grad_norm": 0.5734935998916626,
+      "learning_rate": 0.00045434214786832164,
+      "loss": 3.5868,
       "step": 22600
     },
     {
-      "epoch": 2.437843073942525,
-      "grad_norm": 0.6237375736236572,
-      "learning_rate": 0.00045426354918651007,
-      "loss": 3.5464,
+      "epoch": 2.442048517520216,
+      "grad_norm": 0.5584690570831299,
+      "learning_rate": 0.00045401834862385314,
+      "loss": 3.5469,
       "step": 22650
     },
     {
-      "epoch": 2.443224625982133,
-      "grad_norm": 0.5933444499969482,
-      "learning_rate": 0.0004539467729770499,
-      "loss": 3.5417,
+      "epoch": 2.4474393530997305,
+      "grad_norm": 0.5913657546043396,
+      "learning_rate": 0.00045369454937938474,
+      "loss": 3.5375,
       "step": 22700
     },
     {
-      "epoch": 2.4486061780217416,
-      "grad_norm": 0.5668746829032898,
-      "learning_rate": 0.0004536235319469884,
-      "loss": 3.5352,
+      "epoch": 2.452830188679245,
+      "grad_norm": 0.5663593411445618,
+      "learning_rate": 0.0004533707501349163,
+      "loss": 3.5422,
       "step": 22750
     },
     {
-      "epoch": 2.4539877300613497,
-      "grad_norm": 0.6203579902648926,
-      "learning_rate": 0.000453300290916927,
-      "loss": 3.574,
+      "epoch": 2.4582210242587603,
+      "grad_norm": 0.5713441371917725,
+      "learning_rate": 0.0004530469508904479,
+      "loss": 3.5542,
       "step": 22800
     },
     {
-      "epoch": 2.459369282100958,
-      "grad_norm": 0.5573844909667969,
-      "learning_rate": 0.0004529770498868656,
-      "loss": 3.5529,
+      "epoch": 2.463611859838275,
+      "grad_norm": 0.5829697847366333,
+      "learning_rate": 0.00045272315164597945,
+      "loss": 3.5379,
       "step": 22850
     },
     {
-      "epoch": 2.464750834140566,
-      "grad_norm": 0.6508901715278625,
-      "learning_rate": 0.00045265380885680414,
-      "loss": 3.531,
+      "epoch": 2.4690026954177897,
+      "grad_norm": 0.5785667300224304,
+      "learning_rate": 0.000452399352401511,
+      "loss": 3.5559,
       "step": 22900
     },
     {
-      "epoch": 2.4701323861801745,
-      "grad_norm": 0.5782315135002136,
-      "learning_rate": 0.0004523305678267428,
-      "loss": 3.5689,
+      "epoch": 2.4743935309973044,
+      "grad_norm": 0.6283608675003052,
+      "learning_rate": 0.0004520755531570426,
+      "loss": 3.5512,
       "step": 22950
     },
     {
-      "epoch": 2.4755139382197826,
-      "grad_norm": 0.6106559038162231,
-      "learning_rate": 0.0004520073267966814,
-      "loss": 3.5521,
+      "epoch": 2.4797843665768196,
+      "grad_norm": 0.5529918670654297,
+      "learning_rate": 0.00045175175391257415,
+      "loss": 3.544,
       "step": 23000
     },
     {
-      "epoch": 2.4755139382197826,
-      "eval_accuracy": 0.3663403784227969,
-      "eval_loss": 3.5380020141601562,
-      "eval_runtime": 179.612,
-      "eval_samples_per_second": 100.277,
-      "eval_steps_per_second": 6.269,
+      "epoch": 2.4797843665768196,
+      "eval_accuracy": 0.3652747108991124,
+      "eval_loss": 3.542670726776123,
+      "eval_runtime": 184.9689,
+      "eval_samples_per_second": 97.373,
+      "eval_steps_per_second": 6.088,
       "step": 23000
     },
     {
-      "epoch": 2.4808954902593907,
-      "grad_norm": 0.5985417366027832,
-      "learning_rate": 0.00045168408576661993,
-      "loss": 3.5309,
+      "epoch": 2.4851752021563343,
+      "grad_norm": 0.5712370872497559,
+      "learning_rate": 0.00045142795466810576,
+      "loss": 3.5568,
       "step": 23050
     },
     {
-      "epoch": 2.4862770422989993,
-      "grad_norm": 0.5568486452102661,
-      "learning_rate": 0.0004513608447365585,
-      "loss": 3.5504,
+      "epoch": 2.490566037735849,
+      "grad_norm": 0.589131236076355,
+      "learning_rate": 0.0004511041554236373,
+      "loss": 3.5592,
       "step": 23100
     },
     {
-      "epoch": 2.4916585943386074,
-      "grad_norm": 0.607467770576477,
-      "learning_rate": 0.00045103760370649706,
-      "loss": 3.5585,
+      "epoch": 2.4959568733153636,
+      "grad_norm": 0.5965885519981384,
+      "learning_rate": 0.0004507803561791689,
+      "loss": 3.5562,
       "step": 23150
     },
     {
-      "epoch": 2.4970401463782155,
-      "grad_norm": 0.6122623085975647,
-      "learning_rate": 0.00045071436267643566,
-      "loss": 3.55,
+      "epoch": 2.501347708894879,
+      "grad_norm": 0.591253936290741,
+      "learning_rate": 0.00045045655693470046,
+      "loss": 3.5516,
       "step": 23200
     },
     {
-      "epoch": 2.5024216984178236,
-      "grad_norm": 0.6299822330474854,
-      "learning_rate": 0.0004503911216463743,
-      "loss": 3.5361,
+      "epoch": 2.5067385444743935,
+      "grad_norm": 0.575657308101654,
+      "learning_rate": 0.00045013275769023207,
+      "loss": 3.5523,
       "step": 23250
     },
     {
-      "epoch": 2.5078032504574317,
-      "grad_norm": 0.5975040793418884,
-      "learning_rate": 0.00045006788061631285,
-      "loss": 3.5452,
+      "epoch": 2.512129380053908,
+      "grad_norm": 0.549932599067688,
+      "learning_rate": 0.00044980895844576356,
+      "loss": 3.5368,
       "step": 23300
     },
     {
-      "epoch": 2.5131848024970402,
-      "grad_norm": 0.6091236472129822,
-      "learning_rate": 0.00044974463958625144,
-      "loss": 3.5612,
+      "epoch": 2.5175202156334233,
+      "grad_norm": 0.5937458276748657,
+      "learning_rate": 0.0004494851592012951,
+      "loss": 3.544,
       "step": 23350
     },
     {
-      "epoch": 2.5185663545366483,
-      "grad_norm": 0.5954560041427612,
-      "learning_rate": 0.00044942139855619004,
-      "loss": 3.5283,
+      "epoch": 2.522911051212938,
+      "grad_norm": 0.576596736907959,
+      "learning_rate": 0.0004491613599568267,
+      "loss": 3.5412,
       "step": 23400
     },
     {
-      "epoch": 2.5239479065762565,
-      "grad_norm": 0.5859009623527527,
-      "learning_rate": 0.0004490981575261286,
-      "loss": 3.537,
+      "epoch": 2.5283018867924527,
+      "grad_norm": 0.5818652510643005,
+      "learning_rate": 0.00044883756071235827,
+      "loss": 3.5493,
       "step": 23450
     },
     {
-      "epoch": 2.529329458615865,
-      "grad_norm": 0.5438385009765625,
-      "learning_rate": 0.0004487749164960672,
-      "loss": 3.5416,
+      "epoch": 2.533692722371968,
+      "grad_norm": 0.5370622873306274,
+      "learning_rate": 0.0004485137614678899,
+      "loss": 3.5539,
       "step": 23500
     },
     {
-      "epoch": 2.534711010655473,
-      "grad_norm": 0.5849915146827698,
-      "learning_rate": 0.0004484516754660058,
-      "loss": 3.544,
+      "epoch": 2.5390835579514826,
+      "grad_norm": 0.5701831579208374,
+      "learning_rate": 0.0004481899622234214,
+      "loss": 3.5573,
       "step": 23550
     },
     {
-      "epoch": 2.540092562695081,
-      "grad_norm": 0.6110696792602539,
-      "learning_rate": 0.00044812843443594436,
-      "loss": 3.5456,
+      "epoch": 2.5444743935309972,
+      "grad_norm": 0.6162128448486328,
+      "learning_rate": 0.00044786616297895303,
+      "loss": 3.558,
       "step": 23600
     },
     {
-      "epoch": 2.5454741147346893,
-      "grad_norm": 0.6262888312339783,
-      "learning_rate": 0.00044780519340588296,
-      "loss": 3.5357,
+      "epoch": 2.5498652291105124,
+      "grad_norm": 0.547835648059845,
+      "learning_rate": 0.0004475423637344846,
+      "loss": 3.5503,
       "step": 23650
     },
     {
-      "epoch": 2.550855666774298,
-      "grad_norm": 0.6227126717567444,
-      "learning_rate": 0.0004474819523758215,
-      "loss": 3.5471,
+      "epoch": 2.555256064690027,
+      "grad_norm": 0.5706479549407959,
+      "learning_rate": 0.00044721856449001613,
+      "loss": 3.5553,
       "step": 23700
     },
     {
-      "epoch": 2.556237218813906,
-      "grad_norm": 0.607038140296936,
-      "learning_rate": 0.0004471587113457601,
-      "loss": 3.5478,
+      "epoch": 2.560646900269542,
+      "grad_norm": 0.5650632381439209,
+      "learning_rate": 0.00044689476524554774,
+      "loss": 3.5479,
       "step": 23750
     },
     {
-      "epoch": 2.561618770853514,
-      "grad_norm": 0.5638023614883423,
-      "learning_rate": 0.00044683547031569874,
-      "loss": 3.5658,
+      "epoch": 2.5660377358490565,
+      "grad_norm": 0.5929921865463257,
+      "learning_rate": 0.0004465709660010793,
+      "loss": 3.5634,
       "step": 23800
     },
     {
-      "epoch": 2.567000322893122,
-      "grad_norm": 0.6163343787193298,
-      "learning_rate": 0.0004465122292856373,
-      "loss": 3.5602,
+      "epoch": 2.571428571428571,
+      "grad_norm": 0.6945008039474487,
+      "learning_rate": 0.0004462471667566109,
+      "loss": 3.5678,
       "step": 23850
     },
     {
-      "epoch": 2.5723818749327307,
-      "grad_norm": 0.5720522999763489,
-      "learning_rate": 0.0004461889882555759,
-      "loss": 3.5562,
+      "epoch": 2.5768194070080863,
+      "grad_norm": 0.5972580313682556,
+      "learning_rate": 0.00044592336751214244,
+      "loss": 3.5669,
       "step": 23900
     },
     {
-      "epoch": 2.577763426972339,
-      "grad_norm": 0.636634111404419,
-      "learning_rate": 0.00044586574722551447,
-      "loss": 3.5685,
+      "epoch": 2.582210242587601,
+      "grad_norm": 0.6179361343383789,
+      "learning_rate": 0.00044559956826767405,
+      "loss": 3.5411,
       "step": 23950
     },
     {
-      "epoch": 2.583144979011947,
-      "grad_norm": 0.5891180634498596,
-      "learning_rate": 0.000445542506195453,
-      "loss": 3.5455,
+      "epoch": 2.5876010781671157,
+      "grad_norm": 0.538170337677002,
+      "learning_rate": 0.00044527576902320554,
+      "loss": 3.5367,
       "step": 24000
     },
     {
-      "epoch": 2.583144979011947,
-      "eval_accuracy": 0.36704879525134726,
-      "eval_loss": 3.526322364807129,
-      "eval_runtime": 179.7285,
-      "eval_samples_per_second": 100.212,
-      "eval_steps_per_second": 6.265,
+      "epoch": 2.5876010781671157,
+      "eval_accuracy": 0.3663797107681857,
+      "eval_loss": 3.533842086791992,
+      "eval_runtime": 184.8514,
+      "eval_samples_per_second": 97.435,
+      "eval_steps_per_second": 6.091,
       "step": 24000
     },
     {
-      "epoch": 2.5885265310515555,
-      "grad_norm": 0.6043258905410767,
-      "learning_rate": 0.0004452192651653916,
-      "loss": 3.5164,
+      "epoch": 2.592991913746631,
+      "grad_norm": 0.6353852152824402,
+      "learning_rate": 0.00044495196977873715,
+      "loss": 3.5368,
       "step": 24050
     },
     {
-      "epoch": 2.5939080830911636,
-      "grad_norm": 0.6400256752967834,
-      "learning_rate": 0.00044489602413533025,
-      "loss": 3.5409,
+      "epoch": 2.5983827493261455,
+      "grad_norm": 0.5333035588264465,
+      "learning_rate": 0.0004446281705342687,
+      "loss": 3.5384,
       "step": 24100
     },
     {
-      "epoch": 2.5992896351307717,
-      "grad_norm": 0.6273722648620605,
-      "learning_rate": 0.0004445727831052688,
-      "loss": 3.5391,
+      "epoch": 2.6037735849056602,
+      "grad_norm": 0.5871497988700867,
+      "learning_rate": 0.00044430437128980025,
+      "loss": 3.5439,
       "step": 24150
     },
     {
-      "epoch": 2.60467118717038,
-      "grad_norm": 0.6210452318191528,
-      "learning_rate": 0.0004442495420752074,
-      "loss": 3.5436,
+      "epoch": 2.6091644204851754,
+      "grad_norm": 0.6557198166847229,
+      "learning_rate": 0.00044398057204533185,
+      "loss": 3.5563,
       "step": 24200
     },
     {
-      "epoch": 2.610052739209988,
-      "grad_norm": 0.5660796761512756,
-      "learning_rate": 0.00044392630104514593,
-      "loss": 3.5256,
+      "epoch": 2.61455525606469,
+      "grad_norm": 0.5965690016746521,
+      "learning_rate": 0.0004436567728008634,
+      "loss": 3.5611,
       "step": 24250
     },
     {
-      "epoch": 2.6154342912495965,
-      "grad_norm": 0.5474342107772827,
-      "learning_rate": 0.0004436030600150845,
-      "loss": 3.526,
+      "epoch": 2.6199460916442048,
+      "grad_norm": 0.5756832957267761,
+      "learning_rate": 0.000443332973556395,
+      "loss": 3.5472,
       "step": 24300
     },
     {
-      "epoch": 2.6208158432892046,
-      "grad_norm": 0.5688294768333435,
-      "learning_rate": 0.0004432798189850231,
-      "loss": 3.5341,
+      "epoch": 2.62533692722372,
+      "grad_norm": 0.5462123155593872,
+      "learning_rate": 0.00044300917431192656,
+      "loss": 3.5405,
       "step": 24350
     },
     {
-      "epoch": 2.6261973953288127,
-      "grad_norm": 0.5909520387649536,
-      "learning_rate": 0.0004429565779549617,
-      "loss": 3.5394,
+      "epoch": 2.6307277628032346,
+      "grad_norm": 0.56703120470047,
+      "learning_rate": 0.00044268537506745816,
+      "loss": 3.5526,
       "step": 24400
     },
     {
-      "epoch": 2.6315789473684212,
-      "grad_norm": 0.5881052613258362,
-      "learning_rate": 0.0004426333369249003,
-      "loss": 3.544,
+      "epoch": 2.6361185983827493,
+      "grad_norm": 0.5480033755302429,
+      "learning_rate": 0.0004423615758229897,
+      "loss": 3.5252,
       "step": 24450
     },
     {
-      "epoch": 2.6369604994080293,
-      "grad_norm": 0.5401953458786011,
-      "learning_rate": 0.0004423100958948389,
-      "loss": 3.5518,
+      "epoch": 2.641509433962264,
+      "grad_norm": 0.5893705487251282,
+      "learning_rate": 0.0004420377765785213,
+      "loss": 3.5317,
       "step": 24500
     },
     {
-      "epoch": 2.6423420514476375,
-      "grad_norm": 0.6648808121681213,
-      "learning_rate": 0.00044198685486477744,
-      "loss": 3.5483,
+      "epoch": 2.6469002695417787,
+      "grad_norm": 0.6192256212234497,
+      "learning_rate": 0.00044171397733405287,
+      "loss": 3.5497,
       "step": 24550
     },
     {
-      "epoch": 2.6477236034872456,
-      "grad_norm": 0.5619612336158752,
-      "learning_rate": 0.00044166361383471604,
-      "loss": 3.5388,
+      "epoch": 2.652291105121294,
+      "grad_norm": 0.5576728582382202,
+      "learning_rate": 0.00044139017808958437,
+      "loss": 3.5294,
       "step": 24600
     },
     {
-      "epoch": 2.653105155526854,
-      "grad_norm": 0.5457707643508911,
-      "learning_rate": 0.0004413403728046547,
-      "loss": 3.5491,
+      "epoch": 2.6576819407008085,
+      "grad_norm": 0.5619601607322693,
+      "learning_rate": 0.00044106637884511597,
+      "loss": 3.5434,
       "step": 24650
     },
     {
-      "epoch": 2.658486707566462,
-      "grad_norm": 0.6164374947547913,
-      "learning_rate": 0.0004410171317745932,
-      "loss": 3.5383,
+      "epoch": 2.6630727762803232,
+      "grad_norm": 0.5698447227478027,
+      "learning_rate": 0.0004407425796006475,
+      "loss": 3.5462,
       "step": 24700
     },
     {
-      "epoch": 2.6638682596060703,
-      "grad_norm": 0.5915449261665344,
-      "learning_rate": 0.0004406938907445318,
-      "loss": 3.5412,
+      "epoch": 2.6684636118598384,
+      "grad_norm": 0.585787832736969,
+      "learning_rate": 0.00044041878035617913,
+      "loss": 3.5486,
       "step": 24750
     },
     {
-      "epoch": 2.6692498116456784,
-      "grad_norm": 0.6016857028007507,
-      "learning_rate": 0.00044037064971447036,
-      "loss": 3.5451,
+      "epoch": 2.673854447439353,
+      "grad_norm": 0.5916891694068909,
+      "learning_rate": 0.0004400949811117107,
+      "loss": 3.5477,
       "step": 24800
     },
     {
-      "epoch": 2.674631363685287,
-      "grad_norm": 0.5480591058731079,
-      "learning_rate": 0.00044004740868440896,
-      "loss": 3.538,
+      "epoch": 2.6792452830188678,
+      "grad_norm": 0.5576286315917969,
+      "learning_rate": 0.0004397711818672423,
+      "loss": 3.5487,
       "step": 24850
     },
     {
-      "epoch": 2.680012915724895,
-      "grad_norm": 0.5656848549842834,
-      "learning_rate": 0.00043972416765434755,
-      "loss": 3.5566,
+      "epoch": 2.684636118598383,
+      "grad_norm": 0.5873832702636719,
+      "learning_rate": 0.00043944738262277383,
+      "loss": 3.5629,
       "step": 24900
     },
     {
-      "epoch": 2.685394467764503,
-      "grad_norm": 0.6277399063110352,
-      "learning_rate": 0.00043940092662428615,
-      "loss": 3.5456,
+      "epoch": 2.6900269541778976,
+      "grad_norm": 0.5807034969329834,
+      "learning_rate": 0.00043912358337830544,
+      "loss": 3.5528,
       "step": 24950
     },
     {
-      "epoch": 2.6907760198041117,
-      "grad_norm": 0.6056678295135498,
-      "learning_rate": 0.00043907768559422474,
-      "loss": 3.5473,
+      "epoch": 2.6954177897574123,
+      "grad_norm": 0.5750570297241211,
+      "learning_rate": 0.000438799784133837,
+      "loss": 3.5471,
       "step": 25000
     },
     {
-      "epoch": 2.6907760198041117,
-      "eval_accuracy": 0.36839630836602244,
-      "eval_loss": 3.5200185775756836,
-      "eval_runtime": 179.6896,
-      "eval_samples_per_second": 100.234,
-      "eval_steps_per_second": 6.266,
+      "epoch": 2.6954177897574123,
+      "eval_accuracy": 0.3674458129034214,
+      "eval_loss": 3.522408962249756,
+      "eval_runtime": 184.8934,
+      "eval_samples_per_second": 97.413,
+      "eval_steps_per_second": 6.09,
       "step": 25000
     },
     {
-      "epoch": 2.69615757184372,
-      "grad_norm": 0.635338544845581,
-      "learning_rate": 0.00043875444456416334,
-      "loss": 3.5616,
+      "epoch": 2.7008086253369274,
+      "grad_norm": 0.5960842967033386,
+      "learning_rate": 0.00043847598488936854,
+      "loss": 3.5387,
       "step": 25050
     },
     {
-      "epoch": 2.701539123883328,
-      "grad_norm": 0.6054742932319641,
-      "learning_rate": 0.0004384312035341019,
-      "loss": 3.5651,
+      "epoch": 2.706199460916442,
+      "grad_norm": 0.5355721712112427,
+      "learning_rate": 0.00043815218564490014,
+      "loss": 3.5478,
       "step": 25100
     },
     {
-      "epoch": 2.706920675922936,
-      "grad_norm": 0.6213764548301697,
-      "learning_rate": 0.00043810796250404047,
-      "loss": 3.5406,
+      "epoch": 2.711590296495957,
+      "grad_norm": 0.5989283323287964,
+      "learning_rate": 0.0004378283864004317,
+      "loss": 3.5244,
       "step": 25150
     },
     {
-      "epoch": 2.712302227962544,
-      "grad_norm": 0.5552936792373657,
-      "learning_rate": 0.000437784721473979,
-      "loss": 3.5186,
+      "epoch": 2.7169811320754715,
+      "grad_norm": 0.6410152912139893,
+      "learning_rate": 0.0004375045871559633,
+      "loss": 3.546,
       "step": 25200
     },
     {
-      "epoch": 2.7176837800021527,
-      "grad_norm": 0.588412344455719,
-      "learning_rate": 0.00043746148044391766,
-      "loss": 3.5416,
+      "epoch": 2.7223719676549867,
+      "grad_norm": 0.632588267326355,
+      "learning_rate": 0.00043718078791149485,
+      "loss": 3.5491,
       "step": 25250
     },
     {
-      "epoch": 2.723065332041761,
-      "grad_norm": 0.5414818525314331,
-      "learning_rate": 0.00043713823941385625,
-      "loss": 3.5289,
+      "epoch": 2.7277628032345014,
+      "grad_norm": 0.5347442030906677,
+      "learning_rate": 0.00043685698866702645,
+      "loss": 3.5518,
       "step": 25300
     },
     {
-      "epoch": 2.728446884081369,
-      "grad_norm": 0.6171042919158936,
-      "learning_rate": 0.0004368149983837948,
-      "loss": 3.548,
+      "epoch": 2.733153638814016,
+      "grad_norm": 0.5966982245445251,
+      "learning_rate": 0.00043653318942255795,
+      "loss": 3.5438,
       "step": 25350
     },
     {
-      "epoch": 2.7338284361209775,
-      "grad_norm": 0.6250415444374084,
-      "learning_rate": 0.0004364917573537334,
-      "loss": 3.5393,
+      "epoch": 2.7385444743935308,
+      "grad_norm": 0.6366229057312012,
+      "learning_rate": 0.00043620939017808956,
+      "loss": 3.5336,
       "step": 25400
     },
     {
-      "epoch": 2.7392099881605856,
-      "grad_norm": 0.6534850597381592,
-      "learning_rate": 0.00043616851632367193,
-      "loss": 3.5351,
+      "epoch": 2.743935309973046,
+      "grad_norm": 0.598671019077301,
+      "learning_rate": 0.0004358855909336211,
+      "loss": 3.5338,
       "step": 25450
     },
     {
-      "epoch": 2.7445915402001937,
-      "grad_norm": 0.6056131720542908,
-      "learning_rate": 0.0004358452752936106,
-      "loss": 3.5331,
+      "epoch": 2.7493261455525606,
+      "grad_norm": 0.5955828428268433,
+      "learning_rate": 0.00043556179168915266,
+      "loss": 3.5497,
       "step": 25500
     },
     {
-      "epoch": 2.749973092239802,
-      "grad_norm": 0.6097458004951477,
-      "learning_rate": 0.0004355220342635492,
-      "loss": 3.5229,
+      "epoch": 2.7547169811320753,
+      "grad_norm": 0.5636226534843445,
+      "learning_rate": 0.00043523799244468426,
+      "loss": 3.5418,
       "step": 25550
     },
     {
-      "epoch": 2.7553546442794103,
-      "grad_norm": 0.5821205377578735,
-      "learning_rate": 0.00043519879323348777,
-      "loss": 3.5547,
+      "epoch": 2.7601078167115904,
+      "grad_norm": 0.5939311981201172,
+      "learning_rate": 0.0004349141932002158,
+      "loss": 3.5449,
       "step": 25600
     },
     {
-      "epoch": 2.7607361963190185,
-      "grad_norm": 0.572923481464386,
-      "learning_rate": 0.0004348755522034263,
-      "loss": 3.5459,
+      "epoch": 2.765498652291105,
+      "grad_norm": 0.5545902848243713,
+      "learning_rate": 0.0004345903939557474,
+      "loss": 3.537,
       "step": 25650
     },
     {
-      "epoch": 2.7661177483586266,
-      "grad_norm": 0.5959768295288086,
-      "learning_rate": 0.0004345523111733649,
-      "loss": 3.53,
+      "epoch": 2.77088948787062,
+      "grad_norm": 0.6240540742874146,
+      "learning_rate": 0.00043426659471127897,
+      "loss": 3.5434,
       "step": 25700
     },
     {
-      "epoch": 2.7714993003982347,
-      "grad_norm": 0.6123167872428894,
-      "learning_rate": 0.00043422907014330344,
-      "loss": 3.5332,
+      "epoch": 2.776280323450135,
+      "grad_norm": 0.5980251431465149,
+      "learning_rate": 0.00043394279546681057,
+      "loss": 3.53,
       "step": 25750
     },
     {
-      "epoch": 2.776880852437843,
-      "grad_norm": 0.5567452311515808,
-      "learning_rate": 0.0004339058291132421,
-      "loss": 3.5329,
+      "epoch": 2.7816711590296497,
+      "grad_norm": 0.5463680028915405,
+      "learning_rate": 0.0004336189962223421,
+      "loss": 3.5552,
       "step": 25800
     },
     {
-      "epoch": 2.7822624044774513,
-      "grad_norm": 0.6061723232269287,
-      "learning_rate": 0.0004335825880831807,
-      "loss": 3.5429,
+      "epoch": 2.7870619946091644,
+      "grad_norm": 0.5872852802276611,
+      "learning_rate": 0.0004332951969778737,
+      "loss": 3.5262,
       "step": 25850
     },
     {
-      "epoch": 2.7876439565170594,
-      "grad_norm": 0.5593557357788086,
-      "learning_rate": 0.00043325934705311923,
-      "loss": 3.5072,
+      "epoch": 2.7924528301886795,
+      "grad_norm": 0.5788702368736267,
+      "learning_rate": 0.0004329713977334053,
+      "loss": 3.538,
       "step": 25900
     },
     {
-      "epoch": 2.793025508556668,
-      "grad_norm": 0.5608957409858704,
-      "learning_rate": 0.0004329361060230578,
-      "loss": 3.5159,
+      "epoch": 2.797843665768194,
+      "grad_norm": 0.6034548282623291,
+      "learning_rate": 0.0004326475984889368,
+      "loss": 3.5353,
       "step": 25950
     },
     {
-      "epoch": 2.798407060596276,
-      "grad_norm": 0.5900284051895142,
-      "learning_rate": 0.00043261286499299636,
-      "loss": 3.5301,
+      "epoch": 2.803234501347709,
+      "grad_norm": 0.5133172869682312,
+      "learning_rate": 0.0004323237992444684,
+      "loss": 3.551,
       "step": 26000
     },
     {
-      "epoch": 2.798407060596276,
-      "eval_accuracy": 0.36886427635383945,
-      "eval_loss": 3.5095598697662354,
-      "eval_runtime": 180.5366,
-      "eval_samples_per_second": 99.764,
-      "eval_steps_per_second": 6.237,
+      "epoch": 2.803234501347709,
+      "eval_accuracy": 0.36823169924098353,
+      "eval_loss": 3.515547037124634,
+      "eval_runtime": 184.9718,
+      "eval_samples_per_second": 97.372,
+      "eval_steps_per_second": 6.087,
       "step": 26000
     },
     {
-      "epoch": 2.803788612635884,
-      "grad_norm": 0.5952501893043518,
-      "learning_rate": 0.00043228962396293496,
-      "loss": 3.5283,
+      "epoch": 2.8086253369272236,
+      "grad_norm": 0.6219244003295898,
+      "learning_rate": 0.00043199999999999993,
+      "loss": 3.556,
       "step": 26050
     },
     {
-      "epoch": 2.8091701646754923,
-      "grad_norm": 0.5934816598892212,
-      "learning_rate": 0.00043197284775347476,
-      "loss": 3.5287,
+      "epoch": 2.8140161725067383,
+      "grad_norm": 0.5854507684707642,
+      "learning_rate": 0.00043167620075553153,
+      "loss": 3.5406,
       "step": 26100
     },
     {
-      "epoch": 2.8145517167151004,
-      "grad_norm": 0.6665016412734985,
-      "learning_rate": 0.00043164960672341336,
-      "loss": 3.5178,
+      "epoch": 2.8194070080862534,
+      "grad_norm": 0.5761322379112244,
+      "learning_rate": 0.0004313524015110631,
+      "loss": 3.5364,
       "step": 26150
     },
     {
-      "epoch": 2.819933268754709,
-      "grad_norm": 0.5784537196159363,
-      "learning_rate": 0.000431326365693352,
-      "loss": 3.5368,
+      "epoch": 2.824797843665768,
+      "grad_norm": 0.5397259593009949,
+      "learning_rate": 0.0004310286022665947,
+      "loss": 3.5472,
       "step": 26200
     },
     {
-      "epoch": 2.825314820794317,
-      "grad_norm": 0.5916054248809814,
-      "learning_rate": 0.00043100312466329055,
-      "loss": 3.5361,
+      "epoch": 2.830188679245283,
+      "grad_norm": 0.5752435326576233,
+      "learning_rate": 0.0004307112790070156,
+      "loss": 3.5301,
       "step": 26250
     },
     {
-      "epoch": 2.830696372833925,
-      "grad_norm": 0.5684788227081299,
-      "learning_rate": 0.00043067988363322914,
-      "loss": 3.5204,
+      "epoch": 2.835579514824798,
+      "grad_norm": 0.6607756018638611,
+      "learning_rate": 0.0004303874797625472,
+      "loss": 3.5288,
       "step": 26300
     },
     {
-      "epoch": 2.8360779248735337,
-      "grad_norm": 0.5924267172813416,
-      "learning_rate": 0.00043035664260316774,
-      "loss": 3.5382,
+      "epoch": 2.8409703504043127,
+      "grad_norm": 0.5861449837684631,
+      "learning_rate": 0.00043006368051807874,
+      "loss": 3.5494,
       "step": 26350
     },
     {
-      "epoch": 2.841459476913142,
-      "grad_norm": 0.6063621640205383,
-      "learning_rate": 0.0004300334015731063,
-      "loss": 3.5343,
+      "epoch": 2.8463611859838274,
+      "grad_norm": 0.5686273574829102,
+      "learning_rate": 0.00042973988127361035,
+      "loss": 3.5248,
       "step": 26400
     },
     {
-      "epoch": 2.84684102895275,
-      "grad_norm": 0.6259608268737793,
-      "learning_rate": 0.0004297101605430449,
-      "loss": 3.5212,
+      "epoch": 2.8517520215633425,
+      "grad_norm": 0.5789076685905457,
+      "learning_rate": 0.0004294160820291419,
+      "loss": 3.5457,
       "step": 26450
     },
     {
-      "epoch": 2.852222580992358,
-      "grad_norm": 0.5912953019142151,
-      "learning_rate": 0.0004293869195129835,
-      "loss": 3.5333,
+      "epoch": 2.857142857142857,
+      "grad_norm": 0.589009165763855,
+      "learning_rate": 0.0004290922827846735,
+      "loss": 3.5533,
       "step": 26500
     },
     {
-      "epoch": 2.857604133031966,
-      "grad_norm": 0.6470860838890076,
-      "learning_rate": 0.00042906367848292206,
-      "loss": 3.5356,
+      "epoch": 2.862533692722372,
+      "grad_norm": 0.5996458530426025,
+      "learning_rate": 0.00042876848354020505,
+      "loss": 3.5342,
       "step": 26550
     },
     {
-      "epoch": 2.8629856850715747,
-      "grad_norm": 0.6130536198616028,
-      "learning_rate": 0.00042874043745286066,
-      "loss": 3.5259,
+      "epoch": 2.867924528301887,
+      "grad_norm": 0.5394443869590759,
+      "learning_rate": 0.00042844468429573655,
+      "loss": 3.5353,
       "step": 26600
     },
     {
-      "epoch": 2.868367237111183,
-      "grad_norm": 0.5800825357437134,
-      "learning_rate": 0.0004284171964227992,
-      "loss": 3.5425,
+      "epoch": 2.8733153638814017,
+      "grad_norm": 0.5539592504501343,
+      "learning_rate": 0.00042812088505126815,
+      "loss": 3.5198,
       "step": 26650
     },
     {
-      "epoch": 2.873748789150791,
-      "grad_norm": 0.5847500562667847,
-      "learning_rate": 0.0004280939553927378,
-      "loss": 3.5086,
+      "epoch": 2.8787061994609164,
+      "grad_norm": 0.5455945134162903,
+      "learning_rate": 0.0004277970858067997,
+      "loss": 3.561,
       "step": 26700
     },
     {
-      "epoch": 2.8791303411903995,
-      "grad_norm": 0.6748270988464355,
-      "learning_rate": 0.00042777071436267644,
-      "loss": 3.5213,
+      "epoch": 2.884097035040431,
+      "grad_norm": 0.6022780537605286,
+      "learning_rate": 0.0004274732865623313,
+      "loss": 3.5265,
       "step": 26750
     },
     {
-      "epoch": 2.8845118932300076,
-      "grad_norm": 0.576877772808075,
-      "learning_rate": 0.000427447473332615,
-      "loss": 3.5173,
+      "epoch": 2.889487870619946,
+      "grad_norm": 0.5307306051254272,
+      "learning_rate": 0.00042714948731786286,
+      "loss": 3.5104,
       "step": 26800
     },
     {
-      "epoch": 2.8898934452696157,
-      "grad_norm": 0.5994119644165039,
-      "learning_rate": 0.0004271242323025536,
-      "loss": 3.5184,
+      "epoch": 2.894878706199461,
+      "grad_norm": 0.6074073910713196,
+      "learning_rate": 0.00042682568807339447,
+      "loss": 3.5143,
       "step": 26850
     },
     {
-      "epoch": 2.895274997309224,
-      "grad_norm": 0.5870147943496704,
-      "learning_rate": 0.00042680099127249217,
-      "loss": 3.5404,
+      "epoch": 2.9002695417789757,
+      "grad_norm": 0.5870118141174316,
+      "learning_rate": 0.000426501888828926,
+      "loss": 3.5332,
       "step": 26900
     },
     {
-      "epoch": 2.9006565493488323,
-      "grad_norm": 0.5605146884918213,
-      "learning_rate": 0.0004264777502424307,
-      "loss": 3.5338,
+      "epoch": 2.9056603773584904,
+      "grad_norm": 0.5871788859367371,
+      "learning_rate": 0.0004261780895844576,
+      "loss": 3.5547,
       "step": 26950
     },
     {
-      "epoch": 2.9060381013884404,
-      "grad_norm": 0.6184528470039368,
-      "learning_rate": 0.0004261545092123693,
-      "loss": 3.5226,
+      "epoch": 2.9110512129380055,
+      "grad_norm": 0.5941997766494751,
+      "learning_rate": 0.00042585429033998917,
+      "loss": 3.5355,
       "step": 27000
     },
     {
-      "epoch": 2.9060381013884404,
-      "eval_accuracy": 0.3700053489816671,
-      "eval_loss": 3.502581834793091,
-      "eval_runtime": 180.637,
-      "eval_samples_per_second": 99.708,
-      "eval_steps_per_second": 6.233,
+      "epoch": 2.9110512129380055,
+      "eval_accuracy": 0.3690959243106599,
+      "eval_loss": 3.506704807281494,
+      "eval_runtime": 184.548,
+      "eval_samples_per_second": 97.595,
+      "eval_steps_per_second": 6.101,
       "step": 27000
     },
     {
-      "epoch": 2.9114196534280485,
-      "grad_norm": 0.6496872305870056,
-      "learning_rate": 0.00042583126818230795,
-      "loss": 3.5242,
+      "epoch": 2.91644204851752,
+      "grad_norm": 0.5948578119277954,
+      "learning_rate": 0.0004255304910955207,
+      "loss": 3.522,
       "step": 27050
     },
     {
-      "epoch": 2.9168012054676566,
-      "grad_norm": 0.6035163402557373,
-      "learning_rate": 0.0004255080271522465,
-      "loss": 3.5265,
+      "epoch": 2.921832884097035,
+      "grad_norm": 0.6063885688781738,
+      "learning_rate": 0.0004252066918510523,
+      "loss": 3.5315,
       "step": 27100
     },
     {
-      "epoch": 2.922182757507265,
-      "grad_norm": 0.5869918465614319,
-      "learning_rate": 0.0004251847861221851,
-      "loss": 3.5208,
+      "epoch": 2.92722371967655,
+      "grad_norm": 0.5720027685165405,
+      "learning_rate": 0.0004248828926065839,
+      "loss": 3.5281,
       "step": 27150
     },
     {
-      "epoch": 2.9275643095468733,
-      "grad_norm": 0.6070534586906433,
-      "learning_rate": 0.00042486154509212363,
-      "loss": 3.4932,
+      "epoch": 2.9326145552560647,
+      "grad_norm": 0.6208416819572449,
+      "learning_rate": 0.0004245590933621155,
+      "loss": 3.5356,
       "step": 27200
     },
     {
-      "epoch": 2.9329458615864814,
-      "grad_norm": 0.6247556805610657,
-      "learning_rate": 0.0004245383040620622,
-      "loss": 3.5168,
+      "epoch": 2.9380053908355794,
+      "grad_norm": 0.6040082573890686,
+      "learning_rate": 0.00042423529411764703,
+      "loss": 3.5406,
       "step": 27250
     },
     {
-      "epoch": 2.93832741362609,
-      "grad_norm": 0.606638491153717,
-      "learning_rate": 0.0004242150630320009,
-      "loss": 3.5379,
+      "epoch": 2.9433962264150946,
+      "grad_norm": 0.6055092811584473,
+      "learning_rate": 0.00042391149487317864,
+      "loss": 3.5071,
       "step": 27300
     },
     {
-      "epoch": 2.943708965665698,
-      "grad_norm": 0.5896838307380676,
-      "learning_rate": 0.0004238918220019394,
-      "loss": 3.5439,
+      "epoch": 2.9487870619946093,
+      "grad_norm": 0.5801677703857422,
+      "learning_rate": 0.00042358769562871013,
+      "loss": 3.5266,
       "step": 27350
     },
     {
-      "epoch": 2.949090517705306,
-      "grad_norm": 0.7161862850189209,
-      "learning_rate": 0.000423568580971878,
-      "loss": 3.524,
+      "epoch": 2.954177897574124,
+      "grad_norm": 0.6085225939750671,
+      "learning_rate": 0.00042326389638424174,
+      "loss": 3.5207,
       "step": 27400
     },
     {
-      "epoch": 2.9544720697449143,
-      "grad_norm": 0.5472960472106934,
-      "learning_rate": 0.00042324533994181655,
-      "loss": 3.5082,
+      "epoch": 2.9595687331536387,
+      "grad_norm": 0.6469091176986694,
+      "learning_rate": 0.0004229400971397733,
+      "loss": 3.5144,
       "step": 27450
     },
     {
-      "epoch": 2.9598536217845224,
-      "grad_norm": 0.6270456910133362,
-      "learning_rate": 0.00042292209891175514,
-      "loss": 3.5133,
+      "epoch": 2.964959568733154,
+      "grad_norm": 0.5324587821960449,
+      "learning_rate": 0.00042261629789530484,
+      "loss": 3.5401,
       "step": 27500
     },
     {
-      "epoch": 2.965235173824131,
-      "grad_norm": 0.6163716912269592,
-      "learning_rate": 0.00042259885788169374,
-      "loss": 3.5185,
+      "epoch": 2.9703504043126685,
+      "grad_norm": 0.5728937387466431,
+      "learning_rate": 0.00042229249865083644,
+      "loss": 3.5412,
       "step": 27550
     },
     {
-      "epoch": 2.970616725863739,
-      "grad_norm": 0.5813859105110168,
-      "learning_rate": 0.00042227561685163233,
-      "loss": 3.5219,
+      "epoch": 2.975741239892183,
+      "grad_norm": 0.5513703227043152,
+      "learning_rate": 0.000421968699406368,
+      "loss": 3.5235,
       "step": 27600
     },
     {
-      "epoch": 2.975998277903347,
-      "grad_norm": 0.6137823462486267,
-      "learning_rate": 0.00042195237582157093,
-      "loss": 3.5238,
+      "epoch": 2.981132075471698,
+      "grad_norm": 0.6015611290931702,
+      "learning_rate": 0.0004216449001618996,
+      "loss": 3.5281,
       "step": 27650
     },
     {
-      "epoch": 2.9813798299429557,
-      "grad_norm": 0.573917031288147,
-      "learning_rate": 0.0004216291347915095,
-      "loss": 3.531,
+      "epoch": 2.986522911051213,
+      "grad_norm": 0.6475524306297302,
+      "learning_rate": 0.00042132110091743115,
+      "loss": 3.5199,
       "step": 27700
     },
     {
-      "epoch": 2.986761381982564,
-      "grad_norm": 0.6336318254470825,
-      "learning_rate": 0.00042130589376144806,
-      "loss": 3.5142,
+      "epoch": 2.9919137466307277,
+      "grad_norm": 0.5860631465911865,
+      "learning_rate": 0.00042099730167296275,
+      "loss": 3.5175,
       "step": 27750
     },
     {
-      "epoch": 2.992142934022172,
-      "grad_norm": 0.5717305541038513,
-      "learning_rate": 0.00042098265273138666,
-      "loss": 3.5073,
+      "epoch": 2.9973045822102424,
+      "grad_norm": 0.6016190052032471,
+      "learning_rate": 0.0004206735024284943,
+      "loss": 3.5275,
       "step": 27800
     },
     {
-      "epoch": 2.9975244860617805,
-      "grad_norm": 0.6153519749641418,
-      "learning_rate": 0.0004206594117013252,
-      "loss": 3.5313,
+      "epoch": 3.0026954177897576,
+      "grad_norm": 0.5374354124069214,
+      "learning_rate": 0.0004203497031840259,
+      "loss": 3.4869,
       "step": 27850
     },
     {
-      "epoch": 3.0029060381013886,
-      "grad_norm": 0.5966550707817078,
-      "learning_rate": 0.00042033617067126385,
-      "loss": 3.4862,
+      "epoch": 3.0080862533692723,
+      "grad_norm": 0.6027006506919861,
+      "learning_rate": 0.00042002590393955746,
+      "loss": 3.4352,
       "step": 27900
     },
     {
-      "epoch": 3.0082875901409967,
-      "grad_norm": 0.5811104774475098,
-      "learning_rate": 0.00042001292964120244,
-      "loss": 3.4131,
+      "epoch": 3.013477088948787,
+      "grad_norm": 0.6323424577713013,
+      "learning_rate": 0.00041970210469508896,
+      "loss": 3.4338,
       "step": 27950
     },
     {
-      "epoch": 3.0136691421806048,
-      "grad_norm": 0.5833897590637207,
-      "learning_rate": 0.000419689688611141,
-      "loss": 3.4297,
+      "epoch": 3.018867924528302,
+      "grad_norm": 0.5759109258651733,
+      "learning_rate": 0.00041937830545062056,
+      "loss": 3.4341,
       "step": 28000
     },
     {
-      "epoch": 3.0136691421806048,
-      "eval_accuracy": 0.3707112667937978,
-      "eval_loss": 3.4965574741363525,
-      "eval_runtime": 180.5598,
-      "eval_samples_per_second": 99.751,
-      "eval_steps_per_second": 6.236,
+      "epoch": 3.018867924528302,
+      "eval_accuracy": 0.36992527180334783,
+      "eval_loss": 3.500993251800537,
+      "eval_runtime": 185.1029,
+      "eval_samples_per_second": 97.303,
+      "eval_steps_per_second": 6.083,
       "step": 28000
     },
     {
-      "epoch": 3.0190506942202133,
-      "grad_norm": 0.6142577528953552,
-      "learning_rate": 0.0004193664475810796,
-      "loss": 3.4263,
+      "epoch": 3.024258760107817,
+      "grad_norm": 0.5716021060943604,
+      "learning_rate": 0.0004190545062061521,
+      "loss": 3.4337,
       "step": 28050
     },
     {
-      "epoch": 3.0244322462598214,
-      "grad_norm": 0.60226970911026,
-      "learning_rate": 0.00041904320655101817,
-      "loss": 3.4323,
+      "epoch": 3.0296495956873315,
+      "grad_norm": 0.6018630266189575,
+      "learning_rate": 0.0004187307069616837,
+      "loss": 3.4388,
       "step": 28100
     },
     {
-      "epoch": 3.0298137982994295,
-      "grad_norm": 0.6177240014076233,
-      "learning_rate": 0.0004187199655209567,
-      "loss": 3.4357,
+      "epoch": 3.035040431266846,
+      "grad_norm": 0.6063191890716553,
+      "learning_rate": 0.00041840690771721527,
+      "loss": 3.423,
       "step": 28150
     },
     {
-      "epoch": 3.0351953503390376,
-      "grad_norm": 0.5931535363197327,
-      "learning_rate": 0.00041839672449089536,
-      "loss": 3.4589,
+      "epoch": 3.0404312668463613,
+      "grad_norm": 0.5545841455459595,
+      "learning_rate": 0.00041808310847274687,
+      "loss": 3.4414,
       "step": 28200
     },
     {
-      "epoch": 3.040576902378646,
-      "grad_norm": 0.608650803565979,
-      "learning_rate": 0.00041807348346083395,
-      "loss": 3.4432,
+      "epoch": 3.045822102425876,
+      "grad_norm": 0.5774965882301331,
+      "learning_rate": 0.0004177593092282784,
+      "loss": 3.4324,
       "step": 28250
     },
     {
-      "epoch": 3.0459584544182543,
-      "grad_norm": 0.6467320322990417,
-      "learning_rate": 0.0004177502424307725,
-      "loss": 3.4356,
+      "epoch": 3.0512129380053907,
+      "grad_norm": 0.5643296837806702,
+      "learning_rate": 0.0004174419859686994,
+      "loss": 3.455,
       "step": 28300
     },
     {
-      "epoch": 3.0513400064578624,
-      "grad_norm": 0.6208910346031189,
-      "learning_rate": 0.0004174270014007111,
-      "loss": 3.4335,
+      "epoch": 3.056603773584906,
+      "grad_norm": 0.6419238448143005,
+      "learning_rate": 0.0004171181867242309,
+      "loss": 3.4396,
       "step": 28350
     },
     {
-      "epoch": 3.0567215584974705,
-      "grad_norm": 0.6012397408485413,
-      "learning_rate": 0.00041710376037064963,
-      "loss": 3.4368,
+      "epoch": 3.0619946091644206,
+      "grad_norm": 0.5539114475250244,
+      "learning_rate": 0.00041679438747976253,
+      "loss": 3.4607,
       "step": 28400
     },
     {
-      "epoch": 3.062103110537079,
-      "grad_norm": 0.5883954763412476,
-      "learning_rate": 0.0004167805193405883,
-      "loss": 3.4407,
+      "epoch": 3.0673854447439353,
+      "grad_norm": 0.575046956539154,
+      "learning_rate": 0.0004164705882352941,
+      "loss": 3.4423,
       "step": 28450
     },
     {
-      "epoch": 3.067484662576687,
-      "grad_norm": 0.5667783617973328,
-      "learning_rate": 0.0004164572783105269,
-      "loss": 3.4358,
+      "epoch": 3.07277628032345,
+      "grad_norm": 0.5910366177558899,
+      "learning_rate": 0.0004161467889908257,
+      "loss": 3.4337,
       "step": 28500
     },
     {
-      "epoch": 3.0728662146162953,
-      "grad_norm": 0.6078146696090698,
-      "learning_rate": 0.0004161340372804654,
-      "loss": 3.4422,
+      "epoch": 3.078167115902965,
+      "grad_norm": 0.6324418187141418,
+      "learning_rate": 0.00041582298974635724,
+      "loss": 3.4346,
       "step": 28550
     },
     {
-      "epoch": 3.0782477666559034,
-      "grad_norm": 0.6011075377464294,
-      "learning_rate": 0.000415810796250404,
-      "loss": 3.4444,
+      "epoch": 3.08355795148248,
+      "grad_norm": 0.5757888555526733,
+      "learning_rate": 0.00041549919050188884,
+      "loss": 3.4546,
       "step": 28600
     },
     {
-      "epoch": 3.083629318695512,
-      "grad_norm": 0.6752050518989563,
-      "learning_rate": 0.0004154875552203426,
-      "loss": 3.4557,
+      "epoch": 3.0889487870619945,
+      "grad_norm": 0.6831218600273132,
+      "learning_rate": 0.00041517539125742034,
+      "loss": 3.4394,
       "step": 28650
     },
     {
-      "epoch": 3.08901087073512,
-      "grad_norm": 0.6102807521820068,
-      "learning_rate": 0.00041516431419028114,
-      "loss": 3.4272,
+      "epoch": 3.0943396226415096,
+      "grad_norm": 0.575617253780365,
+      "learning_rate": 0.0004148515920129519,
+      "loss": 3.4627,
       "step": 28700
     },
     {
-      "epoch": 3.094392422774728,
-      "grad_norm": 0.6542812585830688,
-      "learning_rate": 0.0004148410731602198,
-      "loss": 3.4495,
+      "epoch": 3.0997304582210243,
+      "grad_norm": 0.5726577043533325,
+      "learning_rate": 0.0004145277927684835,
+      "loss": 3.4267,
       "step": 28750
     },
     {
-      "epoch": 3.0997739748143363,
-      "grad_norm": 0.6202298998832703,
-      "learning_rate": 0.0004145178321301584,
-      "loss": 3.4493,
+      "epoch": 3.105121293800539,
+      "grad_norm": 0.5732120871543884,
+      "learning_rate": 0.00041420399352401504,
+      "loss": 3.4551,
       "step": 28800
     },
     {
-      "epoch": 3.105155526853945,
-      "grad_norm": 0.6084877848625183,
-      "learning_rate": 0.00041419459110009693,
-      "loss": 3.4437,
+      "epoch": 3.1105121293800537,
+      "grad_norm": 0.6304787397384644,
+      "learning_rate": 0.00041388019427954665,
+      "loss": 3.4438,
       "step": 28850
     },
     {
-      "epoch": 3.110537078893553,
-      "grad_norm": 0.8142428398132324,
-      "learning_rate": 0.0004138713500700355,
-      "loss": 3.4456,
+      "epoch": 3.115902964959569,
+      "grad_norm": 0.5989034175872803,
+      "learning_rate": 0.0004135563950350782,
+      "loss": 3.4592,
       "step": 28900
     },
     {
-      "epoch": 3.115918630933161,
-      "grad_norm": 0.6456471681594849,
-      "learning_rate": 0.00041354810903997406,
-      "loss": 3.4468,
+      "epoch": 3.1212938005390836,
+      "grad_norm": 0.5832788348197937,
+      "learning_rate": 0.0004132325957906098,
+      "loss": 3.4562,
       "step": 28950
     },
     {
-      "epoch": 3.121300182972769,
-      "grad_norm": 0.6340752840042114,
-      "learning_rate": 0.00041322486800991266,
-      "loss": 3.4224,
+      "epoch": 3.1266846361185983,
+      "grad_norm": 0.592212975025177,
+      "learning_rate": 0.00041290879654614135,
+      "loss": 3.4521,
       "step": 29000
     },
     {
-      "epoch": 3.121300182972769,
-      "eval_accuracy": 0.37119194716948906,
-      "eval_loss": 3.4933841228485107,
-      "eval_runtime": 180.305,
-      "eval_samples_per_second": 99.892,
-      "eval_steps_per_second": 6.245,
+      "epoch": 3.1266846361185983,
+      "eval_accuracy": 0.37072582628076495,
+      "eval_loss": 3.496122121810913,
+      "eval_runtime": 184.6302,
+      "eval_samples_per_second": 97.552,
+      "eval_steps_per_second": 6.099,
       "step": 29000
     },
     {
-      "epoch": 3.1266817350123777,
-      "grad_norm": 0.6109960675239563,
-      "learning_rate": 0.0004129016269798513,
-      "loss": 3.4461,
+      "epoch": 3.1320754716981134,
+      "grad_norm": 0.5908344388008118,
+      "learning_rate": 0.00041258499730167296,
+      "loss": 3.4568,
       "step": 29050
     },
     {
-      "epoch": 3.132063287051986,
-      "grad_norm": 0.5882677435874939,
-      "learning_rate": 0.00041257838594978985,
-      "loss": 3.4344,
+      "epoch": 3.137466307277628,
+      "grad_norm": 0.5698777437210083,
+      "learning_rate": 0.0004122611980572045,
+      "loss": 3.4543,
       "step": 29100
     },
     {
-      "epoch": 3.137444839091594,
-      "grad_norm": 0.5873978734016418,
-      "learning_rate": 0.00041225514491972844,
-      "loss": 3.4492,
+      "epoch": 3.142857142857143,
+      "grad_norm": 0.5800032615661621,
+      "learning_rate": 0.00041193739881273606,
+      "loss": 3.46,
       "step": 29150
     },
     {
-      "epoch": 3.1428263911312024,
-      "grad_norm": 0.6426523923873901,
-      "learning_rate": 0.00041193190388966704,
-      "loss": 3.4381,
+      "epoch": 3.1482479784366575,
+      "grad_norm": 0.583019495010376,
+      "learning_rate": 0.00041161359956826766,
+      "loss": 3.455,
       "step": 29200
     },
     {
-      "epoch": 3.1482079431708105,
-      "grad_norm": 0.5868753790855408,
-      "learning_rate": 0.0004116086628596056,
-      "loss": 3.4539,
+      "epoch": 3.1536388140161726,
+      "grad_norm": 0.6571466326713562,
+      "learning_rate": 0.0004112898003237992,
+      "loss": 3.473,
       "step": 29250
     },
     {
-      "epoch": 3.1535894952104186,
-      "grad_norm": 0.6552289128303528,
-      "learning_rate": 0.0004112854218295442,
-      "loss": 3.4576,
+      "epoch": 3.1590296495956873,
+      "grad_norm": 0.6302173137664795,
+      "learning_rate": 0.0004109660010793308,
+      "loss": 3.4594,
       "step": 29300
     },
     {
-      "epoch": 3.1589710472500268,
-      "grad_norm": 0.6329394578933716,
-      "learning_rate": 0.0004109621807994828,
-      "loss": 3.4514,
+      "epoch": 3.164420485175202,
+      "grad_norm": 0.6568088531494141,
+      "learning_rate": 0.0004106422018348623,
+      "loss": 3.453,
       "step": 29350
     },
     {
-      "epoch": 3.1643525992896353,
-      "grad_norm": 0.6486338376998901,
-      "learning_rate": 0.00041063893976942136,
-      "loss": 3.4588,
+      "epoch": 3.169811320754717,
+      "grad_norm": 0.6203985810279846,
+      "learning_rate": 0.0004103184025903939,
+      "loss": 3.449,
       "step": 29400
     },
     {
-      "epoch": 3.1697341513292434,
-      "grad_norm": 0.5754224061965942,
-      "learning_rate": 0.00041031569873935996,
-      "loss": 3.4392,
+      "epoch": 3.175202156334232,
+      "grad_norm": 0.6170281767845154,
+      "learning_rate": 0.00040999460334592547,
+      "loss": 3.4603,
       "step": 29450
     },
     {
-      "epoch": 3.1751157033688515,
-      "grad_norm": 0.6380184888839722,
-      "learning_rate": 0.0004099924577092985,
-      "loss": 3.4547,
+      "epoch": 3.1805929919137466,
+      "grad_norm": 0.5898356437683105,
+      "learning_rate": 0.000409670804101457,
+      "loss": 3.4562,
       "step": 29500
     },
     {
-      "epoch": 3.1804972554084596,
-      "grad_norm": 0.6305249333381653,
-      "learning_rate": 0.0004096692166792371,
-      "loss": 3.4568,
+      "epoch": 3.1859838274932613,
+      "grad_norm": 0.670875608921051,
+      "learning_rate": 0.00040934700485698863,
+      "loss": 3.4667,
       "step": 29550
     },
     {
-      "epoch": 3.185878807448068,
-      "grad_norm": 0.6137855648994446,
-      "learning_rate": 0.00040934597564917574,
+      "epoch": 3.1913746630727764,
+      "grad_norm": 0.6004409790039062,
+      "learning_rate": 0.0004090232056125202,
       "loss": 3.4612,
       "step": 29600
     },
     {
-      "epoch": 3.1912603594876763,
-      "grad_norm": 0.5795332789421082,
-      "learning_rate": 0.0004090227346191143,
-      "loss": 3.4387,
+      "epoch": 3.196765498652291,
+      "grad_norm": 0.6314196586608887,
+      "learning_rate": 0.0004086994063680518,
+      "loss": 3.4508,
       "step": 29650
     },
     {
-      "epoch": 3.1966419115272844,
-      "grad_norm": 0.6044164299964905,
-      "learning_rate": 0.0004086994935890529,
-      "loss": 3.4696,
+      "epoch": 3.202156334231806,
+      "grad_norm": 0.5917602181434631,
+      "learning_rate": 0.00040837560712358333,
+      "loss": 3.4829,
       "step": 29700
     },
     {
-      "epoch": 3.2020234635668925,
-      "grad_norm": 0.60833340883255,
-      "learning_rate": 0.00040837625255899147,
-      "loss": 3.4597,
+      "epoch": 3.207547169811321,
+      "grad_norm": 0.6115114688873291,
+      "learning_rate": 0.00040805180787911494,
+      "loss": 3.4527,
       "step": 29750
     },
     {
-      "epoch": 3.207405015606501,
-      "grad_norm": 0.6233122944831848,
-      "learning_rate": 0.00040805301152893,
-      "loss": 3.4374,
+      "epoch": 3.2129380053908356,
+      "grad_norm": 0.6146586537361145,
+      "learning_rate": 0.0004077280086346465,
+      "loss": 3.4625,
       "step": 29800
     },
     {
-      "epoch": 3.212786567646109,
-      "grad_norm": 0.6239629983901978,
-      "learning_rate": 0.0004077297704988686,
-      "loss": 3.4452,
+      "epoch": 3.2183288409703503,
+      "grad_norm": 0.5785256624221802,
+      "learning_rate": 0.0004074042093901781,
+      "loss": 3.4397,
       "step": 29850
     },
     {
-      "epoch": 3.2181681196857173,
-      "grad_norm": 0.6396247148513794,
-      "learning_rate": 0.00040740652946880725,
-      "loss": 3.4325,
+      "epoch": 3.223719676549865,
+      "grad_norm": 0.5921712517738342,
+      "learning_rate": 0.00040708041014570964,
+      "loss": 3.4511,
       "step": 29900
     },
     {
-      "epoch": 3.2235496717253254,
-      "grad_norm": 0.58219975233078,
-      "learning_rate": 0.0004070832884387458,
-      "loss": 3.466,
+      "epoch": 3.22911051212938,
+      "grad_norm": 0.5875865817070007,
+      "learning_rate": 0.00040675661090124114,
+      "loss": 3.464,
       "step": 29950
     },
     {
-      "epoch": 3.228931223764934,
-      "grad_norm": 0.6302553415298462,
-      "learning_rate": 0.0004067600474086844,
-      "loss": 3.456,
+      "epoch": 3.234501347708895,
+      "grad_norm": 0.5804892182350159,
+      "learning_rate": 0.0004064328116567728,
+      "loss": 3.4666,
       "step": 30000
     },
     {
-      "epoch": 3.228931223764934,
-      "eval_accuracy": 0.3716267760265226,
-      "eval_loss": 3.489248275756836,
-      "eval_runtime": 180.7912,
-      "eval_samples_per_second": 99.623,
-      "eval_steps_per_second": 6.228,
+      "epoch": 3.234501347708895,
+      "eval_accuracy": 0.37127311087668646,
+      "eval_loss": 3.491586923599243,
+      "eval_runtime": 185.2775,
+      "eval_samples_per_second": 97.211,
+      "eval_steps_per_second": 6.077,
       "step": 30000
     },
     {
-      "epoch": 3.234312775804542,
-      "grad_norm": 0.6658568382263184,
-      "learning_rate": 0.00040643680637862293,
-      "loss": 3.4516,
+      "epoch": 3.2398921832884096,
+      "grad_norm": 0.6126497983932495,
+      "learning_rate": 0.0004061090124123043,
+      "loss": 3.4724,
       "step": 30050
     },
     {
-      "epoch": 3.23969432784415,
-      "grad_norm": 0.6323238611221313,
-      "learning_rate": 0.0004061135653485615,
-      "loss": 3.457,
+      "epoch": 3.2452830188679247,
+      "grad_norm": 0.5532984733581543,
+      "learning_rate": 0.0004057852131678359,
+      "loss": 3.4516,
       "step": 30100
     },
     {
-      "epoch": 3.2450758798837587,
-      "grad_norm": 0.5995866060256958,
-      "learning_rate": 0.00040579678913910133,
-      "loss": 3.4326,
+      "epoch": 3.2506738544474394,
+      "grad_norm": 0.6137168407440186,
+      "learning_rate": 0.00040546141392336745,
+      "loss": 3.4561,
       "step": 30150
     },
     {
-      "epoch": 3.250457431923367,
-      "grad_norm": 0.6197038888931274,
-      "learning_rate": 0.0004054735481090399,
-      "loss": 3.4274,
+      "epoch": 3.256064690026954,
+      "grad_norm": 0.6394794583320618,
+      "learning_rate": 0.00040513761467889906,
+      "loss": 3.4598,
       "step": 30200
     },
     {
-      "epoch": 3.255838983962975,
-      "grad_norm": 0.5652536749839783,
-      "learning_rate": 0.0004051503070789786,
-      "loss": 3.4548,
+      "epoch": 3.2614555256064692,
+      "grad_norm": 0.600420355796814,
+      "learning_rate": 0.0004048138154344306,
+      "loss": 3.4413,
       "step": 30250
     },
     {
-      "epoch": 3.261220536002583,
-      "grad_norm": 0.6363932490348816,
-      "learning_rate": 0.0004048270660489171,
-      "loss": 3.4593,
+      "epoch": 3.266846361185984,
+      "grad_norm": 0.6034658551216125,
+      "learning_rate": 0.0004044900161899622,
+      "loss": 3.4709,
       "step": 30300
     },
     {
-      "epoch": 3.2666020880421915,
-      "grad_norm": 0.6436594128608704,
-      "learning_rate": 0.0004045038250188557,
-      "loss": 3.4508,
+      "epoch": 3.2722371967654986,
+      "grad_norm": 0.5927817225456238,
+      "learning_rate": 0.0004041726929303831,
+      "loss": 3.4559,
       "step": 30350
     },
     {
-      "epoch": 3.2719836400817996,
-      "grad_norm": 0.627250075340271,
-      "learning_rate": 0.00040418058398879425,
-      "loss": 3.4444,
+      "epoch": 3.2776280323450133,
+      "grad_norm": 0.6020606756210327,
+      "learning_rate": 0.0004038488936859147,
+      "loss": 3.4567,
       "step": 30400
     },
     {
-      "epoch": 3.2773651921214078,
-      "grad_norm": 0.659887433052063,
-      "learning_rate": 0.00040385734295873284,
-      "loss": 3.4687,
+      "epoch": 3.2830188679245285,
+      "grad_norm": 0.6315540671348572,
+      "learning_rate": 0.00040352509444144626,
+      "loss": 3.4581,
       "step": 30450
     },
     {
-      "epoch": 3.282746744161016,
-      "grad_norm": 0.5804165005683899,
-      "learning_rate": 0.00040353410192867144,
-      "loss": 3.4656,
+      "epoch": 3.288409703504043,
+      "grad_norm": 0.6471781730651855,
+      "learning_rate": 0.00040320129519697787,
+      "loss": 3.4872,
       "step": 30500
     },
     {
-      "epoch": 3.2881282962006244,
-      "grad_norm": 0.6306347846984863,
-      "learning_rate": 0.00040321086089861003,
-      "loss": 3.4417,
+      "epoch": 3.293800539083558,
+      "grad_norm": 0.6469458937644958,
+      "learning_rate": 0.0004028774959525094,
+      "loss": 3.4568,
       "step": 30550
     },
     {
-      "epoch": 3.2935098482402325,
-      "grad_norm": 0.6350857019424438,
-      "learning_rate": 0.00040288761986854863,
-      "loss": 3.4682,
+      "epoch": 3.2991913746630726,
+      "grad_norm": 0.6129242181777954,
+      "learning_rate": 0.000402553696708041,
+      "loss": 3.4642,
       "step": 30600
     },
     {
-      "epoch": 3.2988914002798406,
-      "grad_norm": 0.7076759338378906,
-      "learning_rate": 0.0004025643788384872,
-      "loss": 3.4531,
+      "epoch": 3.3045822102425877,
+      "grad_norm": 0.6130914092063904,
+      "learning_rate": 0.0004022298974635726,
+      "loss": 3.4538,
       "step": 30650
     },
     {
-      "epoch": 3.304272952319449,
-      "grad_norm": 0.6139525175094604,
-      "learning_rate": 0.00040224113780842576,
-      "loss": 3.4659,
+      "epoch": 3.3099730458221024,
+      "grad_norm": 0.6233851313591003,
+      "learning_rate": 0.00040190609821910407,
+      "loss": 3.4567,
       "step": 30700
     },
     {
-      "epoch": 3.3096545043590573,
-      "grad_norm": 0.6305450201034546,
-      "learning_rate": 0.00040191789677836436,
-      "loss": 3.4443,
+      "epoch": 3.315363881401617,
+      "grad_norm": 0.6533472537994385,
+      "learning_rate": 0.0004015822989746357,
+      "loss": 3.445,
       "step": 30750
     },
     {
-      "epoch": 3.3150360563986654,
-      "grad_norm": 0.5848320722579956,
-      "learning_rate": 0.0004015946557483029,
-      "loss": 3.453,
+      "epoch": 3.3207547169811322,
+      "grad_norm": 0.6035603284835815,
+      "learning_rate": 0.0004012584997301672,
+      "loss": 3.4614,
       "step": 30800
     },
     {
-      "epoch": 3.3204176084382735,
-      "grad_norm": 0.6427562236785889,
-      "learning_rate": 0.00040127141471824155,
-      "loss": 3.4731,
+      "epoch": 3.326145552560647,
+      "grad_norm": 0.6911303997039795,
+      "learning_rate": 0.00040093470048569883,
+      "loss": 3.4716,
       "step": 30850
     },
     {
-      "epoch": 3.3257991604778816,
-      "grad_norm": 0.6468438506126404,
-      "learning_rate": 0.00040094817368818014,
-      "loss": 3.4487,
+      "epoch": 3.3315363881401616,
+      "grad_norm": 0.6303085088729858,
+      "learning_rate": 0.0004006109012412304,
+      "loss": 3.4612,
       "step": 30900
     },
     {
-      "epoch": 3.33118071251749,
-      "grad_norm": 0.6019175052642822,
-      "learning_rate": 0.0004006249326581187,
-      "loss": 3.4629,
+      "epoch": 3.3369272237196768,
+      "grad_norm": 0.6187211871147156,
+      "learning_rate": 0.000400287101996762,
+      "loss": 3.4651,
       "step": 30950
     },
     {
-      "epoch": 3.3365622645570983,
-      "grad_norm": 0.6581705212593079,
-      "learning_rate": 0.0004003016916280573,
-      "loss": 3.4661,
+      "epoch": 3.3423180592991915,
+      "grad_norm": 0.5940481424331665,
+      "learning_rate": 0.00039996330275229354,
+      "loss": 3.4547,
       "step": 31000
     },
     {
-      "epoch": 3.3365622645570983,
-      "eval_accuracy": 0.37244656206508286,
-      "eval_loss": 3.4832379817962646,
-      "eval_runtime": 180.6125,
-      "eval_samples_per_second": 99.722,
-      "eval_steps_per_second": 6.234,
+      "epoch": 3.3423180592991915,
+      "eval_accuracy": 0.37221404488515986,
+      "eval_loss": 3.4880588054656982,
+      "eval_runtime": 184.5498,
+      "eval_samples_per_second": 97.594,
+      "eval_steps_per_second": 6.101,
       "step": 31000
     },
     {
-      "epoch": 3.3419438165967064,
-      "grad_norm": 0.65335613489151,
-      "learning_rate": 0.00039997845059799587,
-      "loss": 3.4359,
+      "epoch": 3.347708894878706,
+      "grad_norm": 0.567207396030426,
+      "learning_rate": 0.00039963950350782514,
+      "loss": 3.4654,
       "step": 31050
     },
     {
-      "epoch": 3.347325368636315,
-      "grad_norm": 0.6507302522659302,
-      "learning_rate": 0.00039965520956793447,
-      "loss": 3.4525,
+      "epoch": 3.353099730458221,
+      "grad_norm": 0.6383653879165649,
+      "learning_rate": 0.0003993157042633567,
+      "loss": 3.4536,
       "step": 31100
     },
     {
-      "epoch": 3.352706920675923,
-      "grad_norm": 0.5894781351089478,
-      "learning_rate": 0.00039933196853787306,
-      "loss": 3.4516,
+      "epoch": 3.358490566037736,
+      "grad_norm": 0.6113681197166443,
+      "learning_rate": 0.00039899190501888824,
+      "loss": 3.4568,
       "step": 31150
     },
     {
-      "epoch": 3.358088472715531,
-      "grad_norm": 0.6025640964508057,
-      "learning_rate": 0.00039900872750781166,
-      "loss": 3.4648,
+      "epoch": 3.3638814016172507,
+      "grad_norm": 0.5768980979919434,
+      "learning_rate": 0.00039866810577441985,
+      "loss": 3.4645,
       "step": 31200
     },
     {
-      "epoch": 3.3634700247551392,
-      "grad_norm": 0.6366141438484192,
-      "learning_rate": 0.0003986854864777502,
-      "loss": 3.4504,
+      "epoch": 3.3692722371967654,
+      "grad_norm": 0.5896947979927063,
+      "learning_rate": 0.0003983443065299514,
+      "loss": 3.4555,
       "step": 31250
     },
     {
-      "epoch": 3.368851576794748,
-      "grad_norm": 0.6077277064323425,
-      "learning_rate": 0.0003983622454476888,
-      "loss": 3.4439,
+      "epoch": 3.37466307277628,
+      "grad_norm": 0.6593387126922607,
+      "learning_rate": 0.000398020507285483,
+      "loss": 3.4606,
       "step": 31300
     },
     {
-      "epoch": 3.374233128834356,
-      "grad_norm": 0.6532594561576843,
-      "learning_rate": 0.00039803900441762733,
-      "loss": 3.4624,
+      "epoch": 3.3800539083557952,
+      "grad_norm": 0.6088037490844727,
+      "learning_rate": 0.0003976967080410145,
+      "loss": 3.4654,
       "step": 31350
     },
     {
-      "epoch": 3.379614680873964,
-      "grad_norm": 0.6048536896705627,
-      "learning_rate": 0.000397715763387566,
-      "loss": 3.4589,
+      "epoch": 3.38544474393531,
+      "grad_norm": 0.585678219795227,
+      "learning_rate": 0.0003973729087965461,
+      "loss": 3.4782,
       "step": 31400
     },
     {
-      "epoch": 3.384996232913572,
-      "grad_norm": 0.6351385712623596,
-      "learning_rate": 0.0003973925223575046,
-      "loss": 3.4395,
+      "epoch": 3.3908355795148246,
+      "grad_norm": 0.6365036368370056,
+      "learning_rate": 0.00039704910955207765,
+      "loss": 3.4588,
       "step": 31450
     },
     {
-      "epoch": 3.3903777849531807,
-      "grad_norm": 0.6504836082458496,
-      "learning_rate": 0.0003970692813274431,
-      "loss": 3.4588,
+      "epoch": 3.3962264150943398,
+      "grad_norm": 0.5940556526184082,
+      "learning_rate": 0.00039672531030760926,
+      "loss": 3.4557,
       "step": 31500
     },
     {
-      "epoch": 3.3957593369927888,
-      "grad_norm": 0.6237932443618774,
-      "learning_rate": 0.0003967460402973817,
-      "loss": 3.4532,
+      "epoch": 3.4016172506738545,
+      "grad_norm": 0.6067618727684021,
+      "learning_rate": 0.0003964015110631408,
+      "loss": 3.4426,
       "step": 31550
     },
     {
-      "epoch": 3.401140889032397,
-      "grad_norm": 0.5831241011619568,
-      "learning_rate": 0.0003964227992673203,
-      "loss": 3.447,
+      "epoch": 3.407008086253369,
+      "grad_norm": 0.6328786015510559,
+      "learning_rate": 0.00039607771181867236,
+      "loss": 3.4521,
       "step": 31600
     },
     {
-      "epoch": 3.4065224410720054,
-      "grad_norm": 0.6177220344543457,
-      "learning_rate": 0.00039609955823725884,
-      "loss": 3.4438,
+      "epoch": 3.4123989218328843,
+      "grad_norm": 0.5680471658706665,
+      "learning_rate": 0.00039575391257420397,
+      "loss": 3.4677,
       "step": 31650
     },
     {
-      "epoch": 3.4119039931116135,
-      "grad_norm": 0.6046972870826721,
-      "learning_rate": 0.0003957763172071975,
-      "loss": 3.4752,
+      "epoch": 3.417789757412399,
+      "grad_norm": 0.6207356452941895,
+      "learning_rate": 0.0003954301133297355,
+      "loss": 3.4535,
       "step": 31700
     },
     {
-      "epoch": 3.4172855451512216,
-      "grad_norm": 0.6035352349281311,
-      "learning_rate": 0.0003954530761771361,
-      "loss": 3.4588,
+      "epoch": 3.4231805929919137,
+      "grad_norm": 0.6631540656089783,
+      "learning_rate": 0.0003951063140852671,
+      "loss": 3.4616,
       "step": 31750
     },
     {
-      "epoch": 3.4226670971908297,
-      "grad_norm": 0.5911862850189209,
-      "learning_rate": 0.00039512983514707463,
-      "loss": 3.4505,
+      "epoch": 3.4285714285714284,
+      "grad_norm": 0.6220523118972778,
+      "learning_rate": 0.00039478251484079867,
+      "loss": 3.4612,
       "step": 31800
     },
     {
-      "epoch": 3.428048649230438,
-      "grad_norm": 0.5918540358543396,
-      "learning_rate": 0.0003948065941170132,
-      "loss": 3.4555,
+      "epoch": 3.4339622641509435,
+      "grad_norm": 0.5873520374298096,
+      "learning_rate": 0.0003944587155963303,
+      "loss": 3.4504,
       "step": 31850
     },
     {
-      "epoch": 3.4334302012700464,
-      "grad_norm": 0.6266008019447327,
-      "learning_rate": 0.00039448335308695176,
-      "loss": 3.4768,
+      "epoch": 3.439353099730458,
+      "grad_norm": 0.5860127806663513,
+      "learning_rate": 0.0003941349163518618,
+      "loss": 3.4518,
       "step": 31900
     },
     {
-      "epoch": 3.4388117533096545,
-      "grad_norm": 0.6207110285758972,
-      "learning_rate": 0.00039416011205689036,
-      "loss": 3.4568,
+      "epoch": 3.444743935309973,
+      "grad_norm": 0.5693365931510925,
+      "learning_rate": 0.00039381111710739343,
+      "loss": 3.4594,
       "step": 31950
     },
     {
-      "epoch": 3.4441933053492626,
-      "grad_norm": 0.6404582858085632,
-      "learning_rate": 0.000393836871026829,
-      "loss": 3.4577,
+      "epoch": 3.450134770889488,
+      "grad_norm": 0.6177678108215332,
+      "learning_rate": 0.000393487317862925,
+      "loss": 3.4656,
       "step": 32000
     },
     {
-      "epoch": 3.4441933053492626,
-      "eval_accuracy": 0.37327199805380945,
-      "eval_loss": 3.4748642444610596,
-      "eval_runtime": 180.6631,
-      "eval_samples_per_second": 99.694,
-      "eval_steps_per_second": 6.233,
+      "epoch": 3.450134770889488,
+      "eval_accuracy": 0.37241483542184106,
+      "eval_loss": 3.479771375656128,
+      "eval_runtime": 184.9587,
+      "eval_samples_per_second": 97.378,
+      "eval_steps_per_second": 6.088,
       "step": 32000
     },
     {
-      "epoch": 3.449574857388871,
-      "grad_norm": 0.6518675088882446,
-      "learning_rate": 0.00039351362999676755,
-      "loss": 3.4445,
+      "epoch": 3.4555256064690028,
+      "grad_norm": 0.5787439346313477,
+      "learning_rate": 0.0003931635186184565,
+      "loss": 3.4585,
       "step": 32050
     },
     {
-      "epoch": 3.4549564094284793,
-      "grad_norm": 0.6538469195365906,
-      "learning_rate": 0.00039319038896670614,
-      "loss": 3.4755,
+      "epoch": 3.4609164420485174,
+      "grad_norm": 0.5935786962509155,
+      "learning_rate": 0.0003928397193739881,
+      "loss": 3.4594,
       "step": 32100
     },
     {
-      "epoch": 3.4603379614680874,
-      "grad_norm": 0.5812484622001648,
-      "learning_rate": 0.00039287361275724595,
-      "loss": 3.4689,
+      "epoch": 3.466307277628032,
+      "grad_norm": 0.5919837355613708,
+      "learning_rate": 0.00039251592012951963,
+      "loss": 3.4575,
       "step": 32150
     },
     {
-      "epoch": 3.4657195135076955,
-      "grad_norm": 0.6309418678283691,
-      "learning_rate": 0.00039255037172718454,
-      "loss": 3.4447,
+      "epoch": 3.4716981132075473,
+      "grad_norm": 0.6029905080795288,
+      "learning_rate": 0.00039219212088505124,
+      "loss": 3.4581,
       "step": 32200
     },
     {
-      "epoch": 3.471101065547304,
-      "grad_norm": 0.6192950010299683,
-      "learning_rate": 0.0003922271306971231,
-      "loss": 3.4599,
+      "epoch": 3.477088948787062,
+      "grad_norm": 0.6316198706626892,
+      "learning_rate": 0.0003918683216405828,
+      "loss": 3.4671,
       "step": 32250
     },
     {
-      "epoch": 3.476482617586912,
-      "grad_norm": 0.6055722832679749,
-      "learning_rate": 0.0003919038896670617,
-      "loss": 3.462,
+      "epoch": 3.4824797843665767,
+      "grad_norm": 0.6068309545516968,
+      "learning_rate": 0.0003915445223961144,
+      "loss": 3.4777,
       "step": 32300
     },
     {
-      "epoch": 3.4818641696265202,
-      "grad_norm": 0.6457564234733582,
-      "learning_rate": 0.00039158064863700033,
-      "loss": 3.4393,
+      "epoch": 3.487870619946092,
+      "grad_norm": 0.6286913752555847,
+      "learning_rate": 0.00039122072315164594,
+      "loss": 3.4624,
       "step": 32350
     },
     {
-      "epoch": 3.4872457216661283,
-      "grad_norm": 0.6618093848228455,
-      "learning_rate": 0.00039125740760693887,
-      "loss": 3.4606,
+      "epoch": 3.4932614555256065,
+      "grad_norm": 0.5759497880935669,
+      "learning_rate": 0.0003909033998920669,
+      "loss": 3.4627,
       "step": 32400
     },
     {
-      "epoch": 3.492627273705737,
-      "grad_norm": 0.5980201363563538,
-      "learning_rate": 0.00039093416657687746,
-      "loss": 3.4392,
+      "epoch": 3.498652291105121,
+      "grad_norm": 0.669968843460083,
+      "learning_rate": 0.00039057960064759845,
+      "loss": 3.443,
       "step": 32450
     },
     {
-      "epoch": 3.498008825745345,
-      "grad_norm": 0.6752046346664429,
-      "learning_rate": 0.00039061092554681606,
-      "loss": 3.4676,
+      "epoch": 3.5040431266846364,
+      "grad_norm": 0.6047779321670532,
+      "learning_rate": 0.0003902622773880194,
+      "loss": 3.4675,
       "step": 32500
     },
     {
-      "epoch": 3.503390377784953,
-      "grad_norm": 0.6271814703941345,
-      "learning_rate": 0.0003902876845167546,
-      "loss": 3.4844,
+      "epoch": 3.509433962264151,
+      "grad_norm": 0.6554362773895264,
+      "learning_rate": 0.000389938478143551,
+      "loss": 3.4455,
       "step": 32550
     },
     {
-      "epoch": 3.5087719298245617,
-      "grad_norm": 0.6477372050285339,
-      "learning_rate": 0.0003899644434866932,
-      "loss": 3.457,
+      "epoch": 3.5148247978436657,
+      "grad_norm": 0.5947167277336121,
+      "learning_rate": 0.00038961467889908255,
+      "loss": 3.4678,
       "step": 32600
     },
     {
-      "epoch": 3.5141534818641698,
-      "grad_norm": 0.613669753074646,
-      "learning_rate": 0.00038964120245663184,
-      "loss": 3.4533,
+      "epoch": 3.5202156334231804,
+      "grad_norm": 0.6193543672561646,
+      "learning_rate": 0.00038929087965461405,
+      "loss": 3.4529,
       "step": 32650
     },
     {
-      "epoch": 3.519535033903778,
-      "grad_norm": 0.5857597589492798,
-      "learning_rate": 0.0003893179614265704,
-      "loss": 3.4571,
+      "epoch": 3.525606469002695,
+      "grad_norm": 0.5975248217582703,
+      "learning_rate": 0.00038896708041014566,
+      "loss": 3.4469,
       "step": 32700
     },
     {
-      "epoch": 3.524916585943386,
-      "grad_norm": 0.6374267339706421,
-      "learning_rate": 0.000388994720396509,
-      "loss": 3.455,
+      "epoch": 3.5309973045822103,
+      "grad_norm": 0.6135221123695374,
+      "learning_rate": 0.0003886432811656772,
+      "loss": 3.4602,
       "step": 32750
     },
     {
-      "epoch": 3.530298137982994,
-      "grad_norm": 0.6730107665061951,
-      "learning_rate": 0.0003886714793664475,
-      "loss": 3.4661,
+      "epoch": 3.536388140161725,
+      "grad_norm": 0.5899997353553772,
+      "learning_rate": 0.0003883194819212088,
+      "loss": 3.4645,
       "step": 32800
     },
     {
-      "epoch": 3.5356796900226026,
-      "grad_norm": 0.6586828827857971,
-      "learning_rate": 0.0003883482383363861,
-      "loss": 3.4606,
+      "epoch": 3.5417789757412397,
+      "grad_norm": 0.5811609625816345,
+      "learning_rate": 0.00038799568267674036,
+      "loss": 3.451,
       "step": 32850
     },
     {
-      "epoch": 3.5410612420622107,
-      "grad_norm": 0.6340796947479248,
-      "learning_rate": 0.00038802499730632476,
-      "loss": 3.4729,
+      "epoch": 3.547169811320755,
+      "grad_norm": 0.558104932308197,
+      "learning_rate": 0.00038767188343227197,
+      "loss": 3.4529,
       "step": 32900
     },
     {
-      "epoch": 3.546442794101819,
-      "grad_norm": 0.6165903806686401,
-      "learning_rate": 0.0003877017562762633,
-      "loss": 3.4466,
+      "epoch": 3.5525606469002695,
+      "grad_norm": 0.5764511823654175,
+      "learning_rate": 0.0003873480841878035,
+      "loss": 3.474,
       "step": 32950
     },
     {
-      "epoch": 3.5518243461414274,
-      "grad_norm": 0.6035377383232117,
-      "learning_rate": 0.0003873785152462019,
-      "loss": 3.4654,
+      "epoch": 3.557951482479784,
+      "grad_norm": 0.6954588890075684,
+      "learning_rate": 0.0003870242849433351,
+      "loss": 3.4516,
       "step": 33000
     },
     {
-      "epoch": 3.5518243461414274,
-      "eval_accuracy": 0.3737836445025279,
-      "eval_loss": 3.466870069503784,
-      "eval_runtime": 180.8094,
-      "eval_samples_per_second": 99.613,
-      "eval_steps_per_second": 6.228,
+      "epoch": 3.557951482479784,
+      "eval_accuracy": 0.373256460690852,
+      "eval_loss": 3.4739084243774414,
+      "eval_runtime": 184.6218,
+      "eval_samples_per_second": 97.556,
+      "eval_steps_per_second": 6.099,
       "step": 33000
     },
     {
-      "epoch": 3.5572058981810355,
-      "grad_norm": 0.6327840089797974,
-      "learning_rate": 0.0003870552742161405,
-      "loss": 3.4539,
+      "epoch": 3.5633423180592994,
+      "grad_norm": 0.5901345014572144,
+      "learning_rate": 0.00038670048569886667,
+      "loss": 3.4542,
       "step": 33050
     },
     {
-      "epoch": 3.5625874502206436,
-      "grad_norm": 0.6125268936157227,
-      "learning_rate": 0.00038673203318607903,
-      "loss": 3.4392,
+      "epoch": 3.568733153638814,
+      "grad_norm": 0.5970635414123535,
+      "learning_rate": 0.0003863766864543982,
+      "loss": 3.4758,
       "step": 33100
     },
     {
-      "epoch": 3.5679690022602517,
-      "grad_norm": 0.6121278405189514,
-      "learning_rate": 0.0003864087921560176,
-      "loss": 3.452,
+      "epoch": 3.5741239892183287,
+      "grad_norm": 0.6120544672012329,
+      "learning_rate": 0.00038605288720992983,
+      "loss": 3.4519,
       "step": 33150
     },
     {
-      "epoch": 3.57335055429986,
-      "grad_norm": 0.5984029769897461,
-      "learning_rate": 0.0003860855511259563,
-      "loss": 3.4638,
+      "epoch": 3.579514824797844,
+      "grad_norm": 0.6588565707206726,
+      "learning_rate": 0.0003857290879654614,
+      "loss": 3.468,
       "step": 33200
     },
     {
-      "epoch": 3.5787321063394684,
-      "grad_norm": 0.6518922448158264,
-      "learning_rate": 0.0003857623100958948,
-      "loss": 3.4674,
+      "epoch": 3.5849056603773586,
+      "grad_norm": 0.5929945707321167,
+      "learning_rate": 0.000385405288720993,
+      "loss": 3.4663,
       "step": 33250
     },
     {
-      "epoch": 3.5841136583790765,
-      "grad_norm": 0.6262392401695251,
-      "learning_rate": 0.0003854390690658334,
-      "loss": 3.4391,
+      "epoch": 3.5902964959568733,
+      "grad_norm": 0.6324113607406616,
+      "learning_rate": 0.00038508148947652453,
+      "loss": 3.4891,
       "step": 33300
     },
     {
-      "epoch": 3.5894952104186846,
-      "grad_norm": 0.5916134715080261,
-      "learning_rate": 0.00038511582803577195,
-      "loss": 3.4595,
+      "epoch": 3.595687331536388,
+      "grad_norm": 0.6006209254264832,
+      "learning_rate": 0.00038475769023205614,
+      "loss": 3.4537,
       "step": 33350
     },
     {
-      "epoch": 3.594876762458293,
-      "grad_norm": 0.6136584877967834,
-      "learning_rate": 0.00038479258700571054,
-      "loss": 3.4431,
+      "epoch": 3.601078167115903,
+      "grad_norm": 0.5798439383506775,
+      "learning_rate": 0.00038443389098758763,
+      "loss": 3.4653,
       "step": 33400
     },
     {
-      "epoch": 3.6002583144979012,
-      "grad_norm": 0.6699656248092651,
-      "learning_rate": 0.00038446934597564914,
-      "loss": 3.4583,
+      "epoch": 3.606469002695418,
+      "grad_norm": 0.5905429124832153,
+      "learning_rate": 0.00038411009174311924,
+      "loss": 3.4769,
       "step": 33450
     },
     {
-      "epoch": 3.6056398665375093,
-      "grad_norm": 0.5867238640785217,
-      "learning_rate": 0.00038414610494558773,
-      "loss": 3.4409,
+      "epoch": 3.6118598382749325,
+      "grad_norm": 0.6151285767555237,
+      "learning_rate": 0.0003837862924986508,
+      "loss": 3.4692,
       "step": 33500
     },
     {
-      "epoch": 3.611021418577118,
-      "grad_norm": 0.6047042012214661,
-      "learning_rate": 0.00038382286391552633,
-      "loss": 3.4538,
+      "epoch": 3.617250673854447,
+      "grad_norm": 0.6235437393188477,
+      "learning_rate": 0.00038346249325418234,
+      "loss": 3.4569,
       "step": 33550
     },
     {
-      "epoch": 3.616402970616726,
-      "grad_norm": 0.6440869569778442,
-      "learning_rate": 0.0003834996228854649,
-      "loss": 3.4487,
+      "epoch": 3.6226415094339623,
+      "grad_norm": 0.5967689156532288,
+      "learning_rate": 0.00038313869400971395,
+      "loss": 3.4543,
       "step": 33600
     },
     {
-      "epoch": 3.621784522656334,
-      "grad_norm": 0.5783991813659668,
-      "learning_rate": 0.00038317638185540346,
-      "loss": 3.4664,
+      "epoch": 3.628032345013477,
+      "grad_norm": 0.6635581851005554,
+      "learning_rate": 0.0003828148947652455,
+      "loss": 3.4467,
       "step": 33650
     },
     {
-      "epoch": 3.627166074695942,
-      "grad_norm": 0.611685574054718,
-      "learning_rate": 0.00038285314082534206,
-      "loss": 3.4513,
+      "epoch": 3.6334231805929917,
+      "grad_norm": 0.6080300807952881,
+      "learning_rate": 0.0003824910955207771,
+      "loss": 3.4451,
       "step": 33700
     },
     {
-      "epoch": 3.6325476267355503,
-      "grad_norm": 0.6536784172058105,
-      "learning_rate": 0.0003825298997952806,
-      "loss": 3.4523,
+      "epoch": 3.638814016172507,
+      "grad_norm": 0.6412973999977112,
+      "learning_rate": 0.00038216729627630865,
+      "loss": 3.4645,
       "step": 33750
     },
     {
-      "epoch": 3.637929178775159,
-      "grad_norm": 0.557302713394165,
-      "learning_rate": 0.00038220665876521925,
-      "loss": 3.4496,
+      "epoch": 3.6442048517520216,
+      "grad_norm": 0.5894738435745239,
+      "learning_rate": 0.00038184349703184026,
+      "loss": 3.4445,
       "step": 33800
     },
     {
-      "epoch": 3.643310730814767,
-      "grad_norm": 0.6502556204795837,
-      "learning_rate": 0.00038188341773515784,
-      "loss": 3.4551,
+      "epoch": 3.6495956873315363,
+      "grad_norm": 0.6351161599159241,
+      "learning_rate": 0.0003815196977873718,
+      "loss": 3.4575,
       "step": 33850
     },
     {
-      "epoch": 3.648692282854375,
-      "grad_norm": 0.6656308770179749,
-      "learning_rate": 0.0003815601767050964,
-      "loss": 3.4679,
+      "epoch": 3.6549865229110514,
+      "grad_norm": 0.6070971488952637,
+      "learning_rate": 0.0003811958985429034,
+      "loss": 3.4581,
       "step": 33900
     },
     {
-      "epoch": 3.6540738348939836,
-      "grad_norm": 0.6328909993171692,
-      "learning_rate": 0.000381236935675035,
-      "loss": 3.4338,
+      "epoch": 3.660377358490566,
+      "grad_norm": 0.6513911485671997,
+      "learning_rate": 0.00038087209929843496,
+      "loss": 3.4564,
       "step": 33950
     },
     {
-      "epoch": 3.6594553869335917,
-      "grad_norm": 0.5742416977882385,
-      "learning_rate": 0.0003809136946449735,
-      "loss": 3.4522,
+      "epoch": 3.665768194070081,
+      "grad_norm": 0.6223207712173462,
+      "learning_rate": 0.00038054830005396646,
+      "loss": 3.4754,
       "step": 34000
     },
     {
-      "epoch": 3.6594553869335917,
-      "eval_accuracy": 0.3745204197347979,
-      "eval_loss": 3.4621152877807617,
-      "eval_runtime": 180.7271,
-      "eval_samples_per_second": 99.659,
-      "eval_steps_per_second": 6.23,
+      "epoch": 3.665768194070081,
+      "eval_accuracy": 0.37367835985423564,
+      "eval_loss": 3.467810869216919,
+      "eval_runtime": 184.9879,
+      "eval_samples_per_second": 97.363,
+      "eval_steps_per_second": 6.087,
       "step": 34000
     },
     {
-      "epoch": 3.6648369389732,
-      "grad_norm": 0.6359716653823853,
-      "learning_rate": 0.00038059045361491217,
-      "loss": 3.4588,
+      "epoch": 3.671159029649596,
+      "grad_norm": 0.6494529247283936,
+      "learning_rate": 0.00038022450080949806,
+      "loss": 3.4553,
       "step": 34050
     },
     {
-      "epoch": 3.670218491012808,
-      "grad_norm": 0.5943437814712524,
-      "learning_rate": 0.00038026721258485076,
-      "loss": 3.4557,
+      "epoch": 3.6765498652291106,
+      "grad_norm": 0.5826172828674316,
+      "learning_rate": 0.0003799007015650296,
+      "loss": 3.4634,
       "step": 34100
     },
     {
-      "epoch": 3.675600043052416,
-      "grad_norm": 0.6035296320915222,
-      "learning_rate": 0.0003799439715547893,
-      "loss": 3.4482,
+      "epoch": 3.6819407008086253,
+      "grad_norm": 0.6290008425712585,
+      "learning_rate": 0.0003795769023205612,
+      "loss": 3.47,
       "step": 34150
     },
     {
-      "epoch": 3.6809815950920246,
-      "grad_norm": 0.6305634379386902,
-      "learning_rate": 0.00037962719534532916,
-      "loss": 3.4752,
+      "epoch": 3.68733153638814,
+      "grad_norm": 0.5976738333702087,
+      "learning_rate": 0.00037925310307609277,
+      "loss": 3.4675,
       "step": 34200
     },
     {
-      "epoch": 3.6863631471316327,
-      "grad_norm": 0.5965917706489563,
-      "learning_rate": 0.0003793039543152677,
-      "loss": 3.4266,
+      "epoch": 3.6927223719676547,
+      "grad_norm": 0.6034207940101624,
+      "learning_rate": 0.0003789293038316244,
+      "loss": 3.4609,
       "step": 34250
     },
     {
-      "epoch": 3.691744699171241,
-      "grad_norm": 0.6232327818870544,
-      "learning_rate": 0.0003789807132852063,
-      "loss": 3.4366,
+      "epoch": 3.69811320754717,
+      "grad_norm": 0.6554673314094543,
+      "learning_rate": 0.0003786055045871559,
+      "loss": 3.4695,
       "step": 34300
     },
     {
-      "epoch": 3.6971262512108494,
-      "grad_norm": 0.6077519655227661,
-      "learning_rate": 0.0003786574722551449,
-      "loss": 3.4553,
+      "epoch": 3.7035040431266846,
+      "grad_norm": 0.6599771976470947,
+      "learning_rate": 0.0003782817053426875,
+      "loss": 3.4774,
       "step": 34350
     },
     {
-      "epoch": 3.7025078032504575,
-      "grad_norm": 0.6324222087860107,
-      "learning_rate": 0.00037833423122508343,
-      "loss": 3.4602,
+      "epoch": 3.7088948787061993,
+      "grad_norm": 0.576423704624176,
+      "learning_rate": 0.0003779579060982191,
+      "loss": 3.4506,
       "step": 34400
     },
     {
-      "epoch": 3.7078893552900656,
-      "grad_norm": 0.6138796210289001,
-      "learning_rate": 0.0003780109901950221,
-      "loss": 3.4462,
+      "epoch": 3.7142857142857144,
+      "grad_norm": 0.6085602641105652,
+      "learning_rate": 0.00037763410685375063,
+      "loss": 3.4371,
       "step": 34450
     },
     {
-      "epoch": 3.713270907329674,
-      "grad_norm": 0.6130177974700928,
-      "learning_rate": 0.0003776877491649607,
-      "loss": 3.4534,
+      "epoch": 3.719676549865229,
+      "grad_norm": 0.6494543552398682,
+      "learning_rate": 0.00037731030760928223,
+      "loss": 3.4561,
       "step": 34500
     },
     {
-      "epoch": 3.7186524593692822,
-      "grad_norm": 0.585455596446991,
-      "learning_rate": 0.0003773645081348992,
-      "loss": 3.4594,
+      "epoch": 3.725067385444744,
+      "grad_norm": 0.5890125036239624,
+      "learning_rate": 0.0003769865083648138,
+      "loss": 3.4745,
       "step": 34550
     },
     {
-      "epoch": 3.7240340114088903,
-      "grad_norm": 0.6451506614685059,
-      "learning_rate": 0.0003770412671048378,
-      "loss": 3.4623,
+      "epoch": 3.730458221024259,
+      "grad_norm": 0.6737868785858154,
+      "learning_rate": 0.0003766627091203454,
+      "loss": 3.4653,
       "step": 34600
     },
     {
-      "epoch": 3.7294155634484984,
-      "grad_norm": 0.6334204077720642,
-      "learning_rate": 0.00037671802607477635,
-      "loss": 3.4552,
+      "epoch": 3.7358490566037736,
+      "grad_norm": 0.6290357708930969,
+      "learning_rate": 0.00037633890987587694,
+      "loss": 3.4407,
       "step": 34650
     },
     {
-      "epoch": 3.7347971154881066,
-      "grad_norm": 0.6382574439048767,
-      "learning_rate": 0.000376394785044715,
-      "loss": 3.4504,
+      "epoch": 3.7412398921832883,
+      "grad_norm": 0.5991079807281494,
+      "learning_rate": 0.00037601511063140855,
+      "loss": 3.4806,
       "step": 34700
     },
     {
-      "epoch": 3.740178667527715,
-      "grad_norm": 0.6397031545639038,
-      "learning_rate": 0.0003760715440146536,
-      "loss": 3.4533,
+      "epoch": 3.7466307277628035,
+      "grad_norm": 0.6049849987030029,
+      "learning_rate": 0.00037569131138694004,
+      "loss": 3.4436,
       "step": 34750
     },
     {
-      "epoch": 3.745560219567323,
-      "grad_norm": 0.644877016544342,
-      "learning_rate": 0.00037574830298459214,
-      "loss": 3.4596,
+      "epoch": 3.752021563342318,
+      "grad_norm": 0.6415731906890869,
+      "learning_rate": 0.0003753675121424716,
+      "loss": 3.4751,
       "step": 34800
     },
     {
-      "epoch": 3.7509417716069313,
-      "grad_norm": 0.6290867924690247,
-      "learning_rate": 0.00037542506195453073,
-      "loss": 3.4531,
+      "epoch": 3.757412398921833,
+      "grad_norm": 0.602078378200531,
+      "learning_rate": 0.0003750437128980032,
+      "loss": 3.4876,
       "step": 34850
     },
     {
-      "epoch": 3.75632332364654,
-      "grad_norm": 0.6182779669761658,
-      "learning_rate": 0.0003751018209244693,
-      "loss": 3.4488,
+      "epoch": 3.7628032345013476,
+      "grad_norm": 0.5774201154708862,
+      "learning_rate": 0.00037471991365353475,
+      "loss": 3.4721,
       "step": 34900
     },
     {
-      "epoch": 3.761704875686148,
-      "grad_norm": 0.6078924536705017,
-      "learning_rate": 0.00037477857989440787,
-      "loss": 3.4599,
+      "epoch": 3.7681940700808623,
+      "grad_norm": 0.5578243136405945,
+      "learning_rate": 0.00037439611440906635,
+      "loss": 3.4525,
       "step": 34950
     },
     {
-      "epoch": 3.767086427725756,
-      "grad_norm": 0.6303339004516602,
-      "learning_rate": 0.0003744553388643465,
-      "loss": 3.4511,
+      "epoch": 3.7735849056603774,
+      "grad_norm": 0.5971388220787048,
+      "learning_rate": 0.0003740723151645979,
+      "loss": 3.4543,
       "step": 35000
     },
     {
-      "epoch": 3.767086427725756,
-      "eval_accuracy": 0.37501337788681216,
-      "eval_loss": 3.45632266998291,
-      "eval_runtime": 180.682,
-      "eval_samples_per_second": 99.683,
-      "eval_steps_per_second": 6.232,
+      "epoch": 3.7735849056603774,
+      "eval_accuracy": 0.37426954021683423,
+      "eval_loss": 3.4600491523742676,
+      "eval_runtime": 184.946,
+      "eval_samples_per_second": 97.385,
+      "eval_steps_per_second": 6.088,
       "step": 35000
     },
     {
-      "epoch": 3.772467979765364,
-      "grad_norm": 0.6259037852287292,
-      "learning_rate": 0.0003741320978342851,
-      "loss": 3.4522,
+      "epoch": 3.778975741239892,
+      "grad_norm": 0.6439304947853088,
+      "learning_rate": 0.0003737485159201295,
+      "loss": 3.4757,
       "step": 35050
     },
     {
-      "epoch": 3.7778495318049723,
-      "grad_norm": 0.6218352913856506,
-      "learning_rate": 0.00037380885680422365,
-      "loss": 3.452,
+      "epoch": 3.784366576819407,
+      "grad_norm": 0.6615363955497742,
+      "learning_rate": 0.00037342471667566106,
+      "loss": 3.4424,
       "step": 35100
     },
     {
-      "epoch": 3.783231083844581,
-      "grad_norm": 0.6394650936126709,
-      "learning_rate": 0.00037348561577416224,
-      "loss": 3.4425,
+      "epoch": 3.789757412398922,
+      "grad_norm": 0.5871990323066711,
+      "learning_rate": 0.00037310091743119266,
+      "loss": 3.4652,
       "step": 35150
     },
     {
-      "epoch": 3.788612635884189,
-      "grad_norm": 0.6433684825897217,
-      "learning_rate": 0.0003731623747441008,
-      "loss": 3.4432,
+      "epoch": 3.7951482479784366,
+      "grad_norm": 0.6038589477539062,
+      "learning_rate": 0.0003727771181867242,
+      "loss": 3.4471,
       "step": 35200
     },
     {
-      "epoch": 3.793994187923797,
-      "grad_norm": 0.5973502397537231,
-      "learning_rate": 0.0003728391337140394,
-      "loss": 3.4626,
+      "epoch": 3.8005390835579513,
+      "grad_norm": 0.6315929889678955,
+      "learning_rate": 0.00037245331894225576,
+      "loss": 3.4738,
       "step": 35250
     },
     {
-      "epoch": 3.7993757399634056,
-      "grad_norm": 0.6366912126541138,
-      "learning_rate": 0.00037251589268397803,
-      "loss": 3.4434,
+      "epoch": 3.8059299191374665,
+      "grad_norm": 0.6901034116744995,
+      "learning_rate": 0.00037212951969778737,
+      "loss": 3.4629,
       "step": 35300
     },
     {
-      "epoch": 3.8047572920030137,
-      "grad_norm": 0.6271637082099915,
-      "learning_rate": 0.00037219265165391657,
-      "loss": 3.4478,
+      "epoch": 3.811320754716981,
+      "grad_norm": 0.6126003265380859,
+      "learning_rate": 0.00037180572045331887,
+      "loss": 3.4642,
       "step": 35350
     },
     {
-      "epoch": 3.810138844042622,
-      "grad_norm": 0.6246579885482788,
-      "learning_rate": 0.00037186941062385516,
-      "loss": 3.4664,
+      "epoch": 3.816711590296496,
+      "grad_norm": 0.6850118637084961,
+      "learning_rate": 0.00037148192120885047,
+      "loss": 3.4828,
       "step": 35400
     },
     {
-      "epoch": 3.8155203960822304,
-      "grad_norm": 0.6329902410507202,
-      "learning_rate": 0.0003715461695937937,
-      "loss": 3.4509,
+      "epoch": 3.822102425876011,
+      "grad_norm": 0.6038691401481628,
+      "learning_rate": 0.000371158121964382,
+      "loss": 3.4385,
       "step": 35450
     },
     {
-      "epoch": 3.8209019481218385,
-      "grad_norm": 0.6158545613288879,
-      "learning_rate": 0.0003712229285637323,
-      "loss": 3.4538,
+      "epoch": 3.8274932614555257,
+      "grad_norm": 0.6459229588508606,
+      "learning_rate": 0.0003708343227199136,
+      "loss": 3.4753,
       "step": 35500
     },
     {
-      "epoch": 3.8262835001614466,
-      "grad_norm": 0.6100180149078369,
-      "learning_rate": 0.0003708996875336709,
-      "loss": 3.4539,
+      "epoch": 3.8328840970350404,
+      "grad_norm": 0.6884293556213379,
+      "learning_rate": 0.0003705105234754452,
+      "loss": 3.4546,
       "step": 35550
     },
     {
-      "epoch": 3.8316650522010547,
-      "grad_norm": 0.6152170896530151,
-      "learning_rate": 0.0003705764465036095,
-      "loss": 3.4498,
+      "epoch": 3.838274932614555,
+      "grad_norm": 0.598310649394989,
+      "learning_rate": 0.0003701867242309768,
+      "loss": 3.4446,
       "step": 35600
     },
     {
-      "epoch": 3.837046604240663,
-      "grad_norm": 0.6338374018669128,
-      "learning_rate": 0.0003702532054735481,
-      "loss": 3.4563,
+      "epoch": 3.8436657681940702,
+      "grad_norm": 0.6089202761650085,
+      "learning_rate": 0.00036986292498650833,
+      "loss": 3.4598,
       "step": 35650
     },
     {
-      "epoch": 3.8424281562802713,
-      "grad_norm": 0.6253648996353149,
-      "learning_rate": 0.0003699299644434867,
-      "loss": 3.452,
+      "epoch": 3.849056603773585,
+      "grad_norm": 0.6563847661018372,
+      "learning_rate": 0.0003695391257420399,
+      "loss": 3.4544,
       "step": 35700
     },
     {
-      "epoch": 3.8478097083198795,
-      "grad_norm": 0.6302428245544434,
-      "learning_rate": 0.0003696067234134252,
-      "loss": 3.456,
+      "epoch": 3.8544474393530996,
+      "grad_norm": 0.6375155448913574,
+      "learning_rate": 0.0003692153264975715,
+      "loss": 3.4363,
       "step": 35750
     },
     {
-      "epoch": 3.8531912603594876,
-      "grad_norm": 0.6266033053398132,
-      "learning_rate": 0.0003692834823833638,
-      "loss": 3.4385,
+      "epoch": 3.8598382749326143,
+      "grad_norm": 0.6118590831756592,
+      "learning_rate": 0.00036889152725310304,
+      "loss": 3.4561,
       "step": 35800
     },
     {
-      "epoch": 3.858572812399096,
-      "grad_norm": 0.613212525844574,
-      "learning_rate": 0.00036896024135330246,
-      "loss": 3.4627,
+      "epoch": 3.8652291105121295,
+      "grad_norm": 0.637126624584198,
+      "learning_rate": 0.00036856772800863464,
+      "loss": 3.4628,
       "step": 35850
     },
     {
-      "epoch": 3.863954364438704,
-      "grad_norm": 0.6388803720474243,
-      "learning_rate": 0.000368637000323241,
-      "loss": 3.4467,
+      "epoch": 3.870619946091644,
+      "grad_norm": 0.5772865414619446,
+      "learning_rate": 0.0003682439287641662,
+      "loss": 3.4519,
       "step": 35900
     },
     {
-      "epoch": 3.8693359164783123,
-      "grad_norm": 0.6624432802200317,
-      "learning_rate": 0.0003683137592931796,
-      "loss": 3.4508,
+      "epoch": 3.876010781671159,
+      "grad_norm": 0.5758655667304993,
+      "learning_rate": 0.0003679201295196978,
+      "loss": 3.4571,
       "step": 35950
     },
     {
-      "epoch": 3.8747174685179204,
-      "grad_norm": 0.6262862086296082,
-      "learning_rate": 0.00036799051826311814,
-      "loss": 3.4455,
+      "epoch": 3.881401617250674,
+      "grad_norm": 0.6438195109367371,
+      "learning_rate": 0.00036759633027522935,
+      "loss": 3.4385,
       "step": 36000
     },
     {
-      "epoch": 3.8747174685179204,
-      "eval_accuracy": 0.37557652580435463,
-      "eval_loss": 3.453572988510132,
-      "eval_runtime": 180.586,
-      "eval_samples_per_second": 99.736,
-      "eval_steps_per_second": 6.235,
+      "epoch": 3.881401617250674,
+      "eval_accuracy": 0.375193850333472,
+      "eval_loss": 3.4544591903686523,
+      "eval_runtime": 184.7169,
+      "eval_samples_per_second": 97.506,
+      "eval_steps_per_second": 6.096,
       "step": 36000
     },
     {
-      "epoch": 3.8800990205575285,
-      "grad_norm": 0.6518728137016296,
-      "learning_rate": 0.00036766727723305673,
-      "loss": 3.4398,
+      "epoch": 3.8867924528301887,
+      "grad_norm": 0.6010862588882446,
+      "learning_rate": 0.00036727253103076084,
+      "loss": 3.4421,
       "step": 36050
     },
     {
-      "epoch": 3.885480572597137,
-      "grad_norm": 0.6276541352272034,
-      "learning_rate": 0.0003673440362029953,
-      "loss": 3.4572,
+      "epoch": 3.8921832884097034,
+      "grad_norm": 0.5899602770805359,
+      "learning_rate": 0.00036694873178629245,
+      "loss": 3.4652,
       "step": 36100
     },
     {
-      "epoch": 3.890862124636745,
-      "grad_norm": 0.6711607575416565,
-      "learning_rate": 0.0003670207951729339,
-      "loss": 3.4699,
+      "epoch": 3.8975741239892185,
+      "grad_norm": 0.6755630373954773,
+      "learning_rate": 0.000366624932541824,
+      "loss": 3.4556,
       "step": 36150
     },
     {
-      "epoch": 3.8962436766763533,
-      "grad_norm": 0.610508382320404,
-      "learning_rate": 0.0003667040189634737,
-      "loss": 3.4665,
+      "epoch": 3.9029649595687332,
+      "grad_norm": 0.5775014758110046,
+      "learning_rate": 0.0003663011332973556,
+      "loss": 3.4361,
       "step": 36200
     },
     {
-      "epoch": 3.901625228715962,
-      "grad_norm": 0.6061813235282898,
-      "learning_rate": 0.0003663807779334123,
-      "loss": 3.4368,
+      "epoch": 3.908355795148248,
+      "grad_norm": 0.6255040168762207,
+      "learning_rate": 0.00036597733405288715,
+      "loss": 3.4413,
       "step": 36250
     },
     {
-      "epoch": 3.90700678075557,
-      "grad_norm": 0.6153860092163086,
-      "learning_rate": 0.0003660575369033509,
-      "loss": 3.4408,
+      "epoch": 3.913746630727763,
+      "grad_norm": 0.6343650817871094,
+      "learning_rate": 0.00036565353480841876,
+      "loss": 3.4543,
       "step": 36300
     },
     {
-      "epoch": 3.912388332795178,
-      "grad_norm": 0.6401193141937256,
-      "learning_rate": 0.0003657342958732895,
-      "loss": 3.4316,
+      "epoch": 3.9191374663072778,
+      "grad_norm": 0.6278274059295654,
+      "learning_rate": 0.0003653297355639503,
+      "loss": 3.445,
       "step": 36350
     },
     {
-      "epoch": 3.9177698848347866,
-      "grad_norm": 0.6334563493728638,
-      "learning_rate": 0.00036541105484322805,
-      "loss": 3.4479,
+      "epoch": 3.9245283018867925,
+      "grad_norm": 0.6233466267585754,
+      "learning_rate": 0.0003650059363194819,
+      "loss": 3.4475,
       "step": 36400
     },
     {
-      "epoch": 3.9231514368743947,
-      "grad_norm": 0.6796964406967163,
-      "learning_rate": 0.00036508781381316665,
-      "loss": 3.4225,
+      "epoch": 3.929919137466307,
+      "grad_norm": 0.6265336275100708,
+      "learning_rate": 0.00036468213707501347,
+      "loss": 3.4557,
       "step": 36450
     },
     {
-      "epoch": 3.928532988914003,
-      "grad_norm": 0.5866906046867371,
-      "learning_rate": 0.0003647645727831053,
-      "loss": 3.4297,
+      "epoch": 3.935309973045822,
+      "grad_norm": 0.6003140211105347,
+      "learning_rate": 0.0003643648138154344,
+      "loss": 3.4605,
       "step": 36500
     },
     {
-      "epoch": 3.933914540953611,
-      "grad_norm": 0.6415225267410278,
-      "learning_rate": 0.00036444133175304384,
-      "loss": 3.4563,
+      "epoch": 3.940700808625337,
+      "grad_norm": 0.6145981550216675,
+      "learning_rate": 0.00036404101457096597,
+      "loss": 3.449,
       "step": 36550
     },
     {
-      "epoch": 3.939296092993219,
-      "grad_norm": 0.6429531574249268,
-      "learning_rate": 0.00036411809072298243,
-      "loss": 3.4365,
+      "epoch": 3.9460916442048517,
+      "grad_norm": 0.6301774978637695,
+      "learning_rate": 0.00036371721532649757,
+      "loss": 3.4603,
       "step": 36600
     },
     {
-      "epoch": 3.9446776450328276,
-      "grad_norm": 0.590459942817688,
-      "learning_rate": 0.00036379484969292097,
-      "loss": 3.4546,
+      "epoch": 3.9514824797843664,
+      "grad_norm": 0.5825753211975098,
+      "learning_rate": 0.0003633934160820291,
+      "loss": 3.4527,
       "step": 36650
     },
     {
-      "epoch": 3.9500591970724357,
-      "grad_norm": 0.5942658185958862,
-      "learning_rate": 0.00036347160866285956,
-      "loss": 3.438,
+      "epoch": 3.9568733153638815,
+      "grad_norm": 0.6060313582420349,
+      "learning_rate": 0.00036306961683756073,
+      "loss": 3.453,
       "step": 36700
     },
     {
-      "epoch": 3.955440749112044,
-      "grad_norm": 0.6288406848907471,
-      "learning_rate": 0.0003631483676327981,
-      "loss": 3.4517,
+      "epoch": 3.9622641509433962,
+      "grad_norm": 0.6096614003181458,
+      "learning_rate": 0.0003627458175930922,
+      "loss": 3.4412,
       "step": 36750
     },
     {
-      "epoch": 3.9608223011516523,
-      "grad_norm": 0.6701431274414062,
-      "learning_rate": 0.00036282512660273675,
-      "loss": 3.4485,
+      "epoch": 3.967654986522911,
+      "grad_norm": 0.6492642760276794,
+      "learning_rate": 0.0003624220183486238,
+      "loss": 3.4409,
       "step": 36800
     },
     {
-      "epoch": 3.9662038531912605,
-      "grad_norm": 0.7019712328910828,
-      "learning_rate": 0.00036250188557267535,
-      "loss": 3.4408,
+      "epoch": 3.973045822102426,
+      "grad_norm": 0.6882068514823914,
+      "learning_rate": 0.0003620982191041554,
+      "loss": 3.4385,
       "step": 36850
     },
     {
-      "epoch": 3.9715854052308686,
-      "grad_norm": 0.6082743406295776,
-      "learning_rate": 0.0003621786445426139,
-      "loss": 3.4446,
+      "epoch": 3.9784366576819408,
+      "grad_norm": 0.6335549354553223,
+      "learning_rate": 0.00036177441985968693,
+      "loss": 3.4543,
       "step": 36900
     },
     {
-      "epoch": 3.9769669572704767,
-      "grad_norm": 0.6317737102508545,
-      "learning_rate": 0.0003618554035125525,
-      "loss": 3.4349,
+      "epoch": 3.9838274932614555,
+      "grad_norm": 0.6579206585884094,
+      "learning_rate": 0.00036145062061521854,
+      "loss": 3.4656,
       "step": 36950
     },
     {
-      "epoch": 3.9823485093100848,
-      "grad_norm": 0.619764506816864,
-      "learning_rate": 0.0003615321624824911,
-      "loss": 3.4402,
+      "epoch": 3.9892183288409706,
+      "grad_norm": 0.6292389631271362,
+      "learning_rate": 0.0003611268213707501,
+      "loss": 3.4609,
       "step": 37000
     },
     {
-      "epoch": 3.9823485093100848,
-      "eval_accuracy": 0.3760329765860633,
-      "eval_loss": 3.4432144165039062,
-      "eval_runtime": 180.6742,
-      "eval_samples_per_second": 99.688,
-      "eval_steps_per_second": 6.232,
+      "epoch": 3.9892183288409706,
+      "eval_accuracy": 0.3756760518495927,
+      "eval_loss": 3.4502644538879395,
+      "eval_runtime": 185.0208,
+      "eval_samples_per_second": 97.346,
+      "eval_steps_per_second": 6.086,
       "step": 37000
     },
     {
-      "epoch": 3.9877300613496933,
-      "grad_norm": 0.649409830570221,
-      "learning_rate": 0.0003612089214524296,
-      "loss": 3.4344,
+      "epoch": 3.9946091644204853,
+      "grad_norm": 0.6191860437393188,
+      "learning_rate": 0.0003608030221262817,
+      "loss": 3.4559,
       "step": 37050
     },
     {
-      "epoch": 3.9931116133893014,
-      "grad_norm": 0.6841443777084351,
-      "learning_rate": 0.00036088568042236827,
-      "loss": 3.4393,
+      "epoch": 4.0,
+      "grad_norm": 1.2047762870788574,
+      "learning_rate": 0.00036047922288181324,
+      "loss": 3.4466,
       "step": 37100
     },
     {
-      "epoch": 3.9984931654289095,
-      "grad_norm": 0.6146060824394226,
-      "learning_rate": 0.00036056243939230686,
-      "loss": 3.4337,
+      "epoch": 4.005390835579515,
+      "grad_norm": 0.6499625444412231,
+      "learning_rate": 0.00036015542363734485,
+      "loss": 3.3602,
       "step": 37150
     },
     {
-      "epoch": 4.003874717468518,
-      "grad_norm": 0.6463866829872131,
-      "learning_rate": 0.0003602391983622454,
-      "loss": 3.3701,
+      "epoch": 4.010781671159029,
+      "grad_norm": 0.6357666850090027,
+      "learning_rate": 0.0003598316243928764,
+      "loss": 3.3603,
       "step": 37200
     },
     {
-      "epoch": 4.009256269508126,
-      "grad_norm": 0.6479421257972717,
-      "learning_rate": 0.000359915957332184,
-      "loss": 3.3524,
+      "epoch": 4.0161725067385445,
+      "grad_norm": 0.6046599745750427,
+      "learning_rate": 0.00035950782514840795,
+      "loss": 3.3552,
       "step": 37250
     },
     {
-      "epoch": 4.014637821547734,
-      "grad_norm": 0.7297543883323669,
-      "learning_rate": 0.00035959271630212254,
-      "loss": 3.3483,
+      "epoch": 4.02156334231806,
+      "grad_norm": 0.7040229439735413,
+      "learning_rate": 0.00035918402590393955,
+      "loss": 3.3379,
       "step": 37300
     },
     {
-      "epoch": 4.020019373587343,
-      "grad_norm": 0.6528753638267517,
-      "learning_rate": 0.00035926947527206113,
-      "loss": 3.3525,
+      "epoch": 4.026954177897574,
+      "grad_norm": 0.6770185828208923,
+      "learning_rate": 0.00035886022665947105,
+      "loss": 3.3644,
       "step": 37350
     },
     {
-      "epoch": 4.0254009256269505,
-      "grad_norm": 0.6073748469352722,
-      "learning_rate": 0.0003589462342419998,
-      "loss": 3.3627,
+      "epoch": 4.032345013477089,
+      "grad_norm": 0.6390303373336792,
+      "learning_rate": 0.0003585364274150027,
+      "loss": 3.3712,
       "step": 37400
     },
     {
-      "epoch": 4.030782477666559,
-      "grad_norm": 0.6415776610374451,
-      "learning_rate": 0.0003586229932119383,
-      "loss": 3.3607,
+      "epoch": 4.037735849056604,
+      "grad_norm": 0.6191239953041077,
+      "learning_rate": 0.0003582126281705342,
+      "loss": 3.3588,
       "step": 37450
     },
     {
-      "epoch": 4.036164029706168,
-      "grad_norm": 0.629884660243988,
-      "learning_rate": 0.0003582997521818769,
-      "loss": 3.3572,
+      "epoch": 4.0431266846361185,
+      "grad_norm": 0.6024324893951416,
+      "learning_rate": 0.0003578888289260658,
+      "loss": 3.3723,
       "step": 37500
     },
     {
-      "epoch": 4.041545581745775,
-      "grad_norm": 0.6344863772392273,
-      "learning_rate": 0.0003579765111518155,
-      "loss": 3.3406,
+      "epoch": 4.048517520215634,
+      "grad_norm": 0.6768052577972412,
+      "learning_rate": 0.00035756502968159736,
+      "loss": 3.3643,
       "step": 37550
     },
     {
-      "epoch": 4.046927133785384,
-      "grad_norm": 0.5791500806808472,
-      "learning_rate": 0.00035765327012175405,
-      "loss": 3.3334,
+      "epoch": 4.053908355795148,
+      "grad_norm": 0.646030068397522,
+      "learning_rate": 0.00035724123043712896,
+      "loss": 3.3596,
       "step": 37600
     },
     {
-      "epoch": 4.0523086858249915,
-      "grad_norm": 0.6291230320930481,
-      "learning_rate": 0.0003573300290916927,
-      "loss": 3.364,
+      "epoch": 4.059299191374663,
+      "grad_norm": 0.7075605392456055,
+      "learning_rate": 0.0003569174311926605,
+      "loss": 3.3739,
       "step": 37650
     },
     {
-      "epoch": 4.0576902378646,
-      "grad_norm": 0.6606059670448303,
-      "learning_rate": 0.0003570067880616313,
-      "loss": 3.3702,
+      "epoch": 4.064690026954178,
+      "grad_norm": 0.6510331034660339,
+      "learning_rate": 0.00035659363194819206,
+      "loss": 3.3767,
       "step": 37700
     },
     {
-      "epoch": 4.063071789904209,
-      "grad_norm": 0.66637122631073,
-      "learning_rate": 0.00035668354703156984,
-      "loss": 3.375,
+      "epoch": 4.070080862533692,
+      "grad_norm": 0.633736252784729,
+      "learning_rate": 0.00035626983270372367,
+      "loss": 3.355,
       "step": 37750
     },
     {
-      "epoch": 4.068453341943816,
-      "grad_norm": 0.6079049110412598,
-      "learning_rate": 0.00035636030600150843,
-      "loss": 3.3491,
+      "epoch": 4.0754716981132075,
+      "grad_norm": 0.6637200713157654,
+      "learning_rate": 0.0003559460334592552,
+      "loss": 3.3565,
       "step": 37800
     },
     {
-      "epoch": 4.073834893983425,
-      "grad_norm": 0.6581602692604065,
-      "learning_rate": 0.00035603706497144697,
-      "loss": 3.3601,
+      "epoch": 4.080862533692723,
+      "grad_norm": 0.6610650420188904,
+      "learning_rate": 0.0003556222342147868,
+      "loss": 3.3665,
       "step": 37850
     },
     {
-      "epoch": 4.079216446023033,
-      "grad_norm": 0.5903530716896057,
-      "learning_rate": 0.00035571382394138557,
-      "loss": 3.3685,
+      "epoch": 4.086253369272237,
+      "grad_norm": 0.5748628377914429,
+      "learning_rate": 0.0003552984349703184,
+      "loss": 3.3764,
       "step": 37900
     },
     {
-      "epoch": 4.084597998062641,
-      "grad_norm": 0.6376979947090149,
-      "learning_rate": 0.0003553905829113242,
-      "loss": 3.3583,
+      "epoch": 4.091644204851752,
+      "grad_norm": 0.6597402691841125,
+      "learning_rate": 0.00035497463572585,
+      "loss": 3.3859,
       "step": 37950
     },
     {
-      "epoch": 4.08997955010225,
-      "grad_norm": 0.6378331184387207,
-      "learning_rate": 0.00035506734188126275,
-      "loss": 3.3602,
+      "epoch": 4.097035040431267,
+      "grad_norm": 0.6575984954833984,
+      "learning_rate": 0.00035465083648138153,
+      "loss": 3.3725,
       "step": 38000
     },
     {
-      "epoch": 4.08997955010225,
-      "eval_accuracy": 0.37666750945089983,
-      "eval_loss": 3.4465579986572266,
-      "eval_runtime": 180.5241,
-      "eval_samples_per_second": 99.771,
-      "eval_steps_per_second": 6.237,
+      "epoch": 4.097035040431267,
+      "eval_accuracy": 0.37657613237224785,
+      "eval_loss": 3.45074462890625,
+      "eval_runtime": 185.1239,
+      "eval_samples_per_second": 97.292,
+      "eval_steps_per_second": 6.082,
       "step": 38000
     },
     {
-      "epoch": 4.095361102141858,
-      "grad_norm": 0.6897552013397217,
-      "learning_rate": 0.00035474410085120135,
-      "loss": 3.3881,
+      "epoch": 4.1024258760107815,
+      "grad_norm": 0.6551222205162048,
+      "learning_rate": 0.00035432703723691314,
+      "loss": 3.3677,
       "step": 38050
     },
     {
-      "epoch": 4.100742654181466,
-      "grad_norm": 0.624983549118042,
-      "learning_rate": 0.00035442085982113994,
-      "loss": 3.384,
+      "epoch": 4.107816711590297,
+      "grad_norm": 0.6057955026626587,
+      "learning_rate": 0.00035400323799244463,
+      "loss": 3.3761,
       "step": 38100
     },
     {
-      "epoch": 4.106124206221074,
-      "grad_norm": 0.6599211692810059,
-      "learning_rate": 0.0003540976187910785,
-      "loss": 3.3793,
+      "epoch": 4.113207547169812,
+      "grad_norm": 0.5905152559280396,
+      "learning_rate": 0.0003536794387479762,
+      "loss": 3.3678,
       "step": 38150
     },
     {
-      "epoch": 4.111505758260682,
-      "grad_norm": Infinity,
-      "learning_rate": 0.0003537808425816183,
-      "loss": 3.3838,
+      "epoch": 4.118598382749326,
+      "grad_norm": 0.621264636516571,
+      "learning_rate": 0.0003533556395035078,
+      "loss": 3.3929,
       "step": 38200
     },
     {
-      "epoch": 4.1168873103002905,
-      "grad_norm": 0.6250761151313782,
-      "learning_rate": 0.0003534576015515569,
-      "loss": 3.372,
+      "epoch": 4.123989218328841,
+      "grad_norm": 0.6266987919807434,
+      "learning_rate": 0.00035303184025903934,
+      "loss": 3.3833,
       "step": 38250
     },
     {
-      "epoch": 4.122268862339899,
-      "grad_norm": 0.6939108967781067,
-      "learning_rate": 0.00035313436052149553,
-      "loss": 3.3702,
+      "epoch": 4.129380053908355,
+      "grad_norm": 0.6679258346557617,
+      "learning_rate": 0.00035270804101457094,
+      "loss": 3.386,
       "step": 38300
     },
     {
-      "epoch": 4.127650414379507,
-      "grad_norm": 0.6383840441703796,
-      "learning_rate": 0.0003528111194914341,
-      "loss": 3.3634,
+      "epoch": 4.1347708894878705,
+      "grad_norm": 0.6121634244918823,
+      "learning_rate": 0.0003523842417701025,
+      "loss": 3.3758,
       "step": 38350
     },
     {
-      "epoch": 4.133031966419115,
-      "grad_norm": 0.631736159324646,
-      "learning_rate": 0.00035248787846137267,
-      "loss": 3.3815,
+      "epoch": 4.140161725067386,
+      "grad_norm": 0.6435196995735168,
+      "learning_rate": 0.0003520604425256341,
+      "loss": 3.3692,
       "step": 38400
     },
     {
-      "epoch": 4.138413518458724,
-      "grad_norm": 0.9963732957839966,
-      "learning_rate": 0.00035216463743131126,
-      "loss": 3.3576,
+      "epoch": 4.1455525606469,
+      "grad_norm": 0.6009247303009033,
+      "learning_rate": 0.00035173664328116565,
+      "loss": 3.3782,
       "step": 38450
     },
     {
-      "epoch": 4.1437950704983315,
-      "grad_norm": 0.6733307838439941,
-      "learning_rate": 0.0003518413964012498,
-      "loss": 3.3926,
+      "epoch": 4.150943396226415,
+      "grad_norm": 0.630068302154541,
+      "learning_rate": 0.00035141284403669725,
+      "loss": 3.3858,
       "step": 38500
     },
     {
-      "epoch": 4.14917662253794,
-      "grad_norm": 0.5962156653404236,
-      "learning_rate": 0.0003515181553711884,
-      "loss": 3.3779,
+      "epoch": 4.15633423180593,
+      "grad_norm": Infinity,
+      "learning_rate": 0.00035109552077711815,
+      "loss": 3.3752,
       "step": 38550
     },
     {
-      "epoch": 4.154558174577549,
-      "grad_norm": 0.6685757637023926,
-      "learning_rate": 0.00035119491434112705,
-      "loss": 3.3795,
+      "epoch": 4.1617250673854445,
+      "grad_norm": 0.6584694981575012,
+      "learning_rate": 0.00035077172153264976,
+      "loss": 3.3719,
       "step": 38600
     },
     {
-      "epoch": 4.159939726617156,
-      "grad_norm": 0.6577488780021667,
-      "learning_rate": 0.0003508716733110656,
-      "loss": 3.3678,
+      "epoch": 4.16711590296496,
+      "grad_norm": 0.680530309677124,
+      "learning_rate": 0.0003504479222881813,
+      "loss": 3.3995,
       "step": 38650
     },
     {
-      "epoch": 4.165321278656765,
-      "grad_norm": 0.7001060247421265,
-      "learning_rate": 0.0003505484322810042,
-      "loss": 3.3803,
+      "epoch": 4.172506738544475,
+      "grad_norm": 0.662716269493103,
+      "learning_rate": 0.0003501241230437129,
+      "loss": 3.3773,
       "step": 38700
     },
     {
-      "epoch": 4.1707028306963725,
-      "grad_norm": 0.6563752889633179,
-      "learning_rate": 0.0003502251912509427,
-      "loss": 3.3882,
+      "epoch": 4.177897574123989,
+      "grad_norm": 0.6013767123222351,
+      "learning_rate": 0.0003498003237992444,
+      "loss": 3.3717,
       "step": 38750
     },
     {
-      "epoch": 4.176084382735981,
-      "grad_norm": 0.6249417066574097,
-      "learning_rate": 0.0003499019502208813,
-      "loss": 3.3791,
+      "epoch": 4.183288409703504,
+      "grad_norm": 0.6527413725852966,
+      "learning_rate": 0.000349476524554776,
+      "loss": 3.392,
       "step": 38800
     },
     {
-      "epoch": 4.18146593477559,
-      "grad_norm": 0.6289679408073425,
-      "learning_rate": 0.0003495787091908199,
-      "loss": 3.3752,
+      "epoch": 4.188679245283019,
+      "grad_norm": 0.6448991298675537,
+      "learning_rate": 0.00034915272531030756,
+      "loss": 3.395,
       "step": 38850
     },
     {
-      "epoch": 4.186847486815197,
-      "grad_norm": 0.6969509720802307,
-      "learning_rate": 0.0003492554681607585,
-      "loss": 3.3764,
+      "epoch": 4.1940700808625335,
+      "grad_norm": 0.6825259923934937,
+      "learning_rate": 0.0003488289260658391,
+      "loss": 3.3939,
       "step": 38900
     },
     {
-      "epoch": 4.192229038854806,
-      "grad_norm": 0.663133442401886,
-      "learning_rate": 0.0003489322271306971,
-      "loss": 3.3928,
+      "epoch": 4.199460916442049,
+      "grad_norm": 0.6313694715499878,
+      "learning_rate": 0.0003485051268213707,
+      "loss": 3.387,
       "step": 38950
     },
     {
-      "epoch": 4.197610590894414,
-      "grad_norm": 0.7347210645675659,
-      "learning_rate": 0.0003486089861006357,
-      "loss": 3.3807,
+      "epoch": 4.204851752021563,
+      "grad_norm": 0.6302531361579895,
+      "learning_rate": 0.00034818132757690227,
+      "loss": 3.3828,
       "step": 39000
     },
     {
-      "epoch": 4.197610590894414,
-      "eval_accuracy": 0.37702878030288295,
-      "eval_loss": 3.4417600631713867,
-      "eval_runtime": 180.8436,
-      "eval_samples_per_second": 99.594,
-      "eval_steps_per_second": 6.226,
+      "epoch": 4.204851752021563,
+      "eval_accuracy": 0.37683353106348033,
+      "eval_loss": 3.4443628787994385,
+      "eval_runtime": 186.1059,
+      "eval_samples_per_second": 96.778,
+      "eval_steps_per_second": 6.05,
       "step": 39000
     },
     {
-      "epoch": 4.202992142934022,
-      "grad_norm": 0.72121262550354,
-      "learning_rate": 0.00034828574507057424,
-      "loss": 3.3713,
+      "epoch": 4.210242587601078,
+      "grad_norm": 0.6770233511924744,
+      "learning_rate": 0.0003478575283324339,
+      "loss": 3.3811,
       "step": 39050
     },
     {
-      "epoch": 4.208373694973631,
-      "grad_norm": 0.6440616250038147,
-      "learning_rate": 0.00034796250404051283,
-      "loss": 3.3781,
+      "epoch": 4.215633423180593,
+      "grad_norm": 0.6239271759986877,
+      "learning_rate": 0.0003475337290879654,
+      "loss": 3.3813,
       "step": 39100
     },
     {
-      "epoch": 4.213755247013238,
-      "grad_norm": 0.6587533950805664,
-      "learning_rate": 0.00034763926301045137,
-      "loss": 3.3838,
+      "epoch": 4.2210242587601075,
+      "grad_norm": 0.6964449286460876,
+      "learning_rate": 0.00034720992984349703,
+      "loss": 3.379,
       "step": 39150
     },
     {
-      "epoch": 4.219136799052847,
-      "grad_norm": 0.6770883798599243,
-      "learning_rate": 0.00034731602198039,
-      "loss": 3.3788,
+      "epoch": 4.226415094339623,
+      "grad_norm": 0.6273817420005798,
+      "learning_rate": 0.0003468861305990286,
+      "loss": 3.3769,
       "step": 39200
     },
     {
-      "epoch": 4.224518351092455,
-      "grad_norm": 0.6280528903007507,
-      "learning_rate": 0.0003469927809503286,
-      "loss": 3.3621,
+      "epoch": 4.231805929919138,
+      "grad_norm": 0.660798966884613,
+      "learning_rate": 0.0003465623313545602,
+      "loss": 3.3748,
       "step": 39250
     },
     {
-      "epoch": 4.229899903132063,
-      "grad_norm": 0.7106229662895203,
-      "learning_rate": 0.00034666953992026716,
-      "loss": 3.3894,
+      "epoch": 4.237196765498652,
+      "grad_norm": 0.6665384769439697,
+      "learning_rate": 0.00034623853211009173,
+      "loss": 3.3878,
       "step": 39300
     },
     {
-      "epoch": 4.2352814551716715,
-      "grad_norm": 0.6384456157684326,
-      "learning_rate": 0.00034634629889020575,
-      "loss": 3.3686,
+      "epoch": 4.242587601078167,
+      "grad_norm": 0.6520787477493286,
+      "learning_rate": 0.00034591473286562323,
+      "loss": 3.3853,
       "step": 39350
     },
     {
-      "epoch": 4.24066300721128,
-      "grad_norm": 0.6792171001434326,
-      "learning_rate": 0.00034602305786014435,
-      "loss": 3.3783,
+      "epoch": 4.247978436657682,
+      "grad_norm": 0.6408469080924988,
+      "learning_rate": 0.0003455909336211549,
+      "loss": 3.3927,
       "step": 39400
     },
     {
-      "epoch": 4.246044559250888,
-      "grad_norm": 0.6978849768638611,
-      "learning_rate": 0.00034569981683008294,
-      "loss": 3.3816,
+      "epoch": 4.2533692722371965,
+      "grad_norm": 0.638053297996521,
+      "learning_rate": 0.0003452671343766864,
+      "loss": 3.3805,
       "step": 39450
     },
     {
-      "epoch": 4.251426111290496,
-      "grad_norm": 0.6810808181762695,
-      "learning_rate": 0.00034537657580002154,
-      "loss": 3.3828,
+      "epoch": 4.258760107816712,
+      "grad_norm": 0.654602587223053,
+      "learning_rate": 0.000344943335132218,
+      "loss": 3.3743,
       "step": 39500
     },
     {
-      "epoch": 4.256807663330104,
-      "grad_norm": 0.6624694466590881,
-      "learning_rate": 0.00034505333476996013,
-      "loss": 3.3848,
+      "epoch": 4.264150943396227,
+      "grad_norm": 0.7286903858184814,
+      "learning_rate": 0.00034461953588774954,
+      "loss": 3.3771,
       "step": 39550
     },
     {
-      "epoch": 4.2621892153697125,
-      "grad_norm": 0.6291442513465881,
-      "learning_rate": 0.00034473009373989867,
-      "loss": 3.3908,
+      "epoch": 4.269541778975741,
+      "grad_norm": 0.672970712184906,
+      "learning_rate": 0.00034429573664328115,
+      "loss": 3.3774,
       "step": 39600
     },
     {
-      "epoch": 4.267570767409321,
-      "grad_norm": 0.6254659295082092,
-      "learning_rate": 0.00034440685270983727,
-      "loss": 3.3732,
+      "epoch": 4.274932614555256,
+      "grad_norm": 0.6813864707946777,
+      "learning_rate": 0.0003439719373988127,
+      "loss": 3.4044,
       "step": 39650
     },
     {
-      "epoch": 4.272952319448929,
-      "grad_norm": 0.6378159523010254,
-      "learning_rate": 0.0003440836116797758,
-      "loss": 3.3713,
+      "epoch": 4.280323450134771,
+      "grad_norm": 0.6745753288269043,
+      "learning_rate": 0.00034364813815434425,
+      "loss": 3.3876,
       "step": 39700
     },
     {
-      "epoch": 4.278333871488537,
-      "grad_norm": 0.6168226003646851,
-      "learning_rate": 0.00034376037064971445,
-      "loss": 3.3812,
+      "epoch": 4.285714285714286,
+      "grad_norm": 0.6817641854286194,
+      "learning_rate": 0.00034332433890987585,
+      "loss": 3.3838,
       "step": 39750
     },
     {
-      "epoch": 4.283715423528146,
-      "grad_norm": 0.6415433287620544,
-      "learning_rate": 0.00034343712961965305,
-      "loss": 3.3846,
+      "epoch": 4.291105121293801,
+      "grad_norm": 0.6120652556419373,
+      "learning_rate": 0.0003430005396654074,
+      "loss": 3.3974,
       "step": 39800
     },
     {
-      "epoch": 4.2890969755677535,
-      "grad_norm": 0.6528810262680054,
-      "learning_rate": 0.0003431138885895916,
-      "loss": 3.377,
+      "epoch": 4.296495956873315,
+      "grad_norm": 0.6508928537368774,
+      "learning_rate": 0.000342676740420939,
+      "loss": 3.4013,
       "step": 39850
     },
     {
-      "epoch": 4.294478527607362,
-      "grad_norm": 0.6433749794960022,
-      "learning_rate": 0.0003427906475595302,
-      "loss": 3.3652,
+      "epoch": 4.30188679245283,
+      "grad_norm": 0.6188628077507019,
+      "learning_rate": 0.00034235294117647056,
+      "loss": 3.392,
       "step": 39900
     },
     {
-      "epoch": 4.299860079646971,
-      "grad_norm": 0.6395750641822815,
-      "learning_rate": 0.0003424674065294688,
-      "loss": 3.381,
+      "epoch": 4.307277628032345,
+      "grad_norm": 0.6334621906280518,
+      "learning_rate": 0.00034202914193200216,
+      "loss": 3.3925,
       "step": 39950
     },
     {
-      "epoch": 4.305241631686578,
-      "grad_norm": 0.6627628207206726,
-      "learning_rate": 0.0003421441654994073,
-      "loss": 3.3776,
+      "epoch": 4.3126684636118595,
+      "grad_norm": 0.6367536187171936,
+      "learning_rate": 0.0003417053426875337,
+      "loss": 3.3856,
       "step": 40000
     },
     {
-      "epoch": 4.305241631686578,
-      "eval_accuracy": 0.37774751915577576,
-      "eval_loss": 3.4375598430633545,
-      "eval_runtime": 180.6158,
-      "eval_samples_per_second": 99.72,
-      "eval_steps_per_second": 6.234,
+      "epoch": 4.3126684636118595,
+      "eval_accuracy": 0.37727640023421216,
+      "eval_loss": 3.440788507461548,
+      "eval_runtime": 187.6321,
+      "eval_samples_per_second": 95.991,
+      "eval_steps_per_second": 6.001,
       "step": 40000
     },
     {
-      "epoch": 4.310623183726187,
-      "grad_norm": 0.6349745392799377,
-      "learning_rate": 0.00034182092446934597,
-      "loss": 3.3783,
+      "epoch": 4.318059299191375,
+      "grad_norm": 0.6316375732421875,
+      "learning_rate": 0.0003413815434430653,
+      "loss": 3.3784,
       "step": 40050
     },
     {
-      "epoch": 4.3160047357657945,
-      "grad_norm": 0.6404194831848145,
-      "learning_rate": 0.00034149768343928456,
-      "loss": 3.3909,
+      "epoch": 4.32345013477089,
+      "grad_norm": 0.6225030422210693,
+      "learning_rate": 0.0003410577441985968,
+      "loss": 3.406,
       "step": 40100
     },
     {
-      "epoch": 4.321386287805403,
-      "grad_norm": 0.6430397033691406,
-      "learning_rate": 0.0003411744424092231,
-      "loss": 3.3771,
+      "epoch": 4.328840970350404,
+      "grad_norm": 0.6359233260154724,
+      "learning_rate": 0.00034073394495412837,
+      "loss": 3.4153,
       "step": 40150
     },
     {
-      "epoch": 4.326767839845012,
-      "grad_norm": 0.660280704498291,
-      "learning_rate": 0.0003408512013791617,
-      "loss": 3.3799,
+      "epoch": 4.334231805929919,
+      "grad_norm": 0.6805363297462463,
+      "learning_rate": 0.00034041014570965997,
+      "loss": 3.383,
       "step": 40200
     },
     {
-      "epoch": 4.332149391884619,
-      "grad_norm": 0.6253730654716492,
-      "learning_rate": 0.0003405344251697015,
-      "loss": 3.3712,
+      "epoch": 4.339622641509434,
+      "grad_norm": 0.665928840637207,
+      "learning_rate": 0.0003400863464651915,
+      "loss": 3.3884,
       "step": 40250
     },
     {
-      "epoch": 4.337530943924228,
-      "grad_norm": 0.6870384216308594,
-      "learning_rate": 0.0003402111841396401,
-      "loss": 3.381,
+      "epoch": 4.345013477088949,
+      "grad_norm": 0.6551325917243958,
+      "learning_rate": 0.0003397625472207231,
+      "loss": 3.3945,
       "step": 40300
     },
     {
-      "epoch": 4.342912495963836,
-      "grad_norm": 0.665998637676239,
-      "learning_rate": 0.00033988794310957864,
-      "loss": 3.3787,
+      "epoch": 4.350404312668464,
+      "grad_norm": 0.6828498840332031,
+      "learning_rate": 0.0003394387479762547,
+      "loss": 3.3882,
       "step": 40350
     },
     {
-      "epoch": 4.348294048003444,
-      "grad_norm": 0.6889680027961731,
-      "learning_rate": 0.0003395647020795173,
-      "loss": 3.3795,
+      "epoch": 4.355795148247978,
+      "grad_norm": 0.6560102701187134,
+      "learning_rate": 0.0003391149487317863,
+      "loss": 3.4021,
       "step": 40400
     },
     {
-      "epoch": 4.3536756000430525,
-      "grad_norm": 0.7430781722068787,
-      "learning_rate": 0.0003392414610494559,
-      "loss": 3.3779,
+      "epoch": 4.361185983827493,
+      "grad_norm": 0.6644808650016785,
+      "learning_rate": 0.00033879114948731783,
+      "loss": 3.3724,
       "step": 40450
     },
     {
-      "epoch": 4.359057152082661,
-      "grad_norm": 0.629690945148468,
-      "learning_rate": 0.0003389182200193944,
-      "loss": 3.3751,
+      "epoch": 4.366576819407008,
+      "grad_norm": 0.6699093580245972,
+      "learning_rate": 0.00033846735024284944,
+      "loss": 3.3788,
       "step": 40500
     },
     {
-      "epoch": 4.364438704122269,
-      "grad_norm": 0.6800459027290344,
-      "learning_rate": 0.000338594978989333,
-      "loss": 3.3752,
+      "epoch": 4.3719676549865225,
+      "grad_norm": 0.677771270275116,
+      "learning_rate": 0.000338143550998381,
+      "loss": 3.4007,
       "step": 40550
     },
     {
-      "epoch": 4.369820256161877,
-      "grad_norm": 0.6480296850204468,
-      "learning_rate": 0.00033827173795927156,
-      "loss": 3.3933,
+      "epoch": 4.377358490566038,
+      "grad_norm": 0.665482759475708,
+      "learning_rate": 0.00033782622773880194,
+      "loss": 3.386,
       "step": 40600
     },
     {
-      "epoch": 4.375201808201485,
-      "grad_norm": 0.6850379705429077,
-      "learning_rate": 0.00033794849692921015,
-      "loss": 3.3732,
+      "epoch": 4.382749326145553,
+      "grad_norm": 0.6514596343040466,
+      "learning_rate": 0.0003375024284943335,
+      "loss": 3.3772,
       "step": 40650
     },
     {
-      "epoch": 4.3805833602410935,
-      "grad_norm": 0.6677266359329224,
-      "learning_rate": 0.0003376252558991488,
-      "loss": 3.3851,
+      "epoch": 4.388140161725067,
+      "grad_norm": 0.608086347579956,
+      "learning_rate": 0.0003371786292498651,
+      "loss": 3.4051,
       "step": 40700
     },
     {
-      "epoch": 4.385964912280702,
-      "grad_norm": 0.6740007996559143,
-      "learning_rate": 0.00033730201486908734,
-      "loss": 3.3816,
+      "epoch": 4.393530997304582,
+      "grad_norm": 0.6772755980491638,
+      "learning_rate": 0.0003368548300053966,
+      "loss": 3.4013,
       "step": 40750
     },
     {
-      "epoch": 4.39134646432031,
-      "grad_norm": 0.6995536684989929,
-      "learning_rate": 0.00033697877383902594,
-      "loss": 3.3898,
+      "epoch": 4.398921832884097,
+      "grad_norm": 0.6337803602218628,
+      "learning_rate": 0.0003365310307609282,
+      "loss": 3.3755,
       "step": 40800
     },
     {
-      "epoch": 4.396728016359918,
-      "grad_norm": 0.6835097074508667,
-      "learning_rate": 0.00033665553280896453,
-      "loss": 3.3867,
+      "epoch": 4.404312668463612,
+      "grad_norm": 0.6308083534240723,
+      "learning_rate": 0.00033620723151645975,
+      "loss": 3.3956,
       "step": 40850
     },
     {
-      "epoch": 4.402109568399527,
-      "grad_norm": 0.6530295014381409,
-      "learning_rate": 0.00033633229177890307,
-      "loss": 3.3737,
+      "epoch": 4.409703504043127,
+      "grad_norm": 0.6387740969657898,
+      "learning_rate": 0.0003358834322719913,
+      "loss": 3.3838,
       "step": 40900
     },
     {
-      "epoch": 4.4074911204391345,
-      "grad_norm": 0.6370556354522705,
-      "learning_rate": 0.00033600905074884167,
-      "loss": 3.3841,
+      "epoch": 4.415094339622642,
+      "grad_norm": 0.6903748512268066,
+      "learning_rate": 0.0003355596330275229,
+      "loss": 3.3916,
       "step": 40950
     },
     {
-      "epoch": 4.412872672478743,
-      "grad_norm": 0.6709283590316772,
-      "learning_rate": 0.0003356858097187803,
-      "loss": 3.3871,
+      "epoch": 4.420485175202156,
+      "grad_norm": 0.6282992959022522,
+      "learning_rate": 0.00033523583378305445,
+      "loss": 3.4027,
       "step": 41000
     },
     {
-      "epoch": 4.412872672478743,
-      "eval_accuracy": 0.37784802307700416,
-      "eval_loss": 3.4330832958221436,
-      "eval_runtime": 180.4989,
-      "eval_samples_per_second": 99.785,
-      "eval_steps_per_second": 6.238,
+      "epoch": 4.420485175202156,
+      "eval_accuracy": 0.3776030108149825,
+      "eval_loss": 3.4350337982177734,
+      "eval_runtime": 188.6899,
+      "eval_samples_per_second": 95.453,
+      "eval_steps_per_second": 5.967,
       "step": 41000
     },
     {
-      "epoch": 4.418254224518351,
-      "grad_norm": 0.6717859506607056,
-      "learning_rate": 0.00033536256868871886,
-      "loss": 3.3859,
+      "epoch": 4.425876010781671,
+      "grad_norm": 0.636599063873291,
+      "learning_rate": 0.00033491203453858606,
+      "loss": 3.4016,
       "step": 41050
     },
     {
-      "epoch": 4.423635776557959,
-      "grad_norm": 0.6793646216392517,
-      "learning_rate": 0.00033503932765865745,
-      "loss": 3.3893,
+      "epoch": 4.431266846361186,
+      "grad_norm": 0.6283525824546814,
+      "learning_rate": 0.0003345882352941176,
+      "loss": 3.3849,
       "step": 41100
     },
     {
-      "epoch": 4.429017328597568,
-      "grad_norm": 0.6531360149383545,
-      "learning_rate": 0.000334716086628596,
-      "loss": 3.4012,
+      "epoch": 4.436657681940701,
+      "grad_norm": 0.6235990524291992,
+      "learning_rate": 0.0003342644360496492,
+      "loss": 3.3655,
       "step": 41150
     },
     {
-      "epoch": 4.4343988806371755,
-      "grad_norm": 0.655354917049408,
-      "learning_rate": 0.0003343928455985346,
-      "loss": 3.3705,
+      "epoch": 4.442048517520216,
+      "grad_norm": 0.6474363803863525,
+      "learning_rate": 0.00033394063680518076,
+      "loss": 3.3996,
       "step": 41200
     },
     {
-      "epoch": 4.439780432676784,
-      "grad_norm": 0.6870447993278503,
-      "learning_rate": 0.00033406960456847324,
-      "loss": 3.3836,
+      "epoch": 4.44743935309973,
+      "grad_norm": 0.7136996984481812,
+      "learning_rate": 0.00033361683756071237,
+      "loss": 3.3883,
       "step": 41250
     },
     {
-      "epoch": 4.445161984716393,
-      "grad_norm": 0.6646298170089722,
-      "learning_rate": 0.0003337463635384118,
-      "loss": 3.3703,
+      "epoch": 4.452830188679245,
+      "grad_norm": 0.6439403295516968,
+      "learning_rate": 0.0003332930383162439,
+      "loss": 3.3851,
       "step": 41300
     },
     {
-      "epoch": 4.450543536756,
-      "grad_norm": 0.7067282795906067,
-      "learning_rate": 0.00033342312250835037,
-      "loss": 3.3779,
+      "epoch": 4.45822102425876,
+      "grad_norm": 0.6484495997428894,
+      "learning_rate": 0.0003329692390717754,
+      "loss": 3.3777,
       "step": 41350
     },
     {
-      "epoch": 4.455925088795609,
-      "grad_norm": 0.6601843237876892,
-      "learning_rate": 0.00033309988147828896,
-      "loss": 3.3896,
+      "epoch": 4.463611859838275,
+      "grad_norm": 0.6435943245887756,
+      "learning_rate": 0.0003326454398273071,
+      "loss": 3.3884,
       "step": 41400
     },
     {
-      "epoch": 4.461306640835216,
-      "grad_norm": 0.6824533343315125,
-      "learning_rate": 0.0003327766404482275,
-      "loss": 3.3798,
+      "epoch": 4.46900269541779,
+      "grad_norm": 0.7229622006416321,
+      "learning_rate": 0.00033232164058283857,
+      "loss": 3.3741,
       "step": 41450
     },
     {
-      "epoch": 4.466688192874825,
-      "grad_norm": 0.6678142547607422,
-      "learning_rate": 0.0003324533994181661,
-      "loss": 3.385,
+      "epoch": 4.474393530997305,
+      "grad_norm": 0.615696132183075,
+      "learning_rate": 0.0003319978413383702,
+      "loss": 3.3795,
       "step": 41500
     },
     {
-      "epoch": 4.4720697449144335,
-      "grad_norm": 0.6664413213729858,
-      "learning_rate": 0.00033213015838810475,
-      "loss": 3.4065,
+      "epoch": 4.479784366576819,
+      "grad_norm": 0.682884931564331,
+      "learning_rate": 0.0003316805180787911,
+      "loss": 3.4081,
       "step": 41550
     },
     {
-      "epoch": 4.477451296954041,
-      "grad_norm": 0.6379727721214294,
-      "learning_rate": 0.0003318069173580433,
-      "loss": 3.3982,
+      "epoch": 4.485175202156334,
+      "grad_norm": 0.6571813225746155,
+      "learning_rate": 0.0003313567188343227,
+      "loss": 3.3991,
       "step": 41600
     },
     {
-      "epoch": 4.48283284899365,
-      "grad_norm": 0.7035315036773682,
-      "learning_rate": 0.0003314836763279819,
-      "loss": 3.3971,
+      "epoch": 4.490566037735849,
+      "grad_norm": 0.6233645081520081,
+      "learning_rate": 0.00033103291958985423,
+      "loss": 3.4019,
       "step": 41650
     },
     {
-      "epoch": 4.488214401033258,
-      "grad_norm": 0.6721235513687134,
-      "learning_rate": 0.0003311604352979204,
-      "loss": 3.3991,
+      "epoch": 4.495956873315364,
+      "grad_norm": 0.6672859787940979,
+      "learning_rate": 0.00033070912034538583,
+      "loss": 3.4047,
       "step": 41700
     },
     {
-      "epoch": 4.493595953072866,
-      "grad_norm": 0.6236030459403992,
-      "learning_rate": 0.000330837194267859,
-      "loss": 3.3777,
+      "epoch": 4.501347708894879,
+      "grad_norm": 0.7237746119499207,
+      "learning_rate": 0.0003303853211009174,
+      "loss": 3.3885,
       "step": 41750
     },
     {
-      "epoch": 4.4989775051124745,
-      "grad_norm": 0.6594476103782654,
-      "learning_rate": 0.0003305139532377976,
-      "loss": 3.392,
+      "epoch": 4.506738544474393,
+      "grad_norm": 0.6526361107826233,
+      "learning_rate": 0.000330061521856449,
+      "loss": 3.4006,
       "step": 41800
     },
     {
-      "epoch": 4.504359057152083,
-      "grad_norm": 0.6631010174751282,
-      "learning_rate": 0.0003301907122077362,
-      "loss": 3.3953,
+      "epoch": 4.512129380053908,
+      "grad_norm": 0.6548643708229065,
+      "learning_rate": 0.00032973772261198054,
+      "loss": 3.3766,
       "step": 41850
     },
     {
-      "epoch": 4.509740609191691,
-      "grad_norm": 0.6775970458984375,
-      "learning_rate": 0.0003298674711776748,
-      "loss": 3.3798,
+      "epoch": 4.517520215633423,
+      "grad_norm": 0.62809157371521,
+      "learning_rate": 0.00032941392336751214,
+      "loss": 3.4004,
       "step": 41900
     },
     {
-      "epoch": 4.515122161231299,
-      "grad_norm": 0.6755871772766113,
-      "learning_rate": 0.0003295442301476134,
-      "loss": 3.376,
+      "epoch": 4.5229110512129385,
+      "grad_norm": 0.7281858921051025,
+      "learning_rate": 0.0003290901241230437,
+      "loss": 3.4013,
       "step": 41950
     },
     {
-      "epoch": 4.520503713270907,
-      "grad_norm": 0.660190761089325,
-      "learning_rate": 0.00032922098911755194,
-      "loss": 3.3824,
+      "epoch": 4.528301886792453,
+      "grad_norm": 0.6370959281921387,
+      "learning_rate": 0.0003287663248785753,
+      "loss": 3.4027,
       "step": 42000
     },
     {
-      "epoch": 4.520503713270907,
-      "eval_accuracy": 0.37855622259977895,
-      "eval_loss": 3.42791485786438,
-      "eval_runtime": 180.7657,
-      "eval_samples_per_second": 99.637,
-      "eval_steps_per_second": 6.229,
+      "epoch": 4.528301886792453,
+      "eval_accuracy": 0.3783059949991422,
+      "eval_loss": 3.430859088897705,
+      "eval_runtime": 184.5354,
+      "eval_samples_per_second": 97.602,
+      "eval_steps_per_second": 6.102,
       "step": 42000
     },
     {
-      "epoch": 4.5258852653105155,
-      "grad_norm": 0.6533682942390442,
-      "learning_rate": 0.00032889774808749053,
-      "loss": 3.3913,
+      "epoch": 4.533692722371968,
+      "grad_norm": 0.6753497123718262,
+      "learning_rate": 0.00032844252563410685,
+      "loss": 3.3765,
       "step": 42050
     },
     {
-      "epoch": 4.531266817350124,
-      "grad_norm": 0.6714156270027161,
-      "learning_rate": 0.0003285745070574292,
-      "loss": 3.3873,
+      "epoch": 4.539083557951482,
+      "grad_norm": 0.6597613096237183,
+      "learning_rate": 0.00032811872638963834,
+      "loss": 3.3869,
       "step": 42100
     },
     {
-      "epoch": 4.536648369389732,
-      "grad_norm": 0.6406194567680359,
-      "learning_rate": 0.0003282512660273677,
-      "loss": 3.3876,
+      "epoch": 4.544474393530997,
+      "grad_norm": 0.6980563402175903,
+      "learning_rate": 0.00032779492714516995,
+      "loss": 3.3917,
       "step": 42150
     },
     {
-      "epoch": 4.54202992142934,
-      "grad_norm": 0.672727108001709,
-      "learning_rate": 0.0003279280249973063,
-      "loss": 3.3871,
+      "epoch": 4.549865229110512,
+      "grad_norm": 0.6639866232872009,
+      "learning_rate": 0.0003274711279007015,
+      "loss": 3.4029,
       "step": 42200
     },
     {
-      "epoch": 4.547411473468949,
-      "grad_norm": 0.6561397314071655,
-      "learning_rate": 0.00032760478396724486,
-      "loss": 3.3847,
+      "epoch": 4.555256064690027,
+      "grad_norm": 0.6949554681777954,
+      "learning_rate": 0.0003271473286562331,
+      "loss": 3.3972,
       "step": 42250
     },
     {
-      "epoch": 4.5527930255085565,
-      "grad_norm": 0.7014278173446655,
-      "learning_rate": 0.0003272880077577847,
-      "loss": 3.4011,
+      "epoch": 4.560646900269542,
+      "grad_norm": 0.6414914131164551,
+      "learning_rate": 0.00032682352941176466,
+      "loss": 3.3903,
       "step": 42300
     },
     {
-      "epoch": 4.558174577548165,
-      "grad_norm": 0.6329679489135742,
-      "learning_rate": 0.00032696476672772326,
-      "loss": 3.3918,
+      "epoch": 4.566037735849057,
+      "grad_norm": 0.6200968623161316,
+      "learning_rate": 0.00032649973016729626,
+      "loss": 3.4121,
       "step": 42350
     },
     {
-      "epoch": 4.563556129587774,
-      "grad_norm": 0.6573483943939209,
-      "learning_rate": 0.00032664152569766185,
-      "loss": 3.4036,
+      "epoch": 4.571428571428571,
+      "grad_norm": 0.6413687467575073,
+      "learning_rate": 0.0003261759309228278,
+      "loss": 3.3893,
       "step": 42400
     },
     {
-      "epoch": 4.568937681627381,
-      "grad_norm": 0.6461696624755859,
-      "learning_rate": 0.0003263182846676004,
-      "loss": 3.3831,
+      "epoch": 4.576819407008086,
+      "grad_norm": 0.6251882910728455,
+      "learning_rate": 0.0003258521316783594,
+      "loss": 3.4008,
       "step": 42450
     },
     {
-      "epoch": 4.57431923366699,
-      "grad_norm": 0.6436484456062317,
-      "learning_rate": 0.00032599504363753904,
-      "loss": 3.4091,
+      "epoch": 4.5822102425876015,
+      "grad_norm": 0.7066326141357422,
+      "learning_rate": 0.00032552833243389097,
+      "loss": 3.3838,
       "step": 42500
     },
     {
-      "epoch": 4.579700785706597,
-      "grad_norm": 0.683167040348053,
-      "learning_rate": 0.00032567180260747764,
-      "loss": 3.3936,
+      "epoch": 4.587601078167116,
+      "grad_norm": 0.7187201380729675,
+      "learning_rate": 0.0003252045331894225,
+      "loss": 3.3958,
       "step": 42550
     },
     {
-      "epoch": 4.585082337746206,
-      "grad_norm": 0.6897136569023132,
-      "learning_rate": 0.0003253485615774162,
-      "loss": 3.3839,
+      "epoch": 4.592991913746631,
+      "grad_norm": 0.7195764183998108,
+      "learning_rate": 0.0003248807339449541,
+      "loss": 3.3767,
       "step": 42600
     },
     {
-      "epoch": 4.5904638897858145,
-      "grad_norm": 0.7163823246955872,
-      "learning_rate": 0.00032502532054735477,
-      "loss": 3.3933,
+      "epoch": 4.598382749326145,
+      "grad_norm": 0.6276376247406006,
+      "learning_rate": 0.00032455693470048567,
+      "loss": 3.4029,
       "step": 42650
     },
     {
-      "epoch": 4.595845441825422,
-      "grad_norm": 0.6382591724395752,
-      "learning_rate": 0.00032470207951729337,
-      "loss": 3.3879,
+      "epoch": 4.60377358490566,
+      "grad_norm": 0.628013551235199,
+      "learning_rate": 0.0003242331354560173,
+      "loss": 3.3959,
       "step": 42700
     },
     {
-      "epoch": 4.601226993865031,
-      "grad_norm": 0.6253649592399597,
-      "learning_rate": 0.0003243788384872319,
-      "loss": 3.3937,
+      "epoch": 4.609164420485175,
+      "grad_norm": 0.6547573804855347,
+      "learning_rate": 0.0003239093362115488,
+      "loss": 3.387,
       "step": 42750
     },
     {
-      "epoch": 4.606608545904638,
-      "grad_norm": 0.6715789437294006,
-      "learning_rate": 0.00032405559745717056,
-      "loss": 3.3917,
+      "epoch": 4.6145552560646905,
+      "grad_norm": 0.6338829398155212,
+      "learning_rate": 0.00032358553696708043,
+      "loss": 3.389,
       "step": 42800
     },
     {
-      "epoch": 4.611990097944247,
-      "grad_norm": 0.7270500659942627,
-      "learning_rate": 0.00032373235642710915,
-      "loss": 3.3983,
+      "epoch": 4.619946091644205,
+      "grad_norm": 0.6385948061943054,
+      "learning_rate": 0.00032326173772261193,
+      "loss": 3.3959,
       "step": 42850
     },
     {
-      "epoch": 4.6173716499838555,
-      "grad_norm": 0.6829226613044739,
-      "learning_rate": 0.0003234091153970477,
-      "loss": 3.4023,
+      "epoch": 4.62533692722372,
+      "grad_norm": 0.7105038166046143,
+      "learning_rate": 0.00032293793847814353,
+      "loss": 3.4078,
       "step": 42900
     },
     {
-      "epoch": 4.622753202023463,
-      "grad_norm": 0.7018654942512512,
-      "learning_rate": 0.0003230858743669863,
-      "loss": 3.3912,
+      "epoch": 4.630727762803234,
+      "grad_norm": 0.644783616065979,
+      "learning_rate": 0.0003226141392336751,
+      "loss": 3.417,
       "step": 42950
     },
     {
-      "epoch": 4.628134754063072,
-      "grad_norm": 0.6380162239074707,
-      "learning_rate": 0.0003227626333369248,
-      "loss": 3.3943,
+      "epoch": 4.636118598382749,
+      "grad_norm": 0.6668684482574463,
+      "learning_rate": 0.00032229033998920663,
+      "loss": 3.3913,
       "step": 43000
     },
     {
-      "epoch": 4.628134754063072,
-      "eval_accuracy": 0.37904103178520715,
-      "eval_loss": 3.421799421310425,
-      "eval_runtime": 180.2956,
-      "eval_samples_per_second": 99.897,
-      "eval_steps_per_second": 6.245,
+      "epoch": 4.636118598382749,
+      "eval_accuracy": 0.3783892231112081,
+      "eval_loss": 3.4260003566741943,
+      "eval_runtime": 184.7846,
+      "eval_samples_per_second": 97.47,
+      "eval_steps_per_second": 6.094,
       "step": 43000
     },
     {
-      "epoch": 4.63351630610268,
-      "grad_norm": 0.6637802124023438,
-      "learning_rate": 0.0003224393923068635,
-      "loss": 3.3806,
+      "epoch": 4.6415094339622645,
+      "grad_norm": 0.6323847770690918,
+      "learning_rate": 0.00032196654074473824,
+      "loss": 3.3957,
       "step": 43050
     },
     {
-      "epoch": 4.638897858142288,
-      "grad_norm": 0.6968106031417847,
-      "learning_rate": 0.00032211615127680207,
-      "loss": 3.3677,
+      "epoch": 4.646900269541779,
+      "grad_norm": 0.6507082581520081,
+      "learning_rate": 0.0003216427415002698,
+      "loss": 3.3885,
       "step": 43100
     },
     {
-      "epoch": 4.6442794101818965,
-      "grad_norm": 0.6918413043022156,
-      "learning_rate": 0.0003217929102467406,
-      "loss": 3.3913,
+      "epoch": 4.652291105121294,
+      "grad_norm": 0.6631735563278198,
+      "learning_rate": 0.0003213189422558014,
+      "loss": 3.3978,
       "step": 43150
     },
     {
-      "epoch": 4.649660962221505,
-      "grad_norm": 0.637776255607605,
-      "learning_rate": 0.0003214696692166792,
-      "loss": 3.3909,
+      "epoch": 4.657681940700809,
+      "grad_norm": 0.7026782631874084,
+      "learning_rate": 0.00032099514301133295,
+      "loss": 3.3962,
       "step": 43200
     },
     {
-      "epoch": 4.655042514261113,
-      "grad_norm": 0.666828453540802,
-      "learning_rate": 0.0003211464281866178,
-      "loss": 3.3732,
+      "epoch": 4.663072776280323,
+      "grad_norm": 0.6135462522506714,
+      "learning_rate": 0.00032067134376686455,
+      "loss": 3.3845,
       "step": 43250
     },
     {
-      "epoch": 4.660424066300721,
-      "grad_norm": 0.6439411640167236,
-      "learning_rate": 0.00032082318715655634,
-      "loss": 3.3776,
+      "epoch": 4.668463611859838,
+      "grad_norm": 0.6497666239738464,
+      "learning_rate": 0.0003203475445223961,
+      "loss": 3.3851,
       "step": 43300
     },
     {
-      "epoch": 4.665805618340329,
-      "grad_norm": 0.6902055740356445,
-      "learning_rate": 0.000320499946126495,
-      "loss": 3.3948,
+      "epoch": 4.6738544474393535,
+      "grad_norm": 0.6379224061965942,
+      "learning_rate": 0.0003200237452779277,
+      "loss": 3.3819,
       "step": 43350
     },
     {
-      "epoch": 4.6711871703799375,
-      "grad_norm": 0.7745203375816345,
-      "learning_rate": 0.0003201767050964336,
-      "loss": 3.3795,
+      "epoch": 4.679245283018868,
+      "grad_norm": 0.6403704881668091,
+      "learning_rate": 0.00031969994603345926,
+      "loss": 3.4085,
       "step": 43400
     },
     {
-      "epoch": 4.676568722419546,
-      "grad_norm": 0.671711266040802,
-      "learning_rate": 0.0003198534640663721,
-      "loss": 3.3798,
+      "epoch": 4.684636118598383,
+      "grad_norm": 0.6845625638961792,
+      "learning_rate": 0.00031937614678899075,
+      "loss": 3.4026,
       "step": 43450
     },
     {
-      "epoch": 4.681950274459154,
-      "grad_norm": 0.6714461445808411,
-      "learning_rate": 0.00031953668785691193,
-      "loss": 3.3852,
+      "epoch": 4.690026954177897,
+      "grad_norm": 0.6924819350242615,
+      "learning_rate": 0.00031905234754452236,
+      "loss": 3.4098,
       "step": 43500
     },
     {
-      "epoch": 4.687331826498762,
-      "grad_norm": 0.6817046999931335,
-      "learning_rate": 0.0003192134468268505,
-      "loss": 3.3894,
+      "epoch": 4.695417789757412,
+      "grad_norm": 0.6407290101051331,
+      "learning_rate": 0.0003187285483000539,
+      "loss": 3.3758,
       "step": 43550
     },
     {
-      "epoch": 4.692713378538371,
-      "grad_norm": 0.6941414475440979,
-      "learning_rate": 0.0003188902057967891,
-      "loss": 3.4095,
+      "epoch": 4.7008086253369274,
+      "grad_norm": 0.6609850525856018,
+      "learning_rate": 0.0003184047490555855,
+      "loss": 3.3884,
       "step": 43600
     },
     {
-      "epoch": 4.6980949305779784,
-      "grad_norm": 0.7007185220718384,
-      "learning_rate": 0.00031856696476672766,
-      "loss": 3.3747,
+      "epoch": 4.706199460916442,
+      "grad_norm": 0.6443186402320862,
+      "learning_rate": 0.00031808094981111706,
+      "loss": 3.4078,
       "step": 43650
     },
     {
-      "epoch": 4.703476482617587,
-      "grad_norm": 0.7390422821044922,
-      "learning_rate": 0.0003182437237366663,
-      "loss": 3.3933,
+      "epoch": 4.711590296495957,
+      "grad_norm": 0.6441913843154907,
+      "learning_rate": 0.00031775715056664867,
+      "loss": 3.3966,
       "step": 43700
     },
     {
-      "epoch": 4.7088580346571955,
-      "grad_norm": 0.6899697780609131,
-      "learning_rate": 0.0003179204827066049,
-      "loss": 3.3846,
+      "epoch": 4.716981132075472,
+      "grad_norm": 0.6135480999946594,
+      "learning_rate": 0.0003174333513221802,
+      "loss": 3.3789,
       "step": 43750
     },
     {
-      "epoch": 4.714239586696803,
-      "grad_norm": 0.6393377184867859,
-      "learning_rate": 0.00031759724167654344,
-      "loss": 3.3877,
+      "epoch": 4.722371967654986,
+      "grad_norm": 0.6646343469619751,
+      "learning_rate": 0.00031710955207771177,
+      "loss": 3.3885,
       "step": 43800
     },
     {
-      "epoch": 4.719621138736412,
-      "grad_norm": 0.6704504489898682,
-      "learning_rate": 0.00031727400064648204,
-      "loss": 3.3905,
+      "epoch": 4.727762803234501,
+      "grad_norm": 0.6587278842926025,
+      "learning_rate": 0.0003167857528332434,
+      "loss": 3.4045,
       "step": 43850
     },
     {
-      "epoch": 4.725002690776019,
-      "grad_norm": 0.6651193499565125,
-      "learning_rate": 0.0003169507596164206,
-      "loss": 3.3773,
+      "epoch": 4.7331536388140165,
+      "grad_norm": 0.6713588833808899,
+      "learning_rate": 0.0003164619535887749,
+      "loss": 3.3826,
       "step": 43900
     },
     {
-      "epoch": 4.730384242815628,
-      "grad_norm": 0.6523035168647766,
-      "learning_rate": 0.0003166275185863592,
-      "loss": 3.4,
+      "epoch": 4.738544474393531,
+      "grad_norm": 0.6753560900688171,
+      "learning_rate": 0.00031613815434430653,
+      "loss": 3.3928,
       "step": 43950
     },
     {
-      "epoch": 4.7357657948552365,
-      "grad_norm": 0.6419060230255127,
-      "learning_rate": 0.0003163042775562978,
-      "loss": 3.3728,
+      "epoch": 4.743935309973046,
+      "grad_norm": 0.6197715401649475,
+      "learning_rate": 0.0003158143550998381,
+      "loss": 3.3965,
       "step": 44000
     },
     {
-      "epoch": 4.7357657948552365,
-      "eval_accuracy": 0.37933276478898903,
-      "eval_loss": 3.417551040649414,
-      "eval_runtime": 180.8101,
-      "eval_samples_per_second": 99.613,
-      "eval_steps_per_second": 6.228,
+      "epoch": 4.743935309973046,
+      "eval_accuracy": 0.3794112121739911,
+      "eval_loss": 3.420830726623535,
+      "eval_runtime": 185.017,
+      "eval_samples_per_second": 97.348,
+      "eval_steps_per_second": 6.086,
       "step": 44000
     },
     {
-      "epoch": 4.741147346894844,
-      "grad_norm": 0.6831957697868347,
-      "learning_rate": 0.00031598103652623636,
-      "loss": 3.3938,
+      "epoch": 4.74932614555256,
+      "grad_norm": 0.6172311305999756,
+      "learning_rate": 0.0003154905558553697,
+      "loss": 3.4111,
       "step": 44050
     },
     {
-      "epoch": 4.746528898934453,
-      "grad_norm": 0.6325632929801941,
-      "learning_rate": 0.00031565779549617496,
-      "loss": 3.4015,
+      "epoch": 4.754716981132075,
+      "grad_norm": 0.7061458230018616,
+      "learning_rate": 0.0003151667566109012,
+      "loss": 3.3835,
       "step": 44100
     },
     {
-      "epoch": 4.751910450974061,
-      "grad_norm": 0.687362790107727,
-      "learning_rate": 0.00031533455446611355,
-      "loss": 3.3946,
+      "epoch": 4.7601078167115904,
+      "grad_norm": 0.6917078495025635,
+      "learning_rate": 0.00031484295736643284,
+      "loss": 3.3896,
       "step": 44150
     },
     {
-      "epoch": 4.757292003013669,
-      "grad_norm": 0.6923133134841919,
-      "learning_rate": 0.0003150113134360521,
-      "loss": 3.3823,
+      "epoch": 4.765498652291106,
+      "grad_norm": 0.6253639459609985,
+      "learning_rate": 0.00031451915812196434,
+      "loss": 3.3832,
       "step": 44200
     },
     {
-      "epoch": 4.7626735550532775,
-      "grad_norm": 0.709951639175415,
-      "learning_rate": 0.0003146880724059907,
-      "loss": 3.3947,
+      "epoch": 4.77088948787062,
+      "grad_norm": 0.6521839499473572,
+      "learning_rate": 0.0003142018348623853,
+      "loss": 3.4013,
       "step": 44250
     },
     {
-      "epoch": 4.768055107092886,
-      "grad_norm": 0.6787405610084534,
-      "learning_rate": 0.00031436483137592934,
-      "loss": 3.3734,
+      "epoch": 4.776280323450135,
+      "grad_norm": 0.6730448603630066,
+      "learning_rate": 0.00031387803561791684,
+      "loss": 3.3867,
       "step": 44300
     },
     {
-      "epoch": 4.773436659132494,
-      "grad_norm": 0.7011154890060425,
-      "learning_rate": 0.0003140415903458679,
-      "loss": 3.3932,
+      "epoch": 4.781671159029649,
+      "grad_norm": 0.6574965119361877,
+      "learning_rate": 0.00031355423637344844,
+      "loss": 3.3834,
       "step": 44350
     },
     {
-      "epoch": 4.778818211172102,
-      "grad_norm": 0.7104355096817017,
-      "learning_rate": 0.00031371834931580647,
-      "loss": 3.3876,
+      "epoch": 4.787061994609164,
+      "grad_norm": 0.6981621980667114,
+      "learning_rate": 0.00031323043712898,
+      "loss": 3.3903,
       "step": 44400
     },
     {
-      "epoch": 4.78419976321171,
-      "grad_norm": 0.65087890625,
-      "learning_rate": 0.000313395108285745,
-      "loss": 3.3826,
+      "epoch": 4.7924528301886795,
+      "grad_norm": 0.6450235843658447,
+      "learning_rate": 0.0003129066378845116,
+      "loss": 3.4046,
       "step": 44450
     },
     {
-      "epoch": 4.7895813152513185,
-      "grad_norm": 0.7193938493728638,
-      "learning_rate": 0.0003130718672556836,
-      "loss": 3.3899,
+      "epoch": 4.797843665768194,
+      "grad_norm": 0.6787946820259094,
+      "learning_rate": 0.00031258283864004315,
+      "loss": 3.3993,
       "step": 44500
     },
     {
-      "epoch": 4.794962867290927,
-      "grad_norm": 0.6938375234603882,
-      "learning_rate": 0.0003127486262256222,
-      "loss": 3.367,
+      "epoch": 4.803234501347709,
+      "grad_norm": 0.6712046265602112,
+      "learning_rate": 0.0003122590393955747,
+      "loss": 3.3929,
       "step": 44550
     },
     {
-      "epoch": 4.800344419330535,
-      "grad_norm": 0.6607935428619385,
-      "learning_rate": 0.0003124253851955608,
-      "loss": 3.3828,
+      "epoch": 4.808625336927224,
+      "grad_norm": 0.6471379399299622,
+      "learning_rate": 0.0003119352401511063,
+      "loss": 3.4003,
       "step": 44600
     },
     {
-      "epoch": 4.805725971370143,
-      "grad_norm": 0.7184892296791077,
-      "learning_rate": 0.0003121021441654994,
-      "loss": 3.4025,
+      "epoch": 4.814016172506738,
+      "grad_norm": 0.665787935256958,
+      "learning_rate": 0.00031161144090663786,
+      "loss": 3.3974,
       "step": 44650
     },
     {
-      "epoch": 4.811107523409751,
-      "grad_norm": 0.6583475470542908,
-      "learning_rate": 0.000311778903135438,
-      "loss": 3.4065,
+      "epoch": 4.819407008086253,
+      "grad_norm": 0.6562718152999878,
+      "learning_rate": 0.00031128764166216946,
+      "loss": 3.3865,
       "step": 44700
     },
     {
-      "epoch": 4.8164890754493594,
-      "grad_norm": 0.746714174747467,
-      "learning_rate": 0.0003114556621053765,
-      "loss": 3.4003,
+      "epoch": 4.824797843665769,
+      "grad_norm": 0.665611743927002,
+      "learning_rate": 0.00031096384241770096,
+      "loss": 3.403,
       "step": 44750
     },
     {
-      "epoch": 4.821870627488968,
-      "grad_norm": 0.667801558971405,
-      "learning_rate": 0.0003111324210753151,
-      "loss": 3.3777,
+      "epoch": 4.830188679245283,
+      "grad_norm": 0.6378157734870911,
+      "learning_rate": 0.0003106400431732326,
+      "loss": 3.4124,
       "step": 44800
     },
     {
-      "epoch": 4.827252179528576,
-      "grad_norm": 0.6681658029556274,
-      "learning_rate": 0.00031080918004525377,
-      "loss": 3.3846,
+      "epoch": 4.835579514824798,
+      "grad_norm": 0.6545462608337402,
+      "learning_rate": 0.0003103162439287641,
+      "loss": 3.4008,
       "step": 44850
     },
     {
-      "epoch": 4.832633731568184,
-      "grad_norm": 0.6859903931617737,
-      "learning_rate": 0.0003104859390151923,
-      "loss": 3.3843,
+      "epoch": 4.840970350404312,
+      "grad_norm": 0.6839264035224915,
+      "learning_rate": 0.0003099924446842957,
+      "loss": 3.4062,
       "step": 44900
     },
     {
-      "epoch": 4.838015283607793,
-      "grad_norm": 0.6311453580856323,
-      "learning_rate": 0.0003101626979851309,
-      "loss": 3.3754,
+      "epoch": 4.846361185983827,
+      "grad_norm": 0.6348554491996765,
+      "learning_rate": 0.00030966864543982727,
+      "loss": 3.4098,
       "step": 44950
     },
     {
-      "epoch": 4.8433968356474,
-      "grad_norm": 0.6764549016952515,
-      "learning_rate": 0.00030983945695506945,
-      "loss": 3.3897,
+      "epoch": 4.8517520215633425,
+      "grad_norm": 0.6661449670791626,
+      "learning_rate": 0.0003093448461953588,
+      "loss": 3.4006,
       "step": 45000
     },
     {
-      "epoch": 4.8433968356474,
-      "eval_accuracy": 0.3799441545887211,
-      "eval_loss": 3.4145073890686035,
-      "eval_runtime": 180.8088,
-      "eval_samples_per_second": 99.614,
-      "eval_steps_per_second": 6.228,
+      "epoch": 4.8517520215633425,
+      "eval_accuracy": 0.3796366669162061,
+      "eval_loss": 3.4154839515686035,
+      "eval_runtime": 184.8867,
+      "eval_samples_per_second": 97.416,
+      "eval_steps_per_second": 6.09,
       "step": 45000
     },
     {
-      "epoch": 4.848778387687009,
-      "grad_norm": 0.7071652412414551,
-      "learning_rate": 0.00030951621592500804,
-      "loss": 3.3874,
+      "epoch": 4.857142857142857,
+      "grad_norm": 0.7101733684539795,
+      "learning_rate": 0.0003090210469508904,
+      "loss": 3.3889,
       "step": 45050
     },
     {
-      "epoch": 4.8541599397266175,
-      "grad_norm": 0.6975886821746826,
-      "learning_rate": 0.00030919297489494663,
-      "loss": 3.381,
+      "epoch": 4.862533692722372,
+      "grad_norm": 0.6474313139915466,
+      "learning_rate": 0.00030869724770642197,
+      "loss": 3.394,
       "step": 45100
     },
     {
-      "epoch": 4.859541491766225,
-      "grad_norm": 0.6888333559036255,
-      "learning_rate": 0.00030886973386488523,
-      "loss": 3.3775,
+      "epoch": 4.867924528301887,
+      "grad_norm": 0.6456648707389832,
+      "learning_rate": 0.0003083734484619536,
+      "loss": 3.3986,
       "step": 45150
     },
     {
-      "epoch": 4.864923043805834,
-      "grad_norm": 0.6672408580780029,
-      "learning_rate": 0.0003085464928348238,
-      "loss": 3.4037,
+      "epoch": 4.873315363881401,
+      "grad_norm": 0.7081680297851562,
+      "learning_rate": 0.00030804964921748513,
+      "loss": 3.3801,
       "step": 45200
     },
     {
-      "epoch": 4.870304595845441,
-      "grad_norm": 0.710537850856781,
-      "learning_rate": 0.0003082232518047624,
-      "loss": 3.391,
+      "epoch": 4.878706199460916,
+      "grad_norm": 0.652946412563324,
+      "learning_rate": 0.00030772584997301673,
+      "loss": 3.3928,
       "step": 45250
     },
     {
-      "epoch": 4.87568614788505,
-      "grad_norm": 0.6636947393417358,
-      "learning_rate": 0.00030790001077470096,
-      "loss": 3.3911,
+      "epoch": 4.884097035040432,
+      "grad_norm": 0.6638479828834534,
+      "learning_rate": 0.0003074020507285483,
+      "loss": 3.3753,
       "step": 45300
     },
     {
-      "epoch": 4.8810676999246585,
-      "grad_norm": 0.6961594223976135,
-      "learning_rate": 0.00030757676974463955,
-      "loss": 3.3829,
+      "epoch": 4.889487870619946,
+      "grad_norm": 0.6589642763137817,
+      "learning_rate": 0.0003070782514840799,
+      "loss": 3.3901,
       "step": 45350
     },
     {
-      "epoch": 4.886449251964266,
-      "grad_norm": 0.8008875846862793,
-      "learning_rate": 0.0003072535287145781,
-      "loss": 3.3904,
+      "epoch": 4.894878706199461,
+      "grad_norm": 0.6825507283210754,
+      "learning_rate": 0.00030675445223961144,
+      "loss": 3.3957,
       "step": 45400
     },
     {
-      "epoch": 4.891830804003875,
-      "grad_norm": 0.6139016151428223,
-      "learning_rate": 0.00030693028768451674,
-      "loss": 3.3811,
+      "epoch": 4.900269541778976,
+      "grad_norm": 0.6379856467247009,
+      "learning_rate": 0.00030643065299514294,
+      "loss": 3.3954,
       "step": 45450
     },
     {
-      "epoch": 4.897212356043483,
-      "grad_norm": 0.6992993354797363,
-      "learning_rate": 0.00030660704665445534,
-      "loss": 3.3865,
+      "epoch": 4.90566037735849,
+      "grad_norm": 0.6712074279785156,
+      "learning_rate": 0.00030610685375067454,
+      "loss": 3.3868,
       "step": 45500
     },
     {
-      "epoch": 4.902593908083091,
-      "grad_norm": 0.6530194282531738,
-      "learning_rate": 0.0003062838056243939,
-      "loss": 3.3902,
+      "epoch": 4.9110512129380055,
+      "grad_norm": 0.6530936360359192,
+      "learning_rate": 0.0003057830545062061,
+      "loss": 3.3737,
       "step": 45550
     },
     {
-      "epoch": 4.9079754601226995,
-      "grad_norm": 0.6725596189498901,
-      "learning_rate": 0.00030596056459433247,
-      "loss": 3.3907,
+      "epoch": 4.916442048517521,
+      "grad_norm": 0.6042382717132568,
+      "learning_rate": 0.0003054592552617377,
+      "loss": 3.3815,
       "step": 45600
     },
     {
-      "epoch": 4.913357012162308,
-      "grad_norm": 0.670534610748291,
-      "learning_rate": 0.00030563732356427107,
-      "loss": 3.3845,
+      "epoch": 4.921832884097035,
+      "grad_norm": 0.6475586891174316,
+      "learning_rate": 0.00030513545601726925,
+      "loss": 3.3859,
       "step": 45650
     },
     {
-      "epoch": 4.918738564201916,
-      "grad_norm": 0.6930390000343323,
-      "learning_rate": 0.00030531408253420966,
-      "loss": 3.3919,
+      "epoch": 4.92722371967655,
+      "grad_norm": 0.624848484992981,
+      "learning_rate": 0.00030481165677280085,
+      "loss": 3.3982,
       "step": 45700
     },
     {
-      "epoch": 4.924120116241524,
-      "grad_norm": 0.7007985711097717,
-      "learning_rate": 0.00030499084150414826,
-      "loss": 3.375,
+      "epoch": 4.932614555256064,
+      "grad_norm": 0.6341684460639954,
+      "learning_rate": 0.0003044878575283324,
+      "loss": 3.3939,
       "step": 45750
     },
     {
-      "epoch": 4.929501668281132,
-      "grad_norm": 0.6509608030319214,
-      "learning_rate": 0.00030466760047408685,
-      "loss": 3.3936,
+      "epoch": 4.938005390835579,
+      "grad_norm": 0.6945657134056091,
+      "learning_rate": 0.000304164058283864,
+      "loss": 3.3828,
       "step": 45800
     },
     {
-      "epoch": 4.9348832203207404,
-      "grad_norm": 0.7178775668144226,
-      "learning_rate": 0.0003043443594440254,
-      "loss": 3.3867,
+      "epoch": 4.943396226415095,
+      "grad_norm": 0.658644437789917,
+      "learning_rate": 0.00030384025903939556,
+      "loss": 3.3859,
       "step": 45850
     },
     {
-      "epoch": 4.940264772360349,
-      "grad_norm": 0.7314358949661255,
-      "learning_rate": 0.000304021118413964,
-      "loss": 3.3853,
+      "epoch": 4.948787061994609,
+      "grad_norm": 0.6409682035446167,
+      "learning_rate": 0.0003035164597949271,
+      "loss": 3.3902,
       "step": 45900
     },
     {
-      "epoch": 4.945646324399957,
-      "grad_norm": 0.7417964339256287,
-      "learning_rate": 0.0003036978773839025,
-      "loss": 3.3955,
+      "epoch": 4.954177897574124,
+      "grad_norm": 0.6514752507209778,
+      "learning_rate": 0.0003031926605504587,
+      "loss": 3.3919,
       "step": 45950
     },
     {
-      "epoch": 4.951027876439565,
-      "grad_norm": 0.6962642669677734,
-      "learning_rate": 0.0003033746363538412,
-      "loss": 3.3879,
+      "epoch": 4.959568733153639,
+      "grad_norm": 0.6511775851249695,
+      "learning_rate": 0.00030286886130599026,
+      "loss": 3.3816,
       "step": 46000
     },
     {
-      "epoch": 4.951027876439565,
-      "eval_accuracy": 0.3806199755509272,
-      "eval_loss": 3.4076614379882812,
-      "eval_runtime": 180.463,
-      "eval_samples_per_second": 99.804,
-      "eval_steps_per_second": 6.24,
+      "epoch": 4.959568733153639,
+      "eval_accuracy": 0.380506433283161,
+      "eval_loss": 3.412381649017334,
+      "eval_runtime": 184.9517,
+      "eval_samples_per_second": 97.382,
+      "eval_steps_per_second": 6.088,
       "step": 46000
     },
     {
-      "epoch": 4.956409428479174,
-      "grad_norm": 0.6814647912979126,
-      "learning_rate": 0.00030305139532377977,
-      "loss": 3.3941,
+      "epoch": 4.964959568733153,
+      "grad_norm": 0.6196075081825256,
+      "learning_rate": 0.00030254506206152187,
+      "loss": 3.3994,
       "step": 46050
     },
     {
-      "epoch": 4.961790980518781,
-      "grad_norm": 0.6750015020370483,
-      "learning_rate": 0.0003027281542937183,
-      "loss": 3.3918,
+      "epoch": 4.9703504043126685,
+      "grad_norm": 0.6395175457000732,
+      "learning_rate": 0.00030222126281705336,
+      "loss": 3.382,
       "step": 46100
     },
     {
-      "epoch": 4.96717253255839,
-      "grad_norm": 0.6892281174659729,
-      "learning_rate": 0.0003024049132636569,
-      "loss": 3.3945,
+      "epoch": 4.975741239892184,
+      "grad_norm": 0.7017073631286621,
+      "learning_rate": 0.000301897463572585,
+      "loss": 3.3978,
       "step": 46150
     },
     {
-      "epoch": 4.9725540845979985,
-      "grad_norm": 0.6996626257896423,
-      "learning_rate": 0.0003020816722335955,
-      "loss": 3.3739,
+      "epoch": 4.981132075471698,
+      "grad_norm": 0.7031179070472717,
+      "learning_rate": 0.0003015736643281165,
+      "loss": 3.3877,
       "step": 46200
     },
     {
-      "epoch": 4.977935636637606,
-      "grad_norm": 0.6532090902328491,
-      "learning_rate": 0.00030175843120353404,
-      "loss": 3.4021,
+      "epoch": 4.986522911051213,
+      "grad_norm": 0.628025233745575,
+      "learning_rate": 0.00030124986508364807,
+      "loss": 3.3984,
       "step": 46250
     },
     {
-      "epoch": 4.983317188677215,
-      "grad_norm": 0.6986241936683655,
-      "learning_rate": 0.0003014351901734727,
-      "loss": 3.3769,
+      "epoch": 4.991913746630727,
+      "grad_norm": 0.6933519840240479,
+      "learning_rate": 0.0003009260658391797,
+      "loss": 3.4003,
       "step": 46300
     },
     {
-      "epoch": 4.988698740716822,
-      "grad_norm": 0.6778721213340759,
-      "learning_rate": 0.0003011119491434113,
-      "loss": 3.3806,
+      "epoch": 4.997304582210242,
+      "grad_norm": 0.6547604203224182,
+      "learning_rate": 0.0003006022665947112,
+      "loss": 3.3829,
       "step": 46350
     },
     {
-      "epoch": 4.994080292756431,
-      "grad_norm": 0.7227446436882019,
-      "learning_rate": 0.0003007887081133498,
-      "loss": 3.386,
+      "epoch": 5.002695417789758,
+      "grad_norm": 0.6624903678894043,
+      "learning_rate": 0.00030027846735024283,
+      "loss": 3.3527,
       "step": 46400
     },
     {
-      "epoch": 4.9994618447960395,
-      "grad_norm": 0.6771180629730225,
-      "learning_rate": 0.0003004654670832884,
-      "loss": 3.3777,
+      "epoch": 5.008086253369272,
+      "grad_norm": 0.6124040484428406,
+      "learning_rate": 0.00029995466810577443,
+      "loss": 3.2893,
       "step": 46450
     },
     {
-      "epoch": 5.004843396835647,
-      "grad_norm": 0.6935259699821472,
-      "learning_rate": 0.00030014222605322696,
-      "loss": 3.3098,
+      "epoch": 5.013477088948787,
+      "grad_norm": 0.6685596704483032,
+      "learning_rate": 0.00029963086886130593,
+      "loss": 3.2916,
       "step": 46500
     },
     {
-      "epoch": 5.010224948875256,
-      "grad_norm": 0.7119801640510559,
-      "learning_rate": 0.00029981898502316555,
-      "loss": 3.3013,
+      "epoch": 5.018867924528302,
+      "grad_norm": 0.6791301369667053,
+      "learning_rate": 0.00029930706961683754,
+      "loss": 3.2882,
       "step": 46550
     },
     {
-      "epoch": 5.015606500914864,
-      "grad_norm": 0.7470939755439758,
-      "learning_rate": 0.00029949574399310415,
-      "loss": 3.2987,
+      "epoch": 5.024258760107816,
+      "grad_norm": 0.6711662411689758,
+      "learning_rate": 0.0002989832703723691,
+      "loss": 3.2997,
       "step": 46600
     },
     {
-      "epoch": 5.020988052954472,
-      "grad_norm": 0.7172041535377502,
-      "learning_rate": 0.00029917250296304274,
-      "loss": 3.2962,
+      "epoch": 5.0296495956873315,
+      "grad_norm": 0.6637666821479797,
+      "learning_rate": 0.0002986594711279007,
+      "loss": 3.296,
       "step": 46650
     },
     {
-      "epoch": 5.0263696049940805,
-      "grad_norm": 0.708893895149231,
-      "learning_rate": 0.00029884926193298134,
-      "loss": 3.2944,
+      "epoch": 5.035040431266847,
+      "grad_norm": 0.6611721515655518,
+      "learning_rate": 0.00029833567188343224,
+      "loss": 3.3144,
       "step": 46700
     },
     {
-      "epoch": 5.031751157033688,
-      "grad_norm": 0.6856112480163574,
-      "learning_rate": 0.0002985260209029199,
-      "loss": 3.2975,
+      "epoch": 5.040431266846361,
+      "grad_norm": 0.6716077923774719,
+      "learning_rate": 0.00029801187263896385,
+      "loss": 3.3006,
       "step": 46750
     },
     {
-      "epoch": 5.037132709073297,
-      "grad_norm": 0.6789200305938721,
-      "learning_rate": 0.00029820277987285853,
-      "loss": 3.2961,
+      "epoch": 5.045822102425876,
+      "grad_norm": 0.6382834315299988,
+      "learning_rate": 0.0002976880733944954,
+      "loss": 3.3187,
       "step": 46800
     },
     {
-      "epoch": 5.042514261112905,
-      "grad_norm": 0.6745328307151794,
-      "learning_rate": 0.00029787953884279707,
-      "loss": 3.294,
+      "epoch": 5.051212938005391,
+      "grad_norm": 0.6736823916435242,
+      "learning_rate": 0.00029736427415002695,
+      "loss": 3.2997,
       "step": 46850
     },
     {
-      "epoch": 5.047895813152513,
-      "grad_norm": 0.695334792137146,
-      "learning_rate": 0.00029756276263333693,
-      "loss": 3.3119,
+      "epoch": 5.056603773584905,
+      "grad_norm": 0.6748251914978027,
+      "learning_rate": 0.0002970404749055585,
+      "loss": 3.3102,
       "step": 46900
     },
     {
-      "epoch": 5.0532773651921215,
-      "grad_norm": 0.6939895153045654,
-      "learning_rate": 0.00029723952160327547,
-      "loss": 3.2977,
+      "epoch": 5.061994609164421,
+      "grad_norm": 0.6604644060134888,
+      "learning_rate": 0.0002967166756610901,
+      "loss": 3.3199,
       "step": 46950
     },
     {
-      "epoch": 5.05865891723173,
-      "grad_norm": 0.7092863917350769,
-      "learning_rate": 0.00029691628057321406,
-      "loss": 3.3093,
+      "epoch": 5.067385444743936,
+      "grad_norm": 0.6938402652740479,
+      "learning_rate": 0.00029639287641662165,
+      "loss": 3.308,
       "step": 47000
     },
     {
-      "epoch": 5.05865891723173,
-      "eval_accuracy": 0.3807579647184516,
-      "eval_loss": 3.4131321907043457,
-      "eval_runtime": 180.5823,
-      "eval_samples_per_second": 99.738,
-      "eval_steps_per_second": 6.235,
+      "epoch": 5.067385444743936,
+      "eval_accuracy": 0.38030357834161144,
+      "eval_loss": 3.4177451133728027,
+      "eval_runtime": 184.7954,
+      "eval_samples_per_second": 97.465,
+      "eval_steps_per_second": 6.093,
       "step": 47000
     },
     {
-      "epoch": 5.064040469271338,
-      "grad_norm": 0.7017537355422974,
-      "learning_rate": 0.00029659303954315266,
-      "loss": 3.3113,
+      "epoch": 5.07277628032345,
+      "grad_norm": 0.6623622179031372,
+      "learning_rate": 0.00029606907717215326,
+      "loss": 3.3111,
       "step": 47050
     },
     {
-      "epoch": 5.069422021310946,
-      "grad_norm": 0.7066820859909058,
-      "learning_rate": 0.00029626979851309125,
-      "loss": 3.3025,
+      "epoch": 5.078167115902965,
+      "grad_norm": 0.6709842085838318,
+      "learning_rate": 0.0002957452779276848,
+      "loss": 3.3137,
       "step": 47100
     },
     {
-      "epoch": 5.074803573350554,
-      "grad_norm": 0.7007570862770081,
-      "learning_rate": 0.00029594655748302985,
-      "loss": 3.3077,
+      "epoch": 5.083557951482479,
+      "grad_norm": 0.7319429516792297,
+      "learning_rate": 0.00029542147868321636,
+      "loss": 3.3017,
       "step": 47150
     },
     {
-      "epoch": 5.080185125390162,
-      "grad_norm": 0.7130218744277954,
-      "learning_rate": 0.0002956233164529684,
-      "loss": 3.2919,
+      "epoch": 5.0889487870619945,
+      "grad_norm": 0.6315300464630127,
+      "learning_rate": 0.00029509767943874796,
+      "loss": 3.3109,
       "step": 47200
     },
     {
-      "epoch": 5.085566677429771,
-      "grad_norm": 0.7107332944869995,
-      "learning_rate": 0.000295300075422907,
-      "loss": 3.3093,
+      "epoch": 5.09433962264151,
+      "grad_norm": 0.6517070531845093,
+      "learning_rate": 0.0002947738801942795,
+      "loss": 3.3021,
       "step": 47250
     },
     {
-      "epoch": 5.090948229469379,
-      "grad_norm": 0.6778996586799622,
-      "learning_rate": 0.0002949768343928456,
-      "loss": 3.3089,
+      "epoch": 5.099730458221024,
+      "grad_norm": 0.6561506986618042,
+      "learning_rate": 0.0002944500809498111,
+      "loss": 3.3161,
       "step": 47300
     },
     {
-      "epoch": 5.096329781508987,
-      "grad_norm": 0.7302454113960266,
-      "learning_rate": 0.00029465359336278417,
-      "loss": 3.321,
+      "epoch": 5.105121293800539,
+      "grad_norm": 0.6593621969223022,
+      "learning_rate": 0.00029412628170534267,
+      "loss": 3.3206,
       "step": 47350
     },
     {
-      "epoch": 5.101711333548596,
-      "grad_norm": 0.7293109893798828,
-      "learning_rate": 0.0002943303523327227,
-      "loss": 3.3307,
+      "epoch": 5.110512129380054,
+      "grad_norm": 0.6380136013031006,
+      "learning_rate": 0.0002938024824608742,
+      "loss": 3.3299,
       "step": 47400
     },
     {
-      "epoch": 5.107092885588203,
-      "grad_norm": 0.7125155329704285,
-      "learning_rate": 0.00029400711130266136,
-      "loss": 3.3088,
+      "epoch": 5.115902964959568,
+      "grad_norm": 0.703005313873291,
+      "learning_rate": 0.00029347868321640577,
+      "loss": 3.3193,
       "step": 47450
     },
     {
-      "epoch": 5.112474437627812,
-      "grad_norm": 0.7028940916061401,
-      "learning_rate": 0.0002936838702725999,
-      "loss": 3.3038,
+      "epoch": 5.121293800539084,
+      "grad_norm": 0.6386775374412537,
+      "learning_rate": 0.0002931548839719374,
+      "loss": 3.3235,
       "step": 47500
     },
     {
-      "epoch": 5.1178559896674205,
-      "grad_norm": 0.6821138858795166,
-      "learning_rate": 0.0002933606292425385,
-      "loss": 3.3197,
+      "epoch": 5.126684636118599,
+      "grad_norm": 0.7004210352897644,
+      "learning_rate": 0.0002928310847274689,
+      "loss": 3.32,
       "step": 47550
     },
     {
-      "epoch": 5.123237541707028,
-      "grad_norm": 0.6831609010696411,
-      "learning_rate": 0.0002930373882124771,
-      "loss": 3.3003,
+      "epoch": 5.132075471698113,
+      "grad_norm": 0.7443287968635559,
+      "learning_rate": 0.00029250728548300053,
+      "loss": 3.3218,
       "step": 47600
     },
     {
-      "epoch": 5.128619093746637,
-      "grad_norm": 0.6890237331390381,
-      "learning_rate": 0.0002927141471824157,
-      "loss": 3.3091,
+      "epoch": 5.137466307277628,
+      "grad_norm": 0.6982861161231995,
+      "learning_rate": 0.0002921834862385321,
+      "loss": 3.3123,
       "step": 47650
     },
     {
-      "epoch": 5.134000645786244,
-      "grad_norm": 0.7032915949821472,
-      "learning_rate": 0.0002923909061523542,
-      "loss": 3.3137,
+      "epoch": 5.142857142857143,
+      "grad_norm": 0.7046205997467041,
+      "learning_rate": 0.0002918596869940637,
+      "loss": 3.3081,
       "step": 47700
     },
     {
-      "epoch": 5.139382197825853,
-      "grad_norm": 0.6753126382827759,
-      "learning_rate": 0.0002920676651222928,
-      "loss": 3.3034,
+      "epoch": 5.1482479784366575,
+      "grad_norm": 0.6518100500106812,
+      "learning_rate": 0.0002915358877495952,
+      "loss": 3.3084,
       "step": 47750
     },
     {
-      "epoch": 5.1447637498654615,
-      "grad_norm": 0.7659389972686768,
-      "learning_rate": 0.0002917444240922314,
-      "loss": 3.3154,
+      "epoch": 5.153638814016173,
+      "grad_norm": 0.6555050015449524,
+      "learning_rate": 0.0002912120885051268,
+      "loss": 3.3268,
       "step": 47800
     },
     {
-      "epoch": 5.150145301905069,
-      "grad_norm": 0.688945472240448,
-      "learning_rate": 0.00029142118306216996,
-      "loss": 3.3108,
+      "epoch": 5.159029649595688,
+      "grad_norm": 0.6799789667129517,
+      "learning_rate": 0.00029088828926065834,
+      "loss": 3.3014,
       "step": 47850
     },
     {
-      "epoch": 5.155526853944678,
-      "grad_norm": 0.6884469389915466,
-      "learning_rate": 0.0002910979420321086,
-      "loss": 3.3065,
+      "epoch": 5.164420485175202,
+      "grad_norm": 0.6917016506195068,
+      "learning_rate": 0.00029056449001618994,
+      "loss": 3.3357,
       "step": 47900
     },
     {
-      "epoch": 5.160908405984286,
-      "grad_norm": 0.6957643032073975,
-      "learning_rate": 0.00029077470100204715,
-      "loss": 3.31,
+      "epoch": 5.169811320754717,
+      "grad_norm": 0.6385092735290527,
+      "learning_rate": 0.0002902406907717215,
+      "loss": 3.3204,
       "step": 47950
     },
     {
-      "epoch": 5.166289958023894,
-      "grad_norm": 0.7388163805007935,
-      "learning_rate": 0.00029045145997198574,
-      "loss": 3.2902,
+      "epoch": 5.175202156334231,
+      "grad_norm": 0.6348479986190796,
+      "learning_rate": 0.0002899168915272531,
+      "loss": 3.3136,
       "step": 48000
     },
     {
-      "epoch": 5.166289958023894,
-      "eval_accuracy": 0.3809007346130398,
-      "eval_loss": 3.4100382328033447,
-      "eval_runtime": 180.4717,
-      "eval_samples_per_second": 99.8,
-      "eval_steps_per_second": 6.239,
+      "epoch": 5.175202156334231,
+      "eval_accuracy": 0.3807514455451827,
+      "eval_loss": 3.412294387817383,
+      "eval_runtime": 184.954,
+      "eval_samples_per_second": 97.381,
+      "eval_steps_per_second": 6.088,
       "step": 48000
     },
     {
-      "epoch": 5.1716715100635025,
-      "grad_norm": 0.7176282405853271,
-      "learning_rate": 0.00029012821894192433,
-      "loss": 3.315,
+      "epoch": 5.180592991913747,
+      "grad_norm": 0.7083115577697754,
+      "learning_rate": 0.00028959309228278465,
+      "loss": 3.3247,
       "step": 48050
     },
     {
-      "epoch": 5.17705306210311,
-      "grad_norm": 0.6990656852722168,
-      "learning_rate": 0.00028980497791186293,
-      "loss": 3.3085,
+      "epoch": 5.185983827493262,
+      "grad_norm": 0.7016012668609619,
+      "learning_rate": 0.00028926929303831625,
+      "loss": 3.314,
       "step": 48100
     },
     {
-      "epoch": 5.182434614142719,
-      "grad_norm": 0.7440829277038574,
-      "learning_rate": 0.0002894817368818015,
-      "loss": 3.3026,
+      "epoch": 5.191374663072776,
+      "grad_norm": 0.6694549322128296,
+      "learning_rate": 0.0002889454937938478,
+      "loss": 3.3267,
       "step": 48150
     },
     {
-      "epoch": 5.187816166182327,
-      "grad_norm": 0.7810512781143188,
-      "learning_rate": 0.00028915849585174006,
-      "loss": 3.3067,
+      "epoch": 5.196765498652291,
+      "grad_norm": 0.6478196978569031,
+      "learning_rate": 0.00028862169454937935,
+      "loss": 3.3285,
       "step": 48200
     },
     {
-      "epoch": 5.193197718221935,
-      "grad_norm": 0.7514436841011047,
-      "learning_rate": 0.00028883525482167866,
-      "loss": 3.3155,
+      "epoch": 5.202156334231806,
+      "grad_norm": 0.7021739482879639,
+      "learning_rate": 0.0002883043712898003,
+      "loss": 3.3231,
       "step": 48250
     },
     {
-      "epoch": 5.198579270261543,
-      "grad_norm": 0.7587404251098633,
-      "learning_rate": 0.00028851201379161725,
-      "loss": 3.326,
+      "epoch": 5.2075471698113205,
+      "grad_norm": 0.6745837330818176,
+      "learning_rate": 0.00028798057204533186,
+      "loss": 3.3216,
       "step": 48300
     },
     {
-      "epoch": 5.203960822301152,
-      "grad_norm": 0.728809654712677,
-      "learning_rate": 0.00028818877276155585,
-      "loss": 3.3046,
+      "epoch": 5.212938005390836,
+      "grad_norm": 0.6576200723648071,
+      "learning_rate": 0.00028765677280086346,
+      "loss": 3.3242,
       "step": 48350
     },
     {
-      "epoch": 5.20934237434076,
-      "grad_norm": 0.6925315260887146,
-      "learning_rate": 0.0002878655317314944,
-      "loss": 3.3156,
+      "epoch": 5.218328840970351,
+      "grad_norm": 0.6276342272758484,
+      "learning_rate": 0.000287332973556395,
+      "loss": 3.3337,
       "step": 48400
     },
     {
-      "epoch": 5.214723926380368,
-      "grad_norm": 0.761146605014801,
-      "learning_rate": 0.00028754229070143304,
-      "loss": 3.3213,
+      "epoch": 5.223719676549865,
+      "grad_norm": 0.7644480466842651,
+      "learning_rate": 0.0002870091743119266,
+      "loss": 3.3271,
       "step": 48450
     },
     {
-      "epoch": 5.220105478419977,
-      "grad_norm": 0.7084577679634094,
-      "learning_rate": 0.0002872190496713716,
-      "loss": 3.3169,
+      "epoch": 5.22911051212938,
+      "grad_norm": 0.6814048886299133,
+      "learning_rate": 0.00028668537506745817,
+      "loss": 3.3137,
       "step": 48500
     },
     {
-      "epoch": 5.225487030459584,
-      "grad_norm": 0.6968151926994324,
-      "learning_rate": 0.0002868958086413102,
-      "loss": 3.307,
+      "epoch": 5.234501347708895,
+      "grad_norm": 0.6768558025360107,
+      "learning_rate": 0.0002863615758229897,
+      "loss": 3.3236,
       "step": 48550
     },
     {
-      "epoch": 5.230868582499193,
-      "grad_norm": 0.6933405995368958,
-      "learning_rate": 0.00028657256761124877,
-      "loss": 3.3176,
+      "epoch": 5.2398921832884096,
+      "grad_norm": 0.6881909966468811,
+      "learning_rate": 0.00028603777657852127,
+      "loss": 3.3468,
       "step": 48600
     },
     {
-      "epoch": 5.236250134538801,
-      "grad_norm": 0.7592040300369263,
-      "learning_rate": 0.00028624932658118736,
-      "loss": 3.3159,
+      "epoch": 5.245283018867925,
+      "grad_norm": 0.649848461151123,
+      "learning_rate": 0.0002857139773340529,
+      "loss": 3.3271,
       "step": 48650
     },
     {
-      "epoch": 5.241631686578409,
-      "grad_norm": 0.6854032874107361,
-      "learning_rate": 0.0002859260855511259,
-      "loss": 3.3186,
+      "epoch": 5.250673854447439,
+      "grad_norm": 0.6775457262992859,
+      "learning_rate": 0.0002853966540744738,
+      "loss": 3.3305,
       "step": 48700
     },
     {
-      "epoch": 5.247013238618018,
-      "grad_norm": 0.7083324193954468,
-      "learning_rate": 0.0002856028445210645,
-      "loss": 3.316,
+      "epoch": 5.256064690026954,
+      "grad_norm": 0.6716384887695312,
+      "learning_rate": 0.0002850728548300054,
+      "loss": 3.3351,
       "step": 48750
     },
     {
-      "epoch": 5.252394790657625,
-      "grad_norm": 0.7232556939125061,
-      "learning_rate": 0.0002852796034910031,
-      "loss": 3.3182,
+      "epoch": 5.261455525606469,
+      "grad_norm": 0.6535606384277344,
+      "learning_rate": 0.000284749055585537,
+      "loss": 3.3204,
       "step": 48800
     },
     {
-      "epoch": 5.257776342697234,
-      "grad_norm": 0.6980643272399902,
-      "learning_rate": 0.0002849563624609417,
-      "loss": 3.318,
+      "epoch": 5.2668463611859835,
+      "grad_norm": 0.6380178928375244,
+      "learning_rate": 0.0002844252563410685,
+      "loss": 3.317,
       "step": 48850
     },
     {
-      "epoch": 5.2631578947368425,
-      "grad_norm": 0.6944864988327026,
-      "learning_rate": 0.0002846331214308803,
-      "loss": 3.3329,
+      "epoch": 5.272237196765499,
+      "grad_norm": 0.7210144996643066,
+      "learning_rate": 0.0002841014570966001,
+      "loss": 3.3309,
       "step": 48900
     },
     {
-      "epoch": 5.26853944677645,
-      "grad_norm": 0.749697744846344,
-      "learning_rate": 0.0002843098804008188,
-      "loss": 3.34,
+      "epoch": 5.277628032345014,
+      "grad_norm": 0.6670833826065063,
+      "learning_rate": 0.00028377765785213163,
+      "loss": 3.3307,
       "step": 48950
     },
     {
-      "epoch": 5.273920998816059,
-      "grad_norm": 0.706950843334198,
-      "learning_rate": 0.00028398663937075747,
-      "loss": 3.3097,
+      "epoch": 5.283018867924528,
+      "grad_norm": 0.7309926748275757,
+      "learning_rate": 0.00028345385860766324,
+      "loss": 3.3356,
       "step": 49000
     },
     {
-      "epoch": 5.273920998816059,
-      "eval_accuracy": 0.3812864523647812,
-      "eval_loss": 3.405521869659424,
-      "eval_runtime": 181.1668,
-      "eval_samples_per_second": 99.417,
-      "eval_steps_per_second": 6.215,
+      "epoch": 5.283018867924528,
+      "eval_accuracy": 0.38106197549855647,
+      "eval_loss": 3.4104466438293457,
+      "eval_runtime": 185.3603,
+      "eval_samples_per_second": 97.168,
+      "eval_steps_per_second": 6.075,
       "step": 49000
     },
     {
-      "epoch": 5.279302550855666,
-      "grad_norm": 0.7286232709884644,
-      "learning_rate": 0.000283663398340696,
-      "loss": 3.3282,
+      "epoch": 5.288409703504043,
+      "grad_norm": 0.718997061252594,
+      "learning_rate": 0.0002831300593631948,
+      "loss": 3.3343,
       "step": 49050
     },
     {
-      "epoch": 5.284684102895275,
-      "grad_norm": 0.7539182305335999,
-      "learning_rate": 0.0002833401573106346,
-      "loss": 3.3236,
+      "epoch": 5.293800539083558,
+      "grad_norm": 0.723936140537262,
+      "learning_rate": 0.0002828062601187264,
+      "loss": 3.3476,
       "step": 49100
     },
     {
-      "epoch": 5.2900656549348835,
-      "grad_norm": 0.6944497227668762,
-      "learning_rate": 0.0002830169162805732,
-      "loss": 3.333,
+      "epoch": 5.2991913746630726,
+      "grad_norm": 0.7135783433914185,
+      "learning_rate": 0.00028248246087425794,
+      "loss": 3.3404,
       "step": 49150
     },
     {
-      "epoch": 5.295447206974491,
-      "grad_norm": 0.684943675994873,
-      "learning_rate": 0.0002826936752505118,
-      "loss": 3.3295,
+      "epoch": 5.304582210242588,
+      "grad_norm": 0.7199476361274719,
+      "learning_rate": 0.0002821586616297895,
+      "loss": 3.3478,
       "step": 49200
     },
     {
-      "epoch": 5.3008287590141,
-      "grad_norm": 0.762440025806427,
-      "learning_rate": 0.00028237043422045034,
-      "loss": 3.3233,
+      "epoch": 5.309973045822103,
+      "grad_norm": 0.6716611981391907,
+      "learning_rate": 0.0002818348623853211,
+      "loss": 3.3271,
       "step": 49250
     },
     {
-      "epoch": 5.306210311053708,
-      "grad_norm": 0.7036476135253906,
-      "learning_rate": 0.00028205365801099014,
-      "loss": 3.3298,
+      "epoch": 5.315363881401617,
+      "grad_norm": 0.6387839317321777,
+      "learning_rate": 0.00028151106314085265,
+      "loss": 3.3323,
       "step": 49300
     },
     {
-      "epoch": 5.311591863093316,
-      "grad_norm": 0.7563344836235046,
-      "learning_rate": 0.00028173041698092874,
-      "loss": 3.3374,
+      "epoch": 5.320754716981132,
+      "grad_norm": 0.6785222291946411,
+      "learning_rate": 0.0002811872638963842,
+      "loss": 3.3317,
       "step": 49350
     },
     {
-      "epoch": 5.316973415132924,
-      "grad_norm": 0.7168316841125488,
-      "learning_rate": 0.00028140717595086733,
-      "loss": 3.3358,
+      "epoch": 5.3261455525606465,
+      "grad_norm": 0.6540645956993103,
+      "learning_rate": 0.0002808634646519158,
+      "loss": 3.3245,
       "step": 49400
     },
     {
-      "epoch": 5.322354967172533,
-      "grad_norm": 0.7628238201141357,
-      "learning_rate": 0.0002810839349208059,
-      "loss": 3.3199,
+      "epoch": 5.331536388140162,
+      "grad_norm": 0.641674280166626,
+      "learning_rate": 0.00028053966540744736,
+      "loss": 3.3081,
       "step": 49450
     },
     {
-      "epoch": 5.327736519212141,
-      "grad_norm": 0.7322455644607544,
-      "learning_rate": 0.00028076069389074447,
-      "loss": 3.327,
+      "epoch": 5.336927223719677,
+      "grad_norm": 0.6906507611274719,
+      "learning_rate": 0.0002802158661629789,
+      "loss": 3.3377,
       "step": 49500
     },
     {
-      "epoch": 5.333118071251749,
-      "grad_norm": 0.6967852115631104,
-      "learning_rate": 0.0002804374528606831,
-      "loss": 3.3419,
+      "epoch": 5.342318059299191,
+      "grad_norm": 0.660807192325592,
+      "learning_rate": 0.0002798920669185105,
+      "loss": 3.335,
       "step": 49550
     },
     {
-      "epoch": 5.338499623291357,
-      "grad_norm": 0.7441530227661133,
-      "learning_rate": 0.00028011421183062166,
-      "loss": 3.3185,
+      "epoch": 5.347708894878706,
+      "grad_norm": 0.6709676384925842,
+      "learning_rate": 0.00027956826767404206,
+      "loss": 3.3269,
       "step": 49600
     },
     {
-      "epoch": 5.343881175330965,
-      "grad_norm": 0.7443904280662537,
-      "learning_rate": 0.00027979097080056025,
-      "loss": 3.3199,
+      "epoch": 5.353099730458221,
+      "grad_norm": 0.6719909310340881,
+      "learning_rate": 0.00027924446842957367,
+      "loss": 3.3308,
       "step": 49650
     },
     {
-      "epoch": 5.349262727370574,
-      "grad_norm": 0.7386976480484009,
-      "learning_rate": 0.00027946772977049885,
-      "loss": 3.318,
+      "epoch": 5.3584905660377355,
+      "grad_norm": 0.6856489777565002,
+      "learning_rate": 0.0002789206691851052,
+      "loss": 3.3159,
       "step": 49700
     },
     {
-      "epoch": 5.354644279410182,
-      "grad_norm": 0.7359290719032288,
-      "learning_rate": 0.00027914448874043744,
-      "loss": 3.3406,
+      "epoch": 5.363881401617251,
+      "grad_norm": 0.6632635593414307,
+      "learning_rate": 0.00027859686994063677,
+      "loss": 3.3536,
       "step": 49750
     },
     {
-      "epoch": 5.36002583144979,
-      "grad_norm": 0.730819046497345,
-      "learning_rate": 0.00027882124771037603,
-      "loss": 3.3092,
+      "epoch": 5.369272237196766,
+      "grad_norm": 0.6541867852210999,
+      "learning_rate": 0.0002782730706961683,
+      "loss": 3.3294,
       "step": 49800
     },
     {
-      "epoch": 5.365407383489399,
-      "grad_norm": 0.7585179209709167,
-      "learning_rate": 0.0002784980066803146,
-      "loss": 3.326,
+      "epoch": 5.37466307277628,
+      "grad_norm": 0.717612087726593,
+      "learning_rate": 0.0002779492714516999,
+      "loss": 3.3391,
       "step": 49850
     },
     {
-      "epoch": 5.370788935529006,
-      "grad_norm": 0.7162631750106812,
-      "learning_rate": 0.00027817476565025317,
-      "loss": 3.3382,
+      "epoch": 5.380053908355795,
+      "grad_norm": 0.7243929505348206,
+      "learning_rate": 0.00027762547220723147,
+      "loss": 3.3327,
       "step": 49900
     },
     {
-      "epoch": 5.376170487568615,
-      "grad_norm": 0.7122279405593872,
-      "learning_rate": 0.00027785152462019176,
-      "loss": 3.3342,
+      "epoch": 5.38544474393531,
+      "grad_norm": 0.6813120245933533,
+      "learning_rate": 0.0002773016729627631,
+      "loss": 3.3458,
       "step": 49950
     },
     {
-      "epoch": 5.3815520396082235,
-      "grad_norm": 0.7170236706733704,
-      "learning_rate": 0.00027752828359013036,
-      "loss": 3.3242,
+      "epoch": 5.390835579514825,
+      "grad_norm": 0.6332939863204956,
+      "learning_rate": 0.00027697787371829463,
+      "loss": 3.3386,
       "step": 50000
     },
     {
-      "epoch": 5.3815520396082235,
-      "eval_accuracy": 0.38197922317748634,
-      "eval_loss": 3.400938034057617,
-      "eval_runtime": 180.5748,
-      "eval_samples_per_second": 99.743,
-      "eval_steps_per_second": 6.236,
+      "epoch": 5.390835579514825,
+      "eval_accuracy": 0.38111423753759527,
+      "eval_loss": 3.4083776473999023,
+      "eval_runtime": 185.12,
+      "eval_samples_per_second": 97.294,
+      "eval_steps_per_second": 6.083,
       "step": 50000
     },
     {
-      "epoch": 5.386933591647831,
-      "grad_norm": 0.6691011786460876,
-      "learning_rate": 0.0002772050425600689,
-      "loss": 3.3201,
+      "epoch": 5.39622641509434,
+      "grad_norm": 0.6733863353729248,
+      "learning_rate": 0.00027665407447382623,
+      "loss": 3.3213,
       "step": 50050
     },
     {
-      "epoch": 5.39231514368744,
-      "grad_norm": 0.7036200761795044,
-      "learning_rate": 0.00027688180153000755,
-      "loss": 3.3223,
+      "epoch": 5.401617250673855,
+      "grad_norm": 0.740658164024353,
+      "learning_rate": 0.0002763302752293578,
+      "loss": 3.3451,
       "step": 50100
     },
     {
-      "epoch": 5.397696695727047,
-      "grad_norm": 0.6968643069267273,
-      "learning_rate": 0.0002765585604999461,
-      "loss": 3.34,
+      "epoch": 5.407008086253369,
+      "grad_norm": 0.6690346002578735,
+      "learning_rate": 0.00027600647598488933,
+      "loss": 3.3262,
       "step": 50150
     },
     {
-      "epoch": 5.403078247766656,
-      "grad_norm": 0.7583166360855103,
-      "learning_rate": 0.0002762353194698847,
-      "loss": 3.3211,
+      "epoch": 5.412398921832884,
+      "grad_norm": 0.6874278783798218,
+      "learning_rate": 0.0002756826767404209,
+      "loss": 3.3288,
       "step": 50200
     },
     {
-      "epoch": 5.4084597998062645,
-      "grad_norm": 0.7078987956047058,
-      "learning_rate": 0.0002759120784398233,
-      "loss": 3.3302,
+      "epoch": 5.4177897574123985,
+      "grad_norm": 0.7052708268165588,
+      "learning_rate": 0.0002753588774959525,
+      "loss": 3.3515,
       "step": 50250
     },
     {
-      "epoch": 5.413841351845872,
-      "grad_norm": 0.764525294303894,
-      "learning_rate": 0.00027558883740976187,
-      "loss": 3.344,
+      "epoch": 5.423180592991914,
+      "grad_norm": 0.6739440560340881,
+      "learning_rate": 0.00027503507825148404,
+      "loss": 3.3148,
       "step": 50300
     },
     {
-      "epoch": 5.419222903885481,
-      "grad_norm": 0.7484478950500488,
-      "learning_rate": 0.0002752655963797004,
-      "loss": 3.3345,
+      "epoch": 5.428571428571429,
+      "grad_norm": 0.690232515335083,
+      "learning_rate": 0.00027471127900701564,
+      "loss": 3.3306,
       "step": 50350
     },
     {
-      "epoch": 5.424604455925088,
-      "grad_norm": 0.7260310053825378,
-      "learning_rate": 0.000274942355349639,
-      "loss": 3.3208,
+      "epoch": 5.433962264150943,
+      "grad_norm": 0.7004985213279724,
+      "learning_rate": 0.0002743874797625472,
+      "loss": 3.346,
       "step": 50400
     },
     {
-      "epoch": 5.429986007964697,
-      "grad_norm": 0.6879717707633972,
-      "learning_rate": 0.0002746191143195776,
-      "loss": 3.3432,
+      "epoch": 5.439353099730458,
+      "grad_norm": 0.720639169216156,
+      "learning_rate": 0.0002740636805180788,
+      "loss": 3.3301,
       "step": 50450
     },
     {
-      "epoch": 5.435367560004305,
-      "grad_norm": 0.7341870665550232,
-      "learning_rate": 0.0002742958732895162,
-      "loss": 3.3261,
+      "epoch": 5.444743935309973,
+      "grad_norm": 0.6907075047492981,
+      "learning_rate": 0.00027373988127361035,
+      "loss": 3.3429,
       "step": 50500
     },
     {
-      "epoch": 5.440749112043913,
-      "grad_norm": 0.70168536901474,
-      "learning_rate": 0.0002739726322594548,
-      "loss": 3.3284,
+      "epoch": 5.450134770889488,
+      "grad_norm": 0.7512150406837463,
+      "learning_rate": 0.0002734160820291419,
+      "loss": 3.3379,
       "step": 50550
     },
     {
-      "epoch": 5.446130664083522,
-      "grad_norm": 0.745499312877655,
-      "learning_rate": 0.00027364939122939333,
-      "loss": 3.3374,
+      "epoch": 5.455525606469003,
+      "grad_norm": 0.6673445701599121,
+      "learning_rate": 0.00027309228278467345,
+      "loss": 3.3268,
       "step": 50600
     },
     {
-      "epoch": 5.45151221612313,
-      "grad_norm": 0.7081572413444519,
-      "learning_rate": 0.0002733261501993319,
-      "loss": 3.3199,
+      "epoch": 5.460916442048518,
+      "grad_norm": 0.6765791773796082,
+      "learning_rate": 0.00027276848354020506,
+      "loss": 3.3387,
       "step": 50650
     },
     {
-      "epoch": 5.456893768162738,
-      "grad_norm": 0.7260119915008545,
-      "learning_rate": 0.0002730029091692705,
-      "loss": 3.3515,
+      "epoch": 5.466307277628032,
+      "grad_norm": 0.7130911350250244,
+      "learning_rate": 0.0002724446842957366,
+      "loss": 3.3543,
       "step": 50700
     },
     {
-      "epoch": 5.462275320202346,
-      "grad_norm": 0.7278617024421692,
-      "learning_rate": 0.0002726796681392091,
-      "loss": 3.331,
+      "epoch": 5.471698113207547,
+      "grad_norm": 0.6755728125572205,
+      "learning_rate": 0.0002721208850512682,
+      "loss": 3.3594,
       "step": 50750
     },
     {
-      "epoch": 5.467656872241955,
-      "grad_norm": 0.7048456072807312,
-      "learning_rate": 0.0002723564271091477,
-      "loss": 3.3106,
+      "epoch": 5.4770889487870615,
+      "grad_norm": 0.7097396850585938,
+      "learning_rate": 0.00027179708580679976,
+      "loss": 3.3366,
       "step": 50800
     },
     {
-      "epoch": 5.473038424281563,
-      "grad_norm": 0.7255328893661499,
-      "learning_rate": 0.0002720331860790863,
-      "loss": 3.3247,
+      "epoch": 5.482479784366577,
+      "grad_norm": 0.6999021768569946,
+      "learning_rate": 0.0002714732865623313,
+      "loss": 3.3389,
       "step": 50850
     },
     {
-      "epoch": 5.478419976321171,
-      "grad_norm": 0.7401384115219116,
-      "learning_rate": 0.00027170994504902485,
-      "loss": 3.3402,
+      "epoch": 5.487870619946092,
+      "grad_norm": 0.6593881249427795,
+      "learning_rate": 0.0002711494873178629,
+      "loss": 3.3518,
       "step": 50900
     },
     {
-      "epoch": 5.483801528360779,
-      "grad_norm": 0.786148190498352,
-      "learning_rate": 0.00027138670401896344,
-      "loss": 3.3437,
+      "epoch": 5.493261455525606,
+      "grad_norm": 0.6644842028617859,
+      "learning_rate": 0.00027082568807339447,
+      "loss": 3.3399,
       "step": 50950
     },
     {
-      "epoch": 5.489183080400387,
-      "grad_norm": 0.7473777532577515,
-      "learning_rate": 0.00027106346298890204,
-      "loss": 3.3597,
+      "epoch": 5.498652291105121,
+      "grad_norm": 0.6714942455291748,
+      "learning_rate": 0.000270501888828926,
+      "loss": 3.3408,
       "step": 51000
     },
     {
-      "epoch": 5.489183080400387,
-      "eval_accuracy": 0.38263559527277363,
-      "eval_loss": 3.3964664936065674,
-      "eval_runtime": 180.3564,
-      "eval_samples_per_second": 99.863,
-      "eval_steps_per_second": 6.243,
+      "epoch": 5.498652291105121,
+      "eval_accuracy": 0.38187643754561384,
+      "eval_loss": 3.401001453399658,
+      "eval_runtime": 184.9621,
+      "eval_samples_per_second": 97.377,
+      "eval_steps_per_second": 6.088,
       "step": 51000
     },
     {
-      "epoch": 5.494564632439996,
-      "grad_norm": 0.7068842649459839,
-      "learning_rate": 0.00027074022195884063,
-      "loss": 3.3147,
+      "epoch": 5.504043126684636,
+      "grad_norm": 0.6829885840415955,
+      "learning_rate": 0.0002701780895844576,
+      "loss": 3.3329,
       "step": 51050
     },
     {
-      "epoch": 5.499946184479604,
-      "grad_norm": 0.7448179125785828,
-      "learning_rate": 0.0002704169809287792,
-      "loss": 3.3439,
+      "epoch": 5.509433962264151,
+      "grad_norm": 0.6931278705596924,
+      "learning_rate": 0.0002698542903399892,
+      "loss": 3.3251,
       "step": 51100
     },
     {
-      "epoch": 5.505327736519212,
-      "grad_norm": 0.7172561287879944,
-      "learning_rate": 0.00027009373989871776,
-      "loss": 3.3419,
+      "epoch": 5.514824797843666,
+      "grad_norm": 0.7629396319389343,
+      "learning_rate": 0.0002695304910955207,
+      "loss": 3.3209,
       "step": 51150
     },
     {
-      "epoch": 5.510709288558821,
-      "grad_norm": 0.7486968040466309,
-      "learning_rate": 0.00026977049886865636,
-      "loss": 3.3319,
+      "epoch": 5.520215633423181,
+      "grad_norm": 0.6978450417518616,
+      "learning_rate": 0.00026920669185105233,
+      "loss": 3.3236,
       "step": 51200
     },
     {
-      "epoch": 5.516090840598428,
-      "grad_norm": 0.7076429128646851,
-      "learning_rate": 0.00026944725783859495,
-      "loss": 3.3312,
+      "epoch": 5.525606469002695,
+      "grad_norm": 0.6813058853149414,
+      "learning_rate": 0.0002688893685914733,
+      "loss": 3.3521,
       "step": 51250
     },
     {
-      "epoch": 5.521472392638037,
-      "grad_norm": 0.7071561813354492,
-      "learning_rate": 0.00026912401680853355,
-      "loss": 3.3152,
+      "epoch": 5.53099730458221,
+      "grad_norm": 0.709402859210968,
+      "learning_rate": 0.00026856556934700483,
+      "loss": 3.3436,
       "step": 51300
     },
     {
-      "epoch": 5.5268539446776455,
-      "grad_norm": 0.7227540612220764,
-      "learning_rate": 0.0002688007757784721,
-      "loss": 3.3266,
+      "epoch": 5.536388140161725,
+      "grad_norm": 0.745469868183136,
+      "learning_rate": 0.0002682417701025364,
+      "loss": 3.3448,
       "step": 51350
     },
     {
-      "epoch": 5.532235496717253,
-      "grad_norm": 0.7778589129447937,
-      "learning_rate": 0.00026847753474841074,
-      "loss": 3.345,
+      "epoch": 5.54177897574124,
+      "grad_norm": 0.6713014841079712,
+      "learning_rate": 0.000267917970858068,
+      "loss": 3.3228,
       "step": 51400
     },
     {
-      "epoch": 5.537617048756862,
-      "grad_norm": 0.7239121794700623,
-      "learning_rate": 0.0002681542937183493,
-      "loss": 3.3364,
+      "epoch": 5.547169811320755,
+      "grad_norm": 0.6785526871681213,
+      "learning_rate": 0.00026759417161359954,
+      "loss": 3.3257,
       "step": 51450
     },
     {
-      "epoch": 5.542998600796469,
-      "grad_norm": 0.7021813988685608,
-      "learning_rate": 0.0002678310526882879,
-      "loss": 3.3266,
+      "epoch": 5.55256064690027,
+      "grad_norm": 0.7397943735122681,
+      "learning_rate": 0.0002672703723691311,
+      "loss": 3.3662,
       "step": 51500
     },
     {
-      "epoch": 5.548380152836078,
-      "grad_norm": 0.7057538628578186,
-      "learning_rate": 0.00026750781165822647,
-      "loss": 3.315,
+      "epoch": 5.557951482479784,
+      "grad_norm": 0.6826764941215515,
+      "learning_rate": 0.0002669465731246627,
+      "loss": 3.3281,
       "step": 51550
     },
     {
-      "epoch": 5.553761704875686,
-      "grad_norm": 0.7174096703529358,
-      "learning_rate": 0.00026718457062816506,
-      "loss": 3.3249,
+      "epoch": 5.563342318059299,
+      "grad_norm": 0.6693432927131653,
+      "learning_rate": 0.00026662277388019424,
+      "loss": 3.3424,
       "step": 51600
     },
     {
-      "epoch": 5.559143256915294,
-      "grad_norm": 0.7189500331878662,
-      "learning_rate": 0.00026686132959810366,
-      "loss": 3.3414,
+      "epoch": 5.568733153638814,
+      "grad_norm": 0.6833118200302124,
+      "learning_rate": 0.00026629897463572585,
+      "loss": 3.3281,
       "step": 51650
     },
     {
-      "epoch": 5.564524808954903,
-      "grad_norm": 0.7119573354721069,
-      "learning_rate": 0.0002665380885680422,
-      "loss": 3.3625,
+      "epoch": 5.574123989218329,
+      "grad_norm": 0.6914713382720947,
+      "learning_rate": 0.0002659751753912574,
+      "loss": 3.3337,
       "step": 51700
     },
     {
-      "epoch": 5.569906360994511,
-      "grad_norm": 0.7115148901939392,
-      "learning_rate": 0.0002662148475379808,
-      "loss": 3.3203,
+      "epoch": 5.579514824797844,
+      "grad_norm": 0.6685629487037659,
+      "learning_rate": 0.00026565137614678895,
+      "loss": 3.3468,
       "step": 51750
     },
     {
-      "epoch": 5.575287913034119,
-      "grad_norm": 0.7708024382591248,
-      "learning_rate": 0.0002658916065079194,
-      "loss": 3.3362,
+      "epoch": 5.584905660377358,
+      "grad_norm": 0.6789836287498474,
+      "learning_rate": 0.0002653275769023205,
+      "loss": 3.3202,
       "step": 51800
     },
     {
-      "epoch": 5.580669465073727,
-      "grad_norm": 0.7303184866905212,
-      "learning_rate": 0.000265568365477858,
-      "loss": 3.334,
+      "epoch": 5.590296495956873,
+      "grad_norm": 0.6970906853675842,
+      "learning_rate": 0.0002650037776578521,
+      "loss": 3.3166,
       "step": 51850
     },
     {
-      "epoch": 5.586051017113336,
-      "grad_norm": 0.7622580528259277,
-      "learning_rate": 0.0002652451244477965,
-      "loss": 3.32,
+      "epoch": 5.595687331536388,
+      "grad_norm": 0.6891434788703918,
+      "learning_rate": 0.00026467997841338366,
+      "loss": 3.3685,
       "step": 51900
     },
     {
-      "epoch": 5.591432569152944,
-      "grad_norm": 0.7121456265449524,
-      "learning_rate": 0.00026492188341773517,
-      "loss": 3.316,
+      "epoch": 5.601078167115903,
+      "grad_norm": 0.7000083923339844,
+      "learning_rate": 0.00026435617916891526,
+      "loss": 3.3327,
       "step": 51950
     },
     {
-      "epoch": 5.596814121192552,
-      "grad_norm": 0.7205502390861511,
-      "learning_rate": 0.0002645986423876737,
-      "loss": 3.3013,
+      "epoch": 5.606469002695418,
+      "grad_norm": 0.6774043440818787,
+      "learning_rate": 0.0002640323799244468,
+      "loss": 3.3484,
       "step": 52000
     },
     {
-      "epoch": 5.596814121192552,
-      "eval_accuracy": 0.38279140351389956,
-      "eval_loss": 3.3938612937927246,
-      "eval_runtime": 180.5356,
-      "eval_samples_per_second": 99.764,
-      "eval_steps_per_second": 6.237,
+      "epoch": 5.606469002695418,
+      "eval_accuracy": 0.3823833032672684,
+      "eval_loss": 3.3957502841949463,
+      "eval_runtime": 184.6922,
+      "eval_samples_per_second": 97.519,
+      "eval_steps_per_second": 6.097,
       "step": 52000
     },
     {
-      "epoch": 5.60219567323216,
-      "grad_norm": 0.7592937350273132,
-      "learning_rate": 0.0002642754013576123,
-      "loss": 3.3312,
+      "epoch": 5.611859838274933,
+      "grad_norm": 0.6880715489387512,
+      "learning_rate": 0.0002637085806799784,
+      "loss": 3.3226,
       "step": 52050
     },
     {
-      "epoch": 5.607577225271768,
-      "grad_norm": 0.7347540855407715,
-      "learning_rate": 0.0002639521603275509,
-      "loss": 3.3274,
+      "epoch": 5.617250673854447,
+      "grad_norm": 0.6606606245040894,
+      "learning_rate": 0.00026338478143550997,
+      "loss": 3.3464,
       "step": 52100
     },
     {
-      "epoch": 5.612958777311377,
-      "grad_norm": 0.717491626739502,
-      "learning_rate": 0.0002636289192974895,
-      "loss": 3.3496,
+      "epoch": 5.622641509433962,
+      "grad_norm": 0.6744948029518127,
+      "learning_rate": 0.00026306098219104157,
+      "loss": 3.3165,
       "step": 52150
     },
     {
-      "epoch": 5.618340329350985,
-      "grad_norm": 0.7288353443145752,
-      "learning_rate": 0.00026330567826742804,
-      "loss": 3.3103,
+      "epoch": 5.628032345013477,
+      "grad_norm": 0.7021511793136597,
+      "learning_rate": 0.00026273718294657307,
+      "loss": 3.3176,
       "step": 52200
     },
     {
-      "epoch": 5.623721881390593,
-      "grad_norm": 0.6995804309844971,
-      "learning_rate": 0.00026298243723736663,
-      "loss": 3.3326,
+      "epoch": 5.633423180592992,
+      "grad_norm": 0.7162783741950989,
+      "learning_rate": 0.00026241338370210467,
+      "loss": 3.337,
       "step": 52250
     },
     {
-      "epoch": 5.629103433430201,
-      "grad_norm": 0.7089316844940186,
-      "learning_rate": 0.0002626591962073052,
-      "loss": 3.3313,
+      "epoch": 5.638814016172507,
+      "grad_norm": 0.6945198774337769,
+      "learning_rate": 0.0002620895844576362,
+      "loss": 3.3495,
       "step": 52300
     },
     {
-      "epoch": 5.634484985469809,
-      "grad_norm": 0.7350379824638367,
-      "learning_rate": 0.0002623359551772438,
-      "loss": 3.3358,
+      "epoch": 5.644204851752022,
+      "grad_norm": 0.6781046390533447,
+      "learning_rate": 0.00026176578521316783,
+      "loss": 3.3215,
       "step": 52350
     },
     {
-      "epoch": 5.639866537509418,
-      "grad_norm": 0.7242736220359802,
-      "learning_rate": 0.0002620127141471824,
-      "loss": 3.3154,
+      "epoch": 5.649595687331536,
+      "grad_norm": 0.6661491394042969,
+      "learning_rate": 0.0002614419859686994,
+      "loss": 3.318,
       "step": 52400
     },
     {
-      "epoch": 5.645248089549026,
-      "grad_norm": 0.7510601282119751,
-      "learning_rate": 0.00026168947311712095,
-      "loss": 3.3391,
+      "epoch": 5.654986522911051,
+      "grad_norm": 0.8015744090080261,
+      "learning_rate": 0.000261118186724231,
+      "loss": 3.327,
       "step": 52450
     },
     {
-      "epoch": 5.650629641588634,
-      "grad_norm": 0.7219720482826233,
-      "learning_rate": 0.00026136623208705955,
-      "loss": 3.3203,
+      "epoch": 5.660377358490566,
+      "grad_norm": 0.7487977147102356,
+      "learning_rate": 0.00026079438747976253,
+      "loss": 3.3416,
       "step": 52500
     },
     {
-      "epoch": 5.656011193628243,
-      "grad_norm": 0.7292102575302124,
-      "learning_rate": 0.00026104299105699814,
-      "loss": 3.3382,
+      "epoch": 5.665768194070081,
+      "grad_norm": 0.674368679523468,
+      "learning_rate": 0.0002604705882352941,
+      "loss": 3.3369,
       "step": 52550
     },
     {
-      "epoch": 5.66139274566785,
-      "grad_norm": 0.7652955651283264,
-      "learning_rate": 0.00026071975002693674,
-      "loss": 3.3266,
+      "epoch": 5.671159029649596,
+      "grad_norm": 0.750124454498291,
+      "learning_rate": 0.00026014678899082563,
+      "loss": 3.3364,
       "step": 52600
     },
     {
-      "epoch": 5.666774297707459,
-      "grad_norm": 0.7515937685966492,
-      "learning_rate": 0.00026039650899687533,
-      "loss": 3.3376,
+      "epoch": 5.67654986522911,
+      "grad_norm": 0.6911941170692444,
+      "learning_rate": 0.00025982298974635724,
+      "loss": 3.3505,
       "step": 52650
     },
     {
-      "epoch": 5.672155849747067,
-      "grad_norm": 0.7084229588508606,
-      "learning_rate": 0.00026007326796681393,
-      "loss": 3.3164,
+      "epoch": 5.681940700808625,
+      "grad_norm": 0.7061330080032349,
+      "learning_rate": 0.0002594991905018888,
+      "loss": 3.3523,
       "step": 52700
     },
     {
-      "epoch": 5.677537401786675,
-      "grad_norm": 0.717211127281189,
-      "learning_rate": 0.00025975002693675247,
-      "loss": 3.3411,
+      "epoch": 5.6873315363881405,
+      "grad_norm": 0.7390830516815186,
+      "learning_rate": 0.0002591753912574204,
+      "loss": 3.3449,
       "step": 52750
     },
     {
-      "epoch": 5.682918953826284,
-      "grad_norm": 0.7143423557281494,
-      "learning_rate": 0.00025942678590669106,
-      "loss": 3.3249,
+      "epoch": 5.692722371967655,
+      "grad_norm": 0.7006047368049622,
+      "learning_rate": 0.00025885159201295195,
+      "loss": 3.3225,
       "step": 52800
     },
     {
-      "epoch": 5.688300505865891,
-      "grad_norm": 0.7390002012252808,
-      "learning_rate": 0.00025910354487662966,
-      "loss": 3.3317,
+      "epoch": 5.69811320754717,
+      "grad_norm": 0.7471202611923218,
+      "learning_rate": 0.0002585277927684835,
+      "loss": 3.3285,
       "step": 52850
     },
     {
-      "epoch": 5.6936820579055,
-      "grad_norm": 0.7596791386604309,
-      "learning_rate": 0.00025878676866716946,
-      "loss": 3.3377,
+      "epoch": 5.703504043126685,
+      "grad_norm": 0.6955723762512207,
+      "learning_rate": 0.0002582039935240151,
+      "loss": 3.3498,
       "step": 52900
     },
     {
-      "epoch": 5.699063609945108,
-      "grad_norm": 0.7708361744880676,
-      "learning_rate": 0.00025846352763710806,
-      "loss": 3.3309,
+      "epoch": 5.708894878706199,
+      "grad_norm": 0.6657953262329102,
+      "learning_rate": 0.00025788019427954665,
+      "loss": 3.3469,
       "step": 52950
     },
     {
-      "epoch": 5.704445161984716,
-      "grad_norm": 0.7448739409446716,
-      "learning_rate": 0.0002581402866070466,
-      "loss": 3.3461,
+      "epoch": 5.714285714285714,
+      "grad_norm": 0.6830325126647949,
+      "learning_rate": 0.00025755639503507826,
+      "loss": 3.339,
       "step": 53000
     },
     {
-      "epoch": 5.704445161984716,
-      "eval_accuracy": 0.382959815490012,
-      "eval_loss": 3.3894808292388916,
-      "eval_runtime": 180.4953,
-      "eval_samples_per_second": 99.787,
-      "eval_steps_per_second": 6.238,
+      "epoch": 5.714285714285714,
+      "eval_accuracy": 0.38255432291268837,
+      "eval_loss": 3.392019748687744,
+      "eval_runtime": 185.0125,
+      "eval_samples_per_second": 97.35,
+      "eval_steps_per_second": 6.086,
       "step": 53000
     },
     {
-      "epoch": 5.709826714024325,
-      "grad_norm": 0.7527272701263428,
-      "learning_rate": 0.00025781704557698525,
-      "loss": 3.3118,
+      "epoch": 5.719676549865229,
+      "grad_norm": 0.6657763123512268,
+      "learning_rate": 0.0002572325957906098,
+      "loss": 3.3222,
       "step": 53050
     },
     {
-      "epoch": 5.715208266063933,
-      "grad_norm": 0.7586866021156311,
-      "learning_rate": 0.0002574938045469238,
-      "loss": 3.323,
+      "epoch": 5.725067385444744,
+      "grad_norm": 0.7333123683929443,
+      "learning_rate": 0.00025690879654614136,
+      "loss": 3.3471,
       "step": 53100
     },
     {
-      "epoch": 5.720589818103541,
-      "grad_norm": 0.7504985928535461,
-      "learning_rate": 0.0002571705635168624,
-      "loss": 3.3293,
+      "epoch": 5.730458221024259,
+      "grad_norm": 0.7092816829681396,
+      "learning_rate": 0.0002565849973016729,
+      "loss": 3.3468,
       "step": 53150
     },
     {
-      "epoch": 5.725971370143149,
-      "grad_norm": 0.7618194222450256,
-      "learning_rate": 0.000256847322486801,
-      "loss": 3.344,
+      "epoch": 5.735849056603773,
+      "grad_norm": 0.7320940494537354,
+      "learning_rate": 0.0002562611980572045,
+      "loss": 3.3424,
       "step": 53200
     },
     {
-      "epoch": 5.731352922182758,
-      "grad_norm": 0.7982940673828125,
-      "learning_rate": 0.0002565240814567396,
-      "loss": 3.3382,
+      "epoch": 5.741239892183288,
+      "grad_norm": 0.7620314359664917,
+      "learning_rate": 0.00025593739881273606,
+      "loss": 3.3349,
       "step": 53250
     },
     {
-      "epoch": 5.736734474222366,
-      "grad_norm": 0.7086600661277771,
-      "learning_rate": 0.0002562008404266781,
-      "loss": 3.3341,
+      "epoch": 5.7466307277628035,
+      "grad_norm": 0.6692413091659546,
+      "learning_rate": 0.00025561359956826767,
+      "loss": 3.3568,
       "step": 53300
     },
     {
-      "epoch": 5.742116026261974,
-      "grad_norm": 0.71695476770401,
-      "learning_rate": 0.0002558775993966167,
-      "loss": 3.3213,
+      "epoch": 5.752021563342318,
+      "grad_norm": 0.6936851739883423,
+      "learning_rate": 0.0002552898003237992,
+      "loss": 3.3201,
       "step": 53350
     },
     {
-      "epoch": 5.747497578301582,
-      "grad_norm": 0.7411425709724426,
-      "learning_rate": 0.0002555543583665553,
-      "loss": 3.3474,
+      "epoch": 5.757412398921833,
+      "grad_norm": 0.6667423844337463,
+      "learning_rate": 0.0002549660010793308,
+      "loss": 3.3396,
       "step": 53400
     },
     {
-      "epoch": 5.75287913034119,
-      "grad_norm": 0.7171765565872192,
-      "learning_rate": 0.0002552311173364939,
-      "loss": 3.3428,
+      "epoch": 5.762803234501348,
+      "grad_norm": 0.7328035831451416,
+      "learning_rate": 0.0002546422018348624,
+      "loss": 3.329,
       "step": 53450
     },
     {
-      "epoch": 5.758260682380799,
-      "grad_norm": 0.7351555228233337,
-      "learning_rate": 0.0002549078763064325,
-      "loss": 3.3225,
+      "epoch": 5.768194070080862,
+      "grad_norm": 0.6744652390480042,
+      "learning_rate": 0.0002543184025903939,
+      "loss": 3.3415,
       "step": 53500
     },
     {
-      "epoch": 5.763642234420407,
-      "grad_norm": 0.7267298102378845,
-      "learning_rate": 0.00025458463527637103,
-      "loss": 3.3269,
+      "epoch": 5.773584905660377,
+      "grad_norm": 0.6693131327629089,
+      "learning_rate": 0.0002539946033459255,
+      "loss": 3.3265,
       "step": 53550
     },
     {
-      "epoch": 5.769023786460015,
-      "grad_norm": 0.6977094411849976,
-      "learning_rate": 0.0002542613942463097,
-      "loss": 3.3547,
+      "epoch": 5.7789757412398925,
+      "grad_norm": 0.7442718148231506,
+      "learning_rate": 0.0002536708041014571,
+      "loss": 3.322,
       "step": 53600
     },
     {
-      "epoch": 5.774405338499624,
-      "grad_norm": 0.7931075096130371,
-      "learning_rate": 0.0002539381532162482,
-      "loss": 3.3261,
+      "epoch": 5.784366576819407,
+      "grad_norm": 0.6674184203147888,
+      "learning_rate": 0.00025334700485698863,
+      "loss": 3.3204,
       "step": 53650
     },
     {
-      "epoch": 5.779786890539231,
-      "grad_norm": 0.7646996378898621,
-      "learning_rate": 0.0002536149121861868,
-      "loss": 3.3383,
+      "epoch": 5.789757412398922,
+      "grad_norm": 0.712354838848114,
+      "learning_rate": 0.00025302320561252023,
+      "loss": 3.3311,
       "step": 53700
     },
     {
-      "epoch": 5.78516844257884,
-      "grad_norm": 0.7829656004905701,
-      "learning_rate": 0.0002532916711561254,
-      "loss": 3.3332,
+      "epoch": 5.795148247978437,
+      "grad_norm": 0.710060179233551,
+      "learning_rate": 0.0002526994063680518,
+      "loss": 3.3383,
       "step": 53750
     },
     {
-      "epoch": 5.790549994618448,
-      "grad_norm": 0.7650371789932251,
-      "learning_rate": 0.000252968430126064,
-      "loss": 3.3414,
+      "epoch": 5.800539083557951,
+      "grad_norm": 0.7189279198646545,
+      "learning_rate": 0.0002523756071235834,
+      "loss": 3.3457,
       "step": 53800
     },
     {
-      "epoch": 5.795931546658056,
-      "grad_norm": 0.7446619272232056,
-      "learning_rate": 0.00025264518909600255,
-      "loss": 3.3277,
+      "epoch": 5.8059299191374665,
+      "grad_norm": 0.6352289915084839,
+      "learning_rate": 0.00025205180787911494,
+      "loss": 3.3283,
       "step": 53850
     },
     {
-      "epoch": 5.801313098697665,
-      "grad_norm": 0.7219211459159851,
-      "learning_rate": 0.00025232194806594114,
-      "loss": 3.3408,
+      "epoch": 5.811320754716981,
+      "grad_norm": 0.7231967449188232,
+      "learning_rate": 0.0002517280086346465,
+      "loss": 3.3339,
       "step": 53900
     },
     {
-      "epoch": 5.806694650737272,
-      "grad_norm": 0.7577422857284546,
-      "learning_rate": 0.00025199870703587974,
-      "loss": 3.345,
+      "epoch": 5.816711590296496,
+      "grad_norm": 0.6729211807250977,
+      "learning_rate": 0.00025140420939017804,
+      "loss": 3.3329,
       "step": 53950
     },
     {
-      "epoch": 5.812076202776881,
-      "grad_norm": 0.7506514191627502,
-      "learning_rate": 0.00025167546600581833,
-      "loss": 3.3376,
+      "epoch": 5.822102425876011,
+      "grad_norm": 0.6906619071960449,
+      "learning_rate": 0.00025108041014570965,
+      "loss": 3.3305,
       "step": 54000
     },
     {
-      "epoch": 5.812076202776881,
-      "eval_accuracy": 0.3836133626102161,
-      "eval_loss": 3.3827064037323,
-      "eval_runtime": 180.6271,
-      "eval_samples_per_second": 99.714,
-      "eval_steps_per_second": 6.234,
+      "epoch": 5.822102425876011,
+      "eval_accuracy": 0.3831925499757107,
+      "eval_loss": 3.38727068901062,
+      "eval_runtime": 184.8974,
+      "eval_samples_per_second": 97.411,
+      "eval_steps_per_second": 6.09,
       "step": 54000
     },
     {
-      "epoch": 5.817457754816489,
-      "grad_norm": 0.7082645297050476,
-      "learning_rate": 0.0002513522249757569,
-      "loss": 3.3294,
+      "epoch": 5.827493261455525,
+      "grad_norm": 0.706639289855957,
+      "learning_rate": 0.0002507566109012412,
+      "loss": 3.3496,
       "step": 54050
     },
     {
-      "epoch": 5.822839306856097,
-      "grad_norm": 0.8100736141204834,
-      "learning_rate": 0.00025102898394569547,
-      "loss": 3.3329,
+      "epoch": 5.83288409703504,
+      "grad_norm": 0.7261019945144653,
+      "learning_rate": 0.0002504328116567728,
+      "loss": 3.3316,
       "step": 54100
     },
     {
-      "epoch": 5.828220858895706,
-      "grad_norm": 0.7404306530952454,
-      "learning_rate": 0.00025070574291563406,
-      "loss": 3.3335,
+      "epoch": 5.8382749326145555,
+      "grad_norm": 0.681576669216156,
+      "learning_rate": 0.00025010901241230435,
+      "loss": 3.3525,
       "step": 54150
     },
     {
-      "epoch": 5.833602410935313,
-      "grad_norm": 0.7143645286560059,
-      "learning_rate": 0.00025038250188557265,
-      "loss": 3.3362,
+      "epoch": 5.84366576819407,
+      "grad_norm": 0.6800199151039124,
+      "learning_rate": 0.0002497852131678359,
+      "loss": 3.3567,
       "step": 54200
     },
     {
-      "epoch": 5.838983962974922,
-      "grad_norm": 0.7716860175132751,
-      "learning_rate": 0.00025005926085551125,
-      "loss": 3.3455,
+      "epoch": 5.849056603773585,
+      "grad_norm": 0.6690753698348999,
+      "learning_rate": 0.0002494614139233675,
+      "loss": 3.3331,
       "step": 54250
     },
     {
-      "epoch": 5.84436551501453,
-      "grad_norm": 0.769177258014679,
-      "learning_rate": 0.0002497360198254498,
-      "loss": 3.3358,
+      "epoch": 5.8544474393531,
+      "grad_norm": 0.7050347924232483,
+      "learning_rate": 0.00024913761467889906,
+      "loss": 3.3425,
       "step": 54300
     },
     {
-      "epoch": 5.849747067054138,
-      "grad_norm": 0.7687222957611084,
-      "learning_rate": 0.00024941277879538844,
-      "loss": 3.3366,
+      "epoch": 5.859838274932614,
+      "grad_norm": 0.7527475357055664,
+      "learning_rate": 0.0002488138154344306,
+      "loss": 3.3453,
       "step": 54350
     },
     {
-      "epoch": 5.855128619093747,
-      "grad_norm": 0.7368986010551453,
-      "learning_rate": 0.000249089537765327,
-      "loss": 3.332,
+      "epoch": 5.8652291105121295,
+      "grad_norm": 0.6619974970817566,
+      "learning_rate": 0.0002484900161899622,
+      "loss": 3.3404,
       "step": 54400
     },
     {
-      "epoch": 5.860510171133355,
-      "grad_norm": 0.7428856492042542,
-      "learning_rate": 0.0002487662967352656,
-      "loss": 3.3448,
+      "epoch": 5.870619946091644,
+      "grad_norm": 0.7044657468795776,
+      "learning_rate": 0.00024816621694549376,
+      "loss": 3.3278,
       "step": 54450
     },
     {
-      "epoch": 5.865891723172963,
-      "grad_norm": 0.7492393255233765,
-      "learning_rate": 0.00024844305570520417,
-      "loss": 3.3396,
+      "epoch": 5.876010781671159,
+      "grad_norm": 0.72688227891922,
+      "learning_rate": 0.00024784241770102537,
+      "loss": 3.3391,
       "step": 54500
     },
     {
-      "epoch": 5.871273275212571,
-      "grad_norm": 0.8640391826629639,
-      "learning_rate": 0.00024811981467514276,
-      "loss": 3.3256,
+      "epoch": 5.881401617250674,
+      "grad_norm": 0.7019941806793213,
+      "learning_rate": 0.0002475186184565569,
+      "loss": 3.3388,
       "step": 54550
     },
     {
-      "epoch": 5.87665482725218,
-      "grad_norm": 0.7381857633590698,
-      "learning_rate": 0.00024779657364508136,
-      "loss": 3.3202,
+      "epoch": 5.886792452830189,
+      "grad_norm": 0.7095934748649597,
+      "learning_rate": 0.00024719481921208847,
+      "loss": 3.3388,
       "step": 54600
     },
     {
-      "epoch": 5.882036379291788,
-      "grad_norm": 0.7371327877044678,
-      "learning_rate": 0.0002474733326150199,
-      "loss": 3.3164,
+      "epoch": 5.892183288409703,
+      "grad_norm": 0.7058985829353333,
+      "learning_rate": 0.0002468710199676201,
+      "loss": 3.3453,
       "step": 54650
     },
     {
-      "epoch": 5.887417931331396,
-      "grad_norm": 0.7317863702774048,
-      "learning_rate": 0.0002471500915849585,
-      "loss": 3.314,
+      "epoch": 5.8975741239892185,
+      "grad_norm": 0.7246441841125488,
+      "learning_rate": 0.0002465472207231516,
+      "loss": 3.3331,
       "step": 54700
     },
     {
-      "epoch": 5.892799483371004,
-      "grad_norm": 0.7320008277893066,
-      "learning_rate": 0.0002468268505548971,
-      "loss": 3.3218,
+      "epoch": 5.902964959568733,
+      "grad_norm": 0.7580798268318176,
+      "learning_rate": 0.0002462234214786832,
+      "loss": 3.3339,
       "step": 54750
     },
     {
-      "epoch": 5.898181035410612,
-      "grad_norm": 0.7335460186004639,
-      "learning_rate": 0.0002465036095248357,
-      "loss": 3.339,
+      "epoch": 5.908355795148248,
+      "grad_norm": 0.7288414835929871,
+      "learning_rate": 0.0002458996222342148,
+      "loss": 3.3325,
       "step": 54800
     },
     {
-      "epoch": 5.903562587450221,
-      "grad_norm": 0.7588884234428406,
-      "learning_rate": 0.0002461803684947742,
-      "loss": 3.3418,
+      "epoch": 5.913746630727763,
+      "grad_norm": 0.6888887882232666,
+      "learning_rate": 0.00024557582298974633,
+      "loss": 3.3351,
       "step": 54850
     },
     {
-      "epoch": 5.9089441394898286,
-      "grad_norm": 0.7557958364486694,
-      "learning_rate": 0.0002458571274647128,
-      "loss": 3.3497,
+      "epoch": 5.919137466307277,
+      "grad_norm": 0.7192179560661316,
+      "learning_rate": 0.0002452520237452779,
+      "loss": 3.3365,
       "step": 54900
     },
     {
-      "epoch": 5.914325691529437,
-      "grad_norm": 0.7636759281158447,
-      "learning_rate": 0.0002455338864346514,
-      "loss": 3.3302,
+      "epoch": 5.9245283018867925,
+      "grad_norm": 0.7986792325973511,
+      "learning_rate": 0.0002449282245008095,
+      "loss": 3.3433,
       "step": 54950
     },
     {
-      "epoch": 5.919707243569046,
-      "grad_norm": 0.7415563464164734,
-      "learning_rate": 0.00024521064540459,
-      "loss": 3.3454,
+      "epoch": 5.929919137466308,
+      "grad_norm": 0.6919881105422974,
+      "learning_rate": 0.00024460442525634104,
+      "loss": 3.3495,
       "step": 55000
     },
     {
-      "epoch": 5.919707243569046,
-      "eval_accuracy": 0.38400386108902135,
-      "eval_loss": 3.3798177242279053,
-      "eval_runtime": 180.4085,
-      "eval_samples_per_second": 99.835,
-      "eval_steps_per_second": 6.241,
+      "epoch": 5.929919137466308,
+      "eval_accuracy": 0.38388206120178137,
+      "eval_loss": 3.383330821990967,
+      "eval_runtime": 185.021,
+      "eval_samples_per_second": 97.346,
+      "eval_steps_per_second": 6.086,
       "step": 55000
     },
     {
-      "epoch": 5.925088795608653,
-      "grad_norm": 0.7452832460403442,
-      "learning_rate": 0.0002448874043745286,
-      "loss": 3.3509,
+      "epoch": 5.935309973045822,
+      "grad_norm": 0.7238466739654541,
+      "learning_rate": 0.00024428062601187264,
+      "loss": 3.3563,
       "step": 55050
     },
     {
-      "epoch": 5.930470347648262,
-      "grad_norm": 0.8210157752037048,
-      "learning_rate": 0.00024456416334446714,
-      "loss": 3.3223,
+      "epoch": 5.940700808625337,
+      "grad_norm": 0.675534188747406,
+      "learning_rate": 0.0002439568267674042,
+      "loss": 3.3441,
       "step": 55100
     },
     {
-      "epoch": 5.93585189968787,
-      "grad_norm": 0.734848141670227,
-      "learning_rate": 0.000244247387135007,
-      "loss": 3.3249,
+      "epoch": 5.946091644204852,
+      "grad_norm": 0.7394225001335144,
+      "learning_rate": 0.00024363302752293574,
+      "loss": 3.3394,
       "step": 55150
     },
     {
-      "epoch": 5.941233451727478,
-      "grad_norm": 0.8083829283714294,
-      "learning_rate": 0.00024392414610494557,
-      "loss": 3.329,
+      "epoch": 5.951482479784366,
+      "grad_norm": 0.7124001383781433,
+      "learning_rate": 0.00024330922827846732,
+      "loss": 3.3482,
       "step": 55200
     },
     {
-      "epoch": 5.946615003767087,
-      "grad_norm": 0.7373672723770142,
-      "learning_rate": 0.00024360090507488414,
-      "loss": 3.339,
+      "epoch": 5.9568733153638815,
+      "grad_norm": 0.7209517359733582,
+      "learning_rate": 0.00024299190501888827,
+      "loss": 3.3532,
       "step": 55250
     },
     {
-      "epoch": 5.951996555806694,
-      "grad_norm": 0.7348290681838989,
-      "learning_rate": 0.00024327766404482273,
-      "loss": 3.3362,
+      "epoch": 5.962264150943396,
+      "grad_norm": 0.6774404644966125,
+      "learning_rate": 0.00024266810577441985,
+      "loss": 3.3182,
       "step": 55300
     },
     {
-      "epoch": 5.957378107846303,
-      "grad_norm": 0.7888813018798828,
-      "learning_rate": 0.0002429544230147613,
-      "loss": 3.3275,
+      "epoch": 5.967654986522911,
+      "grad_norm": 0.7080289125442505,
+      "learning_rate": 0.00024234430652995143,
+      "loss": 3.3309,
       "step": 55350
     },
     {
-      "epoch": 5.962759659885911,
-      "grad_norm": 0.7368828058242798,
-      "learning_rate": 0.00024263118198469992,
-      "loss": 3.3428,
+      "epoch": 5.973045822102426,
+      "grad_norm": 0.6646098494529724,
+      "learning_rate": 0.00024202050728548298,
+      "loss": 3.3311,
       "step": 55400
     },
     {
-      "epoch": 5.968141211925519,
-      "grad_norm": 0.726843535900116,
-      "learning_rate": 0.0002423079409546385,
-      "loss": 3.357,
+      "epoch": 5.97843665768194,
+      "grad_norm": 0.7524878978729248,
+      "learning_rate": 0.00024169670804101456,
+      "loss": 3.3239,
       "step": 55450
     },
     {
-      "epoch": 5.973522763965128,
-      "grad_norm": 0.7297669053077698,
-      "learning_rate": 0.00024198469992457706,
-      "loss": 3.3171,
+      "epoch": 5.9838274932614555,
+      "grad_norm": 0.7035167813301086,
+      "learning_rate": 0.0002413729087965461,
+      "loss": 3.3441,
       "step": 55500
     },
     {
-      "epoch": 5.978904316004736,
-      "grad_norm": 0.7472332119941711,
-      "learning_rate": 0.00024166145889451568,
-      "loss": 3.3452,
+      "epoch": 5.989218328840971,
+      "grad_norm": 0.7106602191925049,
+      "learning_rate": 0.00024104910955207768,
+      "loss": 3.3366,
       "step": 55550
     },
     {
-      "epoch": 5.984285868044344,
-      "grad_norm": 0.7742068767547607,
-      "learning_rate": 0.00024133821786445425,
-      "loss": 3.35,
+      "epoch": 5.994609164420485,
+      "grad_norm": 0.6742231249809265,
+      "learning_rate": 0.00024072531030760926,
+      "loss": 3.3182,
       "step": 55600
     },
     {
-      "epoch": 5.989667420083952,
-      "grad_norm": 0.7965028882026672,
-      "learning_rate": 0.0002410149768343928,
-      "loss": 3.3454,
+      "epoch": 6.0,
+      "grad_norm": Infinity,
+      "learning_rate": 0.00024040798704803021,
+      "loss": 3.3324,
       "step": 55650
     },
     {
-      "epoch": 5.995048972123561,
-      "grad_norm": 0.7538635730743408,
-      "learning_rate": 0.0002406917358043314,
-      "loss": 3.3523,
+      "epoch": 6.005390835579515,
+      "grad_norm": 0.730053186416626,
+      "learning_rate": 0.00024008418780356176,
+      "loss": 3.2365,
       "step": 55700
     },
     {
-      "epoch": 6.000430524163169,
-      "grad_norm": 0.7462757229804993,
-      "learning_rate": 0.00024036849477427,
-      "loss": 3.3303,
+      "epoch": 6.010781671159029,
+      "grad_norm": 0.6938073635101318,
+      "learning_rate": 0.00023976038855909334,
+      "loss": 3.2432,
       "step": 55750
     },
     {
-      "epoch": 6.005812076202777,
-      "grad_norm": 0.7339770793914795,
-      "learning_rate": 0.00024004525374420857,
-      "loss": 3.2492,
+      "epoch": 6.0161725067385445,
+      "grad_norm": 0.6961255669593811,
+      "learning_rate": 0.00023943658931462492,
+      "loss": 3.256,
       "step": 55800
     },
     {
-      "epoch": 6.011193628242385,
-      "grad_norm": 0.7494232654571533,
-      "learning_rate": 0.00023972201271414716,
-      "loss": 3.2566,
+      "epoch": 6.02156334231806,
+      "grad_norm": 0.7340931296348572,
+      "learning_rate": 0.00023911279007015647,
+      "loss": 3.2447,
       "step": 55850
     },
     {
-      "epoch": 6.016575180281993,
-      "grad_norm": 0.7096470594406128,
-      "learning_rate": 0.00023939877168408573,
-      "loss": 3.2373,
+      "epoch": 6.026954177897574,
+      "grad_norm": 0.7116497159004211,
+      "learning_rate": 0.00023878899082568805,
+      "loss": 3.2531,
       "step": 55900
     },
     {
-      "epoch": 6.021956732321602,
-      "grad_norm": 0.7256503701210022,
-      "learning_rate": 0.00023907553065402433,
-      "loss": 3.2391,
+      "epoch": 6.032345013477089,
+      "grad_norm": 0.6797800064086914,
+      "learning_rate": 0.00023846519158121963,
+      "loss": 3.2467,
       "step": 55950
     },
     {
-      "epoch": 6.0273382843612096,
-      "grad_norm": 0.7260343432426453,
-      "learning_rate": 0.00023875228962396292,
-      "loss": 3.2501,
+      "epoch": 6.037735849056604,
+      "grad_norm": 0.7329632043838501,
+      "learning_rate": 0.0002381413923367512,
+      "loss": 3.2488,
       "step": 56000
     },
     {
-      "epoch": 6.0273382843612096,
-      "eval_accuracy": 0.3840860026722091,
-      "eval_loss": 3.3830971717834473,
-      "eval_runtime": 180.8499,
-      "eval_samples_per_second": 99.591,
-      "eval_steps_per_second": 6.226,
+      "epoch": 6.037735849056604,
+      "eval_accuracy": 0.3840788315816134,
+      "eval_loss": 3.386367082595825,
+      "eval_runtime": 184.7957,
+      "eval_samples_per_second": 97.464,
+      "eval_steps_per_second": 6.093,
       "step": 56000
     },
     {
-      "epoch": 6.032719836400818,
-      "grad_norm": 0.8156185746192932,
-      "learning_rate": 0.0002384290485939015,
-      "loss": 3.2535,
+      "epoch": 6.0431266846361185,
+      "grad_norm": 0.7919038534164429,
+      "learning_rate": 0.00023781759309228275,
+      "loss": 3.2497,
       "step": 56050
     },
     {
-      "epoch": 6.038101388440427,
-      "grad_norm": 0.7497657537460327,
-      "learning_rate": 0.00023810580756384006,
-      "loss": 3.2479,
+      "epoch": 6.048517520215634,
+      "grad_norm": 0.6978245377540588,
+      "learning_rate": 0.00023749379384781433,
+      "loss": 3.268,
       "step": 56100
     },
     {
-      "epoch": 6.043482940480034,
-      "grad_norm": 0.7156153321266174,
-      "learning_rate": 0.00023778256653377868,
-      "loss": 3.2614,
+      "epoch": 6.053908355795148,
+      "grad_norm": 0.670793354511261,
+      "learning_rate": 0.0002371699946033459,
+      "loss": 3.2439,
       "step": 56150
     },
     {
-      "epoch": 6.048864492519643,
-      "grad_norm": 0.7490081787109375,
-      "learning_rate": 0.00023745932550371725,
-      "loss": 3.2544,
+      "epoch": 6.059299191374663,
+      "grad_norm": 0.6458380222320557,
+      "learning_rate": 0.0002368461953588775,
+      "loss": 3.2512,
       "step": 56200
     },
     {
-      "epoch": 6.0542460445592505,
-      "grad_norm": 0.7716811299324036,
-      "learning_rate": 0.00023713608447365584,
-      "loss": 3.2517,
+      "epoch": 6.064690026954178,
+      "grad_norm": 0.6844601631164551,
+      "learning_rate": 0.00023652239611440904,
+      "loss": 3.2673,
       "step": 56250
     },
     {
-      "epoch": 6.059627596598859,
-      "grad_norm": 0.805608868598938,
-      "learning_rate": 0.00023681284344359444,
-      "loss": 3.2538,
+      "epoch": 6.070080862533692,
+      "grad_norm": 0.7111284732818604,
+      "learning_rate": 0.00023619859686994062,
+      "loss": 3.2516,
       "step": 56300
     },
     {
-      "epoch": 6.065009148638468,
-      "grad_norm": 0.8110917210578918,
-      "learning_rate": 0.000236489602413533,
-      "loss": 3.246,
+      "epoch": 6.0754716981132075,
+      "grad_norm": 0.7226814031600952,
+      "learning_rate": 0.00023587479762547217,
+      "loss": 3.2614,
       "step": 56350
     },
     {
-      "epoch": 6.070390700678075,
-      "grad_norm": 0.8059680461883545,
-      "learning_rate": 0.0002361663613834716,
-      "loss": 3.2608,
+      "epoch": 6.080862533692723,
+      "grad_norm": 0.708491861820221,
+      "learning_rate": 0.00023555099838100374,
+      "loss": 3.2659,
       "step": 56400
     },
     {
-      "epoch": 6.075772252717684,
-      "grad_norm": 0.7581197619438171,
-      "learning_rate": 0.00023584312035341017,
-      "loss": 3.2681,
+      "epoch": 6.086253369272237,
+      "grad_norm": 0.7179819941520691,
+      "learning_rate": 0.00023522719913653532,
+      "loss": 3.2567,
       "step": 56450
     },
     {
-      "epoch": 6.081153804757292,
-      "grad_norm": 0.7696651816368103,
-      "learning_rate": 0.00023551987932334876,
-      "loss": 3.2486,
+      "epoch": 6.091644204851752,
+      "grad_norm": 0.7065309286117554,
+      "learning_rate": 0.0002349033998920669,
+      "loss": 3.2648,
       "step": 56500
     },
     {
-      "epoch": 6.0865353567969,
-      "grad_norm": 0.7502686977386475,
-      "learning_rate": 0.00023519663829328735,
-      "loss": 3.2639,
+      "epoch": 6.097035040431267,
+      "grad_norm": 0.6956079602241516,
+      "learning_rate": 0.00023457960064759848,
+      "loss": 3.2849,
       "step": 56550
     },
     {
-      "epoch": 6.091916908836509,
-      "grad_norm": 0.7801194190979004,
-      "learning_rate": 0.00023487339726322592,
-      "loss": 3.2652,
+      "epoch": 6.1024258760107815,
+      "grad_norm": 0.7147029638290405,
+      "learning_rate": 0.00023425580140313005,
+      "loss": 3.2829,
       "step": 56600
     },
     {
-      "epoch": 6.097298460876116,
-      "grad_norm": 0.7480630874633789,
-      "learning_rate": 0.0002345501562331645,
-      "loss": 3.2495,
+      "epoch": 6.107816711590297,
+      "grad_norm": 0.7096705436706543,
+      "learning_rate": 0.00023393200215866163,
+      "loss": 3.2555,
       "step": 56650
     },
     {
-      "epoch": 6.102680012915725,
-      "grad_norm": 0.7701714038848877,
-      "learning_rate": 0.0002342269152031031,
-      "loss": 3.27,
+      "epoch": 6.113207547169812,
+      "grad_norm": 0.7230135202407837,
+      "learning_rate": 0.00023360820291419316,
+      "loss": 3.2777,
       "step": 56700
     },
     {
-      "epoch": 6.108061564955333,
-      "grad_norm": 0.7400213479995728,
-      "learning_rate": 0.00023390367417304168,
-      "loss": 3.2617,
+      "epoch": 6.118598382749326,
+      "grad_norm": 0.6688002347946167,
+      "learning_rate": 0.00023328440366972473,
+      "loss": 3.2658,
       "step": 56750
     },
     {
-      "epoch": 6.113443116994941,
-      "grad_norm": 0.7020822763442993,
-      "learning_rate": 0.00023358043314298025,
-      "loss": 3.2603,
+      "epoch": 6.123989218328841,
+      "grad_norm": 0.7019734382629395,
+      "learning_rate": 0.0002329606044252563,
+      "loss": 3.258,
       "step": 56800
     },
     {
-      "epoch": 6.11882466903455,
-      "grad_norm": 0.7655128240585327,
-      "learning_rate": 0.00023325719211291887,
-      "loss": 3.2421,
+      "epoch": 6.129380053908355,
+      "grad_norm": 0.7122552990913391,
+      "learning_rate": 0.0002326368051807879,
+      "loss": 3.2585,
       "step": 56850
     },
     {
-      "epoch": 6.124206221074158,
-      "grad_norm": 0.7835574150085449,
-      "learning_rate": 0.00023293395108285744,
-      "loss": 3.2615,
+      "epoch": 6.1347708894878705,
+      "grad_norm": 0.7427650690078735,
+      "learning_rate": 0.00023231300593631947,
+      "loss": 3.2579,
       "step": 56900
     },
     {
-      "epoch": 6.129587773113766,
-      "grad_norm": 0.7708025574684143,
-      "learning_rate": 0.000232610710052796,
-      "loss": 3.2681,
+      "epoch": 6.140161725067386,
+      "grad_norm": 0.7284014225006104,
+      "learning_rate": 0.00023198920669185104,
+      "loss": 3.2703,
       "step": 56950
     },
     {
-      "epoch": 6.134969325153374,
-      "grad_norm": 0.7812603116035461,
-      "learning_rate": 0.0002322874690227346,
-      "loss": 3.2706,
+      "epoch": 6.1455525606469,
+      "grad_norm": 0.7284674048423767,
+      "learning_rate": 0.00023166540744738262,
+      "loss": 3.2657,
       "step": 57000
     },
     {
-      "epoch": 6.134969325153374,
-      "eval_accuracy": 0.3842589780696098,
-      "eval_loss": 3.382967472076416,
-      "eval_runtime": 180.3604,
-      "eval_samples_per_second": 99.861,
-      "eval_steps_per_second": 6.243,
+      "epoch": 6.1455525606469,
+      "eval_accuracy": 0.3841316368850912,
+      "eval_loss": 3.384326457977295,
+      "eval_runtime": 184.9132,
+      "eval_samples_per_second": 97.402,
+      "eval_steps_per_second": 6.089,
       "step": 57000
     },
     {
-      "epoch": 6.140350877192983,
-      "grad_norm": 0.7508556842803955,
-      "learning_rate": 0.0002319642279926732,
-      "loss": 3.2406,
+      "epoch": 6.150943396226415,
+      "grad_norm": 0.7034928798675537,
+      "learning_rate": 0.0002313416082029142,
+      "loss": 3.2762,
       "step": 57050
     },
     {
-      "epoch": 6.1457324292325906,
-      "grad_norm": 0.7822660207748413,
-      "learning_rate": 0.00023164098696261176,
-      "loss": 3.2669,
+      "epoch": 6.15633423180593,
+      "grad_norm": 0.7338201999664307,
+      "learning_rate": 0.00023101780895844572,
+      "loss": 3.2585,
       "step": 57100
     },
     {
-      "epoch": 6.151113981272199,
-      "grad_norm": 0.7589568495750427,
-      "learning_rate": 0.00023131774593255036,
-      "loss": 3.2541,
+      "epoch": 6.1617250673854445,
+      "grad_norm": 0.7155392169952393,
+      "learning_rate": 0.0002306940097139773,
+      "loss": 3.27,
       "step": 57150
     },
     {
-      "epoch": 6.156495533311807,
-      "grad_norm": 0.8089141845703125,
-      "learning_rate": 0.00023099450490248892,
-      "loss": 3.2655,
+      "epoch": 6.16711590296496,
+      "grad_norm": 0.7334120869636536,
+      "learning_rate": 0.00023037021046950888,
+      "loss": 3.2631,
       "step": 57200
     },
     {
-      "epoch": 6.161877085351415,
-      "grad_norm": 0.7813968658447266,
-      "learning_rate": 0.00023067126387242754,
-      "loss": 3.2779,
+      "epoch": 6.172506738544475,
+      "grad_norm": 0.7121962904930115,
+      "learning_rate": 0.00023004641122504046,
+      "loss": 3.2653,
       "step": 57250
     },
     {
-      "epoch": 6.167258637391024,
-      "grad_norm": 0.7517101168632507,
-      "learning_rate": 0.0002303480228423661,
-      "loss": 3.261,
+      "epoch": 6.177897574123989,
+      "grad_norm": 0.7687353491783142,
+      "learning_rate": 0.00022972261198057203,
+      "loss": 3.2666,
       "step": 57300
     },
     {
-      "epoch": 6.1726401894306315,
-      "grad_norm": 0.7565500140190125,
-      "learning_rate": 0.00023002478181230468,
-      "loss": 3.2513,
+      "epoch": 6.183288409703504,
+      "grad_norm": 0.779589831829071,
+      "learning_rate": 0.0002293988127361036,
+      "loss": 3.2713,
       "step": 57350
     },
     {
-      "epoch": 6.17802174147024,
-      "grad_norm": 0.722113311290741,
-      "learning_rate": 0.00022970154078224327,
-      "loss": 3.2755,
+      "epoch": 6.188679245283019,
+      "grad_norm": 0.7740386128425598,
+      "learning_rate": 0.00022907501349163516,
+      "loss": 3.2798,
       "step": 57400
     },
     {
-      "epoch": 6.183403293509849,
-      "grad_norm": 0.805997371673584,
-      "learning_rate": 0.00022937829975218187,
-      "loss": 3.2674,
+      "epoch": 6.1940700808625335,
+      "grad_norm": 0.8117671608924866,
+      "learning_rate": 0.00022875121424716674,
+      "loss": 3.2933,
       "step": 57450
     },
     {
-      "epoch": 6.188784845549456,
-      "grad_norm": 0.7368705868721008,
-      "learning_rate": 0.00022905505872212044,
-      "loss": 3.2627,
+      "epoch": 6.199460916442049,
+      "grad_norm": 0.7242795825004578,
+      "learning_rate": 0.00022842741500269832,
+      "loss": 3.2469,
       "step": 57500
     },
     {
-      "epoch": 6.194166397589065,
-      "grad_norm": 0.7440856099128723,
-      "learning_rate": 0.00022873181769205903,
-      "loss": 3.2576,
+      "epoch": 6.204851752021563,
+      "grad_norm": 0.759215772151947,
+      "learning_rate": 0.00022810361575822987,
+      "loss": 3.2977,
       "step": 57550
     },
     {
-      "epoch": 6.1995479496286725,
-      "grad_norm": 0.8116312623023987,
-      "learning_rate": 0.0002284085766619976,
-      "loss": 3.2743,
+      "epoch": 6.210242587601078,
+      "grad_norm": 0.7306986451148987,
+      "learning_rate": 0.00022777981651376145,
+      "loss": 3.2674,
       "step": 57600
     },
     {
-      "epoch": 6.204929501668281,
-      "grad_norm": 0.7270071506500244,
-      "learning_rate": 0.0002280853356319362,
-      "loss": 3.2859,
+      "epoch": 6.215633423180593,
+      "grad_norm": 0.7148694396018982,
+      "learning_rate": 0.00022745601726929302,
+      "loss": 3.2705,
       "step": 57650
     },
     {
-      "epoch": 6.21031105370789,
-      "grad_norm": 0.7267986536026001,
-      "learning_rate": 0.0002277620946018748,
-      "loss": 3.2578,
+      "epoch": 6.2210242587601075,
+      "grad_norm": 0.7374268770217896,
+      "learning_rate": 0.00022713221802482457,
+      "loss": 3.29,
       "step": 57700
     },
     {
-      "epoch": 6.215692605747497,
-      "grad_norm": 0.7756433486938477,
-      "learning_rate": 0.00022743885357181336,
-      "loss": 3.2718,
+      "epoch": 6.226415094339623,
+      "grad_norm": 0.7002312541007996,
+      "learning_rate": 0.00022681489476524553,
+      "loss": 3.2757,
       "step": 57750
     },
     {
-      "epoch": 6.221074157787106,
-      "grad_norm": 0.7905489206314087,
-      "learning_rate": 0.00022711561254175192,
-      "loss": 3.2755,
+      "epoch": 6.231805929919138,
+      "grad_norm": 0.77752286195755,
+      "learning_rate": 0.0002264910955207771,
+      "loss": 3.28,
       "step": 57800
     },
     {
-      "epoch": 6.226455709826714,
-      "grad_norm": 0.8063260912895203,
-      "learning_rate": 0.00022679237151169054,
-      "loss": 3.2735,
+      "epoch": 6.237196765498652,
+      "grad_norm": 0.7070091366767883,
+      "learning_rate": 0.00022616729627630868,
+      "loss": 3.2798,
       "step": 57850
     },
     {
-      "epoch": 6.231837261866322,
-      "grad_norm": 0.7980824708938599,
-      "learning_rate": 0.0002264691304816291,
-      "loss": 3.2635,
+      "epoch": 6.242587601078167,
+      "grad_norm": 0.7434335350990295,
+      "learning_rate": 0.00022584349703184023,
+      "loss": 3.2808,
       "step": 57900
     },
     {
-      "epoch": 6.237218813905931,
-      "grad_norm": 0.7567322254180908,
-      "learning_rate": 0.00022614588945156768,
-      "loss": 3.2738,
+      "epoch": 6.247978436657682,
+      "grad_norm": 0.7469635009765625,
+      "learning_rate": 0.0002255196977873718,
+      "loss": 3.263,
       "step": 57950
     },
     {
-      "epoch": 6.242600365945538,
-      "grad_norm": 0.7856761813163757,
-      "learning_rate": 0.0002258226484215063,
-      "loss": 3.28,
+      "epoch": 6.2533692722371965,
+      "grad_norm": 0.7343301773071289,
+      "learning_rate": 0.0002251958985429034,
+      "loss": 3.2838,
       "step": 58000
     },
     {
-      "epoch": 6.242600365945538,
-      "eval_accuracy": 0.3846668610104653,
-      "eval_loss": 3.3798773288726807,
-      "eval_runtime": 180.5316,
-      "eval_samples_per_second": 99.766,
-      "eval_steps_per_second": 6.237,
+      "epoch": 6.2533692722371965,
+      "eval_accuracy": 0.38464078431738985,
+      "eval_loss": 3.380790948867798,
+      "eval_runtime": 185.1083,
+      "eval_samples_per_second": 97.3,
+      "eval_steps_per_second": 6.083,
       "step": 58000
     },
     {
-      "epoch": 6.247981917985147,
-      "grad_norm": 0.7729969024658203,
-      "learning_rate": 0.00022549940739144487,
-      "loss": 3.2857,
+      "epoch": 6.258760107816712,
+      "grad_norm": 0.7176439762115479,
+      "learning_rate": 0.00022487209929843494,
+      "loss": 3.2724,
       "step": 58050
     },
     {
-      "epoch": 6.253363470024755,
-      "grad_norm": 0.8097330927848816,
-      "learning_rate": 0.00022517616636138344,
-      "loss": 3.2717,
+      "epoch": 6.264150943396227,
+      "grad_norm": 0.7113427519798279,
+      "learning_rate": 0.00022454830005396652,
+      "loss": 3.278,
       "step": 58100
     },
     {
-      "epoch": 6.258745022064363,
-      "grad_norm": 0.8649580478668213,
-      "learning_rate": 0.00022485292533132203,
-      "loss": 3.2832,
+      "epoch": 6.269541778975741,
+      "grad_norm": 0.7005300521850586,
+      "learning_rate": 0.0002242245008094981,
+      "loss": 3.2739,
       "step": 58150
     },
     {
-      "epoch": 6.264126574103972,
-      "grad_norm": 0.771740734577179,
-      "learning_rate": 0.00022452968430126063,
-      "loss": 3.2734,
+      "epoch": 6.274932614555256,
+      "grad_norm": 0.719237744808197,
+      "learning_rate": 0.00022390070156502967,
+      "loss": 3.2707,
       "step": 58200
     },
     {
-      "epoch": 6.26950812614358,
-      "grad_norm": 0.7663628458976746,
-      "learning_rate": 0.00022420644327119922,
-      "loss": 3.2736,
+      "epoch": 6.280323450134771,
+      "grad_norm": 0.808284342288971,
+      "learning_rate": 0.00022357690232056125,
+      "loss": 3.2804,
       "step": 58250
     },
     {
-      "epoch": 6.274889678183188,
-      "grad_norm": 0.7669972777366638,
-      "learning_rate": 0.0002238832022411378,
-      "loss": 3.2712,
+      "epoch": 6.285714285714286,
+      "grad_norm": 0.7207859754562378,
+      "learning_rate": 0.0002232531030760928,
+      "loss": 3.2658,
       "step": 58300
     },
     {
-      "epoch": 6.280271230222796,
-      "grad_norm": 0.7995150089263916,
-      "learning_rate": 0.00022355996121107636,
-      "loss": 3.2764,
+      "epoch": 6.291105121293801,
+      "grad_norm": 0.7794989943504333,
+      "learning_rate": 0.00022292930383162435,
+      "loss": 3.2928,
       "step": 58350
     },
     {
-      "epoch": 6.285652782262405,
-      "grad_norm": 0.7507680654525757,
-      "learning_rate": 0.00022323672018101498,
-      "loss": 3.2882,
+      "epoch": 6.296495956873315,
+      "grad_norm": 0.7986652851104736,
+      "learning_rate": 0.00022260550458715593,
+      "loss": 3.2741,
       "step": 58400
     },
     {
-      "epoch": 6.2910343343020125,
-      "grad_norm": 0.8180110454559326,
-      "learning_rate": 0.00022291347915095355,
-      "loss": 3.2764,
+      "epoch": 6.30188679245283,
+      "grad_norm": 0.7763763666152954,
+      "learning_rate": 0.0002222817053426875,
+      "loss": 3.2745,
       "step": 58450
     },
     {
-      "epoch": 6.296415886341621,
-      "grad_norm": 0.7749719619750977,
-      "learning_rate": 0.0002225902381208921,
-      "loss": 3.272,
+      "epoch": 6.307277628032345,
+      "grad_norm": 0.7197876572608948,
+      "learning_rate": 0.00022195790609821908,
+      "loss": 3.2668,
       "step": 58500
     },
     {
-      "epoch": 6.301797438381229,
-      "grad_norm": 0.7502884864807129,
-      "learning_rate": 0.00022226699709083073,
-      "loss": 3.2762,
+      "epoch": 6.3126684636118595,
+      "grad_norm": 0.7718214392662048,
+      "learning_rate": 0.00022163410685375066,
+      "loss": 3.2773,
       "step": 58550
     },
     {
-      "epoch": 6.307178990420837,
-      "grad_norm": 0.8392398357391357,
-      "learning_rate": 0.00022195022088137051,
-      "loss": 3.2646,
+      "epoch": 6.318059299191375,
+      "grad_norm": 0.7384246587753296,
+      "learning_rate": 0.00022131030760928224,
+      "loss": 3.2932,
       "step": 58600
     },
     {
-      "epoch": 6.312560542460446,
-      "grad_norm": 0.7521066665649414,
-      "learning_rate": 0.0002216269798513091,
-      "loss": 3.2673,
+      "epoch": 6.32345013477089,
+      "grad_norm": 0.7198066115379333,
+      "learning_rate": 0.00022098650836481382,
+      "loss": 3.2703,
       "step": 58650
     },
     {
-      "epoch": 6.3179420945000535,
-      "grad_norm": 0.7701011300086975,
-      "learning_rate": 0.00022130373882124768,
-      "loss": 3.2645,
+      "epoch": 6.328840970350404,
+      "grad_norm": 0.7592787742614746,
+      "learning_rate": 0.0002206627091203454,
+      "loss": 3.2831,
       "step": 58700
     },
     {
-      "epoch": 6.323323646539662,
-      "grad_norm": 0.8215457797050476,
-      "learning_rate": 0.00022098049779118627,
-      "loss": 3.2603,
+      "epoch": 6.334231805929919,
+      "grad_norm": 0.7230469584465027,
+      "learning_rate": 0.00022033890987587692,
+      "loss": 3.2884,
       "step": 58750
     },
     {
-      "epoch": 6.328705198579271,
-      "grad_norm": 0.7722597122192383,
-      "learning_rate": 0.00022065725676112487,
-      "loss": 3.2791,
+      "epoch": 6.339622641509434,
+      "grad_norm": 0.7920795679092407,
+      "learning_rate": 0.0002200151106314085,
+      "loss": 3.2864,
       "step": 58800
     },
     {
-      "epoch": 6.334086750618878,
-      "grad_norm": 0.7506366968154907,
-      "learning_rate": 0.00022033401573106343,
-      "loss": 3.2869,
+      "epoch": 6.345013477088949,
+      "grad_norm": 0.7017115950584412,
+      "learning_rate": 0.00021969131138694007,
+      "loss": 3.2721,
       "step": 58850
     },
     {
-      "epoch": 6.339468302658487,
-      "grad_norm": 0.7620330452919006,
-      "learning_rate": 0.000220010774701002,
+      "epoch": 6.350404312668464,
+      "grad_norm": 0.7787772417068481,
+      "learning_rate": 0.00021936751214247165,
       "loss": 3.274,
       "step": 58900
     },
     {
-      "epoch": 6.344849854698095,
-      "grad_norm": 0.7641044855117798,
-      "learning_rate": 0.00021968753367094062,
-      "loss": 3.2703,
+      "epoch": 6.355795148247978,
+      "grad_norm": 0.7387892603874207,
+      "learning_rate": 0.00021904371289800323,
+      "loss": 3.2744,
       "step": 58950
     },
     {
-      "epoch": 6.350231406737703,
-      "grad_norm": 0.7357539534568787,
-      "learning_rate": 0.0002193642926408792,
-      "loss": 3.277,
+      "epoch": 6.361185983827493,
+      "grad_norm": 0.7931119203567505,
+      "learning_rate": 0.0002187199136535348,
+      "loss": 3.2684,
       "step": 59000
     },
     {
-      "epoch": 6.350231406737703,
-      "eval_accuracy": 0.3850675728607251,
-      "eval_loss": 3.3782007694244385,
-      "eval_runtime": 180.5172,
-      "eval_samples_per_second": 99.774,
-      "eval_steps_per_second": 6.238,
+      "epoch": 6.361185983827493,
+      "eval_accuracy": 0.38483874987898786,
+      "eval_loss": 3.378488302230835,
+      "eval_runtime": 184.6031,
+      "eval_samples_per_second": 97.566,
+      "eval_steps_per_second": 6.1,
       "step": 59000
     },
     {
-      "epoch": 6.355612958777312,
-      "grad_norm": 0.793239951133728,
-      "learning_rate": 0.00021904105161081778,
-      "loss": 3.2845,
+      "epoch": 6.366576819407008,
+      "grad_norm": 0.7148496508598328,
+      "learning_rate": 0.00021839611440906638,
+      "loss": 3.2747,
       "step": 59050
     },
     {
-      "epoch": 6.360994510816919,
-      "grad_norm": 0.7626572251319885,
-      "learning_rate": 0.00021871781058075638,
-      "loss": 3.2751,
+      "epoch": 6.3719676549865225,
+      "grad_norm": 0.733859658241272,
+      "learning_rate": 0.00021807231516459793,
+      "loss": 3.2728,
       "step": 59100
     },
     {
-      "epoch": 6.366376062856528,
-      "grad_norm": 0.7418571710586548,
-      "learning_rate": 0.00021839456955069495,
-      "loss": 3.285,
+      "epoch": 6.377358490566038,
+      "grad_norm": 0.7047247886657715,
+      "learning_rate": 0.00021774851592012948,
+      "loss": 3.2855,
       "step": 59150
     },
     {
-      "epoch": 6.371757614896136,
-      "grad_norm": 0.7725799679756165,
-      "learning_rate": 0.00021807132852063354,
-      "loss": 3.2829,
+      "epoch": 6.382749326145553,
+      "grad_norm": 0.7953727841377258,
+      "learning_rate": 0.00021742471667566106,
+      "loss": 3.2883,
       "step": 59200
     },
     {
-      "epoch": 6.377139166935744,
-      "grad_norm": 0.731188952922821,
-      "learning_rate": 0.0002177480874905721,
-      "loss": 3.2764,
+      "epoch": 6.388140161725067,
+      "grad_norm": 0.7541425228118896,
+      "learning_rate": 0.00021710091743119264,
+      "loss": 3.2769,
       "step": 59250
     },
     {
-      "epoch": 6.382520718975353,
-      "grad_norm": 0.7765207290649414,
-      "learning_rate": 0.0002174248464605107,
-      "loss": 3.2717,
+      "epoch": 6.393530997304582,
+      "grad_norm": 0.7310684323310852,
+      "learning_rate": 0.00021677711818672422,
+      "loss": 3.298,
       "step": 59300
     },
     {
-      "epoch": 6.387902271014961,
-      "grad_norm": 0.7506345510482788,
-      "learning_rate": 0.0002171016054304493,
-      "loss": 3.2679,
+      "epoch": 6.398921832884097,
+      "grad_norm": 0.7032855749130249,
+      "learning_rate": 0.0002164533189422558,
+      "loss": 3.2742,
       "step": 59350
     },
     {
-      "epoch": 6.393283823054569,
-      "grad_norm": 0.7452055215835571,
-      "learning_rate": 0.00021677836440038787,
-      "loss": 3.2577,
+      "epoch": 6.404312668463612,
+      "grad_norm": 0.7056180834770203,
+      "learning_rate": 0.00021612951969778734,
+      "loss": 3.274,
       "step": 59400
     },
     {
-      "epoch": 6.398665375094177,
-      "grad_norm": 0.8073831796646118,
-      "learning_rate": 0.00021645512337032643,
-      "loss": 3.2606,
+      "epoch": 6.409703504043127,
+      "grad_norm": 0.7354710698127747,
+      "learning_rate": 0.00021580572045331892,
+      "loss": 3.2877,
       "step": 59450
     },
     {
-      "epoch": 6.404046927133785,
-      "grad_norm": 0.7789980173110962,
-      "learning_rate": 0.00021613188234026506,
-      "loss": 3.2712,
+      "epoch": 6.415094339622642,
+      "grad_norm": 0.7771357893943787,
+      "learning_rate": 0.0002154819212088505,
+      "loss": 3.2757,
       "step": 59500
     },
     {
-      "epoch": 6.4094284791733935,
-      "grad_norm": 0.8396959900856018,
-      "learning_rate": 0.00021580864131020362,
-      "loss": 3.2768,
+      "epoch": 6.420485175202156,
+      "grad_norm": 1.259022831916809,
+      "learning_rate": 0.00021515812196438208,
+      "loss": 3.2676,
       "step": 59550
     },
     {
-      "epoch": 6.414810031213002,
-      "grad_norm": 0.797767162322998,
-      "learning_rate": 0.0002154854002801422,
-      "loss": 3.2813,
+      "epoch": 6.425876010781671,
+      "grad_norm": 0.7444080710411072,
+      "learning_rate": 0.00021483432271991363,
+      "loss": 3.2774,
       "step": 59600
     },
     {
-      "epoch": 6.42019158325261,
-      "grad_norm": 0.802899956703186,
-      "learning_rate": 0.0002151621592500808,
-      "loss": 3.2889,
+      "epoch": 6.431266846361186,
+      "grad_norm": 0.7366527915000916,
+      "learning_rate": 0.0002145105234754452,
+      "loss": 3.284,
       "step": 59650
     },
     {
-      "epoch": 6.425573135292218,
-      "grad_norm": 0.8054869771003723,
-      "learning_rate": 0.00021483891822001938,
-      "loss": 3.2716,
+      "epoch": 6.436657681940701,
+      "grad_norm": 0.7495415806770325,
+      "learning_rate": 0.00021418672423097676,
+      "loss": 3.2778,
       "step": 59700
     },
     {
-      "epoch": 6.430954687331827,
-      "grad_norm": 0.7798833250999451,
-      "learning_rate": 0.00021451567718995795,
-      "loss": 3.2781,
+      "epoch": 6.442048517520216,
+      "grad_norm": 0.7302351593971252,
+      "learning_rate": 0.00021386292498650833,
+      "loss": 3.2894,
       "step": 59750
     },
     {
-      "epoch": 6.4363362393714345,
-      "grad_norm": 0.7423691153526306,
-      "learning_rate": 0.00021419243615989654,
-      "loss": 3.2763,
+      "epoch": 6.44743935309973,
+      "grad_norm": 0.7209936380386353,
+      "learning_rate": 0.0002135391257420399,
+      "loss": 3.2789,
       "step": 59800
     },
     {
-      "epoch": 6.441717791411043,
-      "grad_norm": 0.8051731586456299,
-      "learning_rate": 0.00021386919512983514,
-      "loss": 3.2666,
+      "epoch": 6.452830188679245,
+      "grad_norm": 0.7588741779327393,
+      "learning_rate": 0.0002132153264975715,
+      "loss": 3.2869,
       "step": 59850
     },
     {
-      "epoch": 6.447099343450651,
-      "grad_norm": 0.8098742365837097,
-      "learning_rate": 0.0002135459540997737,
-      "loss": 3.2656,
+      "epoch": 6.45822102425876,
+      "grad_norm": 0.7343195080757141,
+      "learning_rate": 0.00021289152725310307,
+      "loss": 3.2743,
       "step": 59900
     },
     {
-      "epoch": 6.452480895490259,
-      "grad_norm": 0.8203098177909851,
-      "learning_rate": 0.0002132227130697123,
-      "loss": 3.2729,
+      "epoch": 6.463611859838275,
+      "grad_norm": 0.6876043081283569,
+      "learning_rate": 0.00021256772800863464,
+      "loss": 3.2879,
       "step": 59950
     },
     {
-      "epoch": 6.457862447529868,
-      "grad_norm": 0.736182689666748,
-      "learning_rate": 0.00021289947203965087,
-      "loss": 3.2727,
+      "epoch": 6.46900269541779,
+      "grad_norm": 0.7846079468727112,
+      "learning_rate": 0.0002122439287641662,
+      "loss": 3.2841,
       "step": 60000
     },
     {
-      "epoch": 6.457862447529868,
-      "eval_accuracy": 0.385509246849691,
-      "eval_loss": 3.372286558151245,
-      "eval_runtime": 180.8685,
-      "eval_samples_per_second": 99.581,
-      "eval_steps_per_second": 6.226,
+      "epoch": 6.46900269541779,
+      "eval_accuracy": 0.3849802159389223,
+      "eval_loss": 3.3735952377319336,
+      "eval_runtime": 184.7204,
+      "eval_samples_per_second": 97.504,
+      "eval_steps_per_second": 6.096,
       "step": 60000
     },
     {
-      "epoch": 6.4632439995694755,
-      "grad_norm": 0.7694886326789856,
-      "learning_rate": 0.0002125762310095895,
-      "loss": 3.2795,
+      "epoch": 6.474393530997305,
+      "grad_norm": 0.7180082201957703,
+      "learning_rate": 0.00021192012951969775,
+      "loss": 3.2929,
       "step": 60050
     },
     {
-      "epoch": 6.468625551609084,
-      "grad_norm": 0.7625947594642639,
-      "learning_rate": 0.00021225298997952806,
-      "loss": 3.2678,
+      "epoch": 6.479784366576819,
+      "grad_norm": 0.7589936852455139,
+      "learning_rate": 0.00021159633027522932,
+      "loss": 3.2898,
       "step": 60100
     },
     {
-      "epoch": 6.474007103648693,
-      "grad_norm": 0.7633724808692932,
-      "learning_rate": 0.00021192974894946662,
-      "loss": 3.2773,
+      "epoch": 6.485175202156334,
+      "grad_norm": 0.748837947845459,
+      "learning_rate": 0.0002112725310307609,
+      "loss": 3.2811,
       "step": 60150
     },
     {
-      "epoch": 6.4793886556883,
-      "grad_norm": 0.7859768271446228,
-      "learning_rate": 0.00021160650791940524,
-      "loss": 3.2929,
+      "epoch": 6.490566037735849,
+      "grad_norm": 0.8100166916847229,
+      "learning_rate": 0.00021094873178629248,
+      "loss": 3.3062,
       "step": 60200
     },
     {
-      "epoch": 6.484770207727909,
-      "grad_norm": 0.7822881937026978,
-      "learning_rate": 0.0002112832668893438,
-      "loss": 3.2923,
+      "epoch": 6.495956873315364,
+      "grad_norm": 0.6930977702140808,
+      "learning_rate": 0.00021062493254182406,
+      "loss": 3.279,
       "step": 60250
     },
     {
-      "epoch": 6.490151759767517,
-      "grad_norm": 0.7258760929107666,
-      "learning_rate": 0.00021096002585928238,
-      "loss": 3.2784,
+      "epoch": 6.501347708894879,
+      "grad_norm": 0.7058539986610413,
+      "learning_rate": 0.00021030113329735563,
+      "loss": 3.2849,
       "step": 60300
     },
     {
-      "epoch": 6.495533311807125,
-      "grad_norm": 0.7965183854103088,
-      "learning_rate": 0.00021063678482922097,
-      "loss": 3.2617,
+      "epoch": 6.506738544474393,
+      "grad_norm": 0.7700808644294739,
+      "learning_rate": 0.0002099773340528872,
+      "loss": 3.2861,
       "step": 60350
     },
     {
-      "epoch": 6.500914863846734,
-      "grad_norm": 0.8430307507514954,
-      "learning_rate": 0.00021031354379915957,
-      "loss": 3.2637,
+      "epoch": 6.512129380053908,
+      "grad_norm": 0.7218989729881287,
+      "learning_rate": 0.0002096535348084188,
+      "loss": 3.2817,
       "step": 60400
     },
     {
-      "epoch": 6.506296415886341,
-      "grad_norm": 0.7958040833473206,
-      "learning_rate": 0.00020999030276909814,
-      "loss": 3.2915,
+      "epoch": 6.517520215633423,
+      "grad_norm": 0.7468814253807068,
+      "learning_rate": 0.0002093297355639503,
+      "loss": 3.2966,
       "step": 60450
     },
     {
-      "epoch": 6.51167796792595,
-      "grad_norm": 0.7817371487617493,
-      "learning_rate": 0.00020966706173903673,
-      "loss": 3.2911,
+      "epoch": 6.5229110512129385,
+      "grad_norm": 0.7302692532539368,
+      "learning_rate": 0.0002090059363194819,
+      "loss": 3.2844,
       "step": 60500
     },
     {
-      "epoch": 6.517059519965558,
-      "grad_norm": 0.7723913192749023,
-      "learning_rate": 0.0002093438207089753,
-      "loss": 3.2695,
+      "epoch": 6.528301886792453,
+      "grad_norm": 0.7506006956100464,
+      "learning_rate": 0.00020868213707501347,
+      "loss": 3.3064,
       "step": 60550
     },
     {
-      "epoch": 6.522441072005166,
-      "grad_norm": 0.7523823380470276,
-      "learning_rate": 0.00020902057967891387,
-      "loss": 3.2835,
+      "epoch": 6.533692722371968,
+      "grad_norm": 0.7327467203140259,
+      "learning_rate": 0.00020835833783054505,
+      "loss": 3.2984,
       "step": 60600
     },
     {
-      "epoch": 6.5278226240447745,
-      "grad_norm": 0.7971324920654297,
-      "learning_rate": 0.0002086973386488525,
-      "loss": 3.2922,
+      "epoch": 6.539083557951482,
+      "grad_norm": 0.7747229933738708,
+      "learning_rate": 0.00020803453858607662,
+      "loss": 3.2889,
       "step": 60650
     },
     {
-      "epoch": 6.533204176084383,
-      "grad_norm": 0.8252760767936707,
-      "learning_rate": 0.00020837409761879106,
-      "loss": 3.2898,
+      "epoch": 6.544474393530997,
+      "grad_norm": 0.7802739143371582,
+      "learning_rate": 0.0002077107393416082,
+      "loss": 3.3025,
       "step": 60700
     },
     {
-      "epoch": 6.538585728123991,
-      "grad_norm": 0.7614521980285645,
-      "learning_rate": 0.00020805085658872962,
-      "loss": 3.2906,
+      "epoch": 6.549865229110512,
+      "grad_norm": 0.7712839245796204,
+      "learning_rate": 0.00020738694009713975,
+      "loss": 3.29,
       "step": 60750
     },
     {
-      "epoch": 6.543967280163599,
-      "grad_norm": 0.8246181011199951,
-      "learning_rate": 0.00020772761555866825,
-      "loss": 3.2725,
+      "epoch": 6.555256064690027,
+      "grad_norm": 0.72076416015625,
+      "learning_rate": 0.00020706314085267133,
+      "loss": 3.2813,
       "step": 60800
     },
     {
-      "epoch": 6.549348832203208,
-      "grad_norm": 0.7531991004943848,
-      "learning_rate": 0.0002074043745286068,
-      "loss": 3.274,
+      "epoch": 6.560646900269542,
+      "grad_norm": 0.7788540720939636,
+      "learning_rate": 0.00020673934160820288,
+      "loss": 3.2969,
       "step": 60850
     },
     {
-      "epoch": 6.5547303842428155,
-      "grad_norm": 0.8107429146766663,
-      "learning_rate": 0.00020708113349854538,
-      "loss": 3.2757,
+      "epoch": 6.566037735849057,
+      "grad_norm": 0.7517972588539124,
+      "learning_rate": 0.00020641554236373446,
+      "loss": 3.2807,
       "step": 60900
     },
     {
-      "epoch": 6.560111936282424,
-      "grad_norm": 0.8336337208747864,
-      "learning_rate": 0.00020675789246848397,
-      "loss": 3.2756,
+      "epoch": 6.571428571428571,
+      "grad_norm": 0.7544095516204834,
+      "learning_rate": 0.00020609174311926604,
+      "loss": 3.2833,
       "step": 60950
     },
     {
-      "epoch": 6.565493488322032,
-      "grad_norm": 0.8028927445411682,
-      "learning_rate": 0.00020643465143842257,
-      "loss": 3.2838,
+      "epoch": 6.576819407008086,
+      "grad_norm": 0.739867091178894,
+      "learning_rate": 0.0002057679438747976,
+      "loss": 3.2765,
       "step": 61000
     },
     {
-      "epoch": 6.565493488322032,
-      "eval_accuracy": 0.3859593957639064,
-      "eval_loss": 3.3682973384857178,
-      "eval_runtime": 180.3806,
-      "eval_samples_per_second": 99.85,
-      "eval_steps_per_second": 6.242,
+      "epoch": 6.576819407008086,
+      "eval_accuracy": 0.3854455762574317,
+      "eval_loss": 3.370424270629883,
+      "eval_runtime": 184.81,
+      "eval_samples_per_second": 97.457,
+      "eval_steps_per_second": 6.093,
       "step": 61000
     },
     {
-      "epoch": 6.57087504036164,
-      "grad_norm": 0.8147567510604858,
-      "learning_rate": 0.00020611141040836116,
-      "loss": 3.2949,
+      "epoch": 6.5822102425876015,
+      "grad_norm": 0.7352069616317749,
+      "learning_rate": 0.00020544414463032916,
+      "loss": 3.2759,
       "step": 61050
     },
     {
-      "epoch": 6.576256592401249,
-      "grad_norm": 0.8347201347351074,
-      "learning_rate": 0.00020578816937829973,
-      "loss": 3.2929,
+      "epoch": 6.587601078167116,
+      "grad_norm": 0.7170955538749695,
+      "learning_rate": 0.00020512034538586074,
+      "loss": 3.2656,
       "step": 61100
     },
     {
-      "epoch": 6.5816381444408565,
-      "grad_norm": 0.8149718046188354,
-      "learning_rate": 0.0002054649283482383,
-      "loss": 3.2933,
+      "epoch": 6.592991913746631,
+      "grad_norm": 0.7378682494163513,
+      "learning_rate": 0.00020479654614139232,
+      "loss": 3.28,
       "step": 61150
     },
     {
-      "epoch": 6.587019696480465,
-      "grad_norm": 0.7641071677207947,
-      "learning_rate": 0.00020514168731817692,
-      "loss": 3.2689,
+      "epoch": 6.598382749326145,
+      "grad_norm": 0.785605788230896,
+      "learning_rate": 0.0002044727468969239,
+      "loss": 3.2807,
       "step": 61200
     },
     {
-      "epoch": 6.592401248520073,
-      "grad_norm": 0.8402169942855835,
-      "learning_rate": 0.0002048184462881155,
-      "loss": 3.2796,
+      "epoch": 6.60377358490566,
+      "grad_norm": 0.7256307005882263,
+      "learning_rate": 0.00020414894765245547,
+      "loss": 3.2822,
       "step": 61250
     },
     {
-      "epoch": 6.597782800559681,
-      "grad_norm": 0.7858189940452576,
-      "learning_rate": 0.00020449520525805406,
-      "loss": 3.29,
+      "epoch": 6.609164420485175,
+      "grad_norm": 0.7596465945243835,
+      "learning_rate": 0.00020382514840798702,
+      "loss": 3.2893,
       "step": 61300
     },
     {
-      "epoch": 6.60316435259929,
-      "grad_norm": 0.7707937359809875,
-      "learning_rate": 0.00020417196422799268,
-      "loss": 3.2823,
+      "epoch": 6.6145552560646905,
+      "grad_norm": 0.7317668199539185,
+      "learning_rate": 0.0002035013491635186,
+      "loss": 3.2854,
       "step": 61350
     },
     {
-      "epoch": 6.608545904638898,
-      "grad_norm": 0.7980074882507324,
-      "learning_rate": 0.00020384872319793125,
-      "loss": 3.2965,
+      "epoch": 6.619946091644205,
+      "grad_norm": 0.8078965544700623,
+      "learning_rate": 0.00020317754991905015,
+      "loss": 3.2705,
       "step": 61400
     },
     {
-      "epoch": 6.613927456678506,
-      "grad_norm": 0.771574854850769,
-      "learning_rate": 0.0002035254821678698,
-      "loss": 3.2852,
+      "epoch": 6.62533692722372,
+      "grad_norm": 0.7488526105880737,
+      "learning_rate": 0.00020285375067458173,
+      "loss": 3.3001,
       "step": 61450
     },
     {
-      "epoch": 6.619309008718115,
-      "grad_norm": 0.7880848050117493,
-      "learning_rate": 0.0002032022411378084,
-      "loss": 3.2822,
+      "epoch": 6.630727762803234,
+      "grad_norm": 0.738555908203125,
+      "learning_rate": 0.0002025299514301133,
+      "loss": 3.2768,
       "step": 61500
     },
     {
-      "epoch": 6.624690560757722,
-      "grad_norm": 0.7877942323684692,
-      "learning_rate": 0.000202879000107747,
-      "loss": 3.264,
+      "epoch": 6.636118598382749,
+      "grad_norm": 0.7701089978218079,
+      "learning_rate": 0.00020220615218564489,
+      "loss": 3.2937,
       "step": 61550
     },
     {
-      "epoch": 6.630072112797331,
-      "grad_norm": 0.7295520901679993,
-      "learning_rate": 0.00020255575907768557,
-      "loss": 3.2691,
+      "epoch": 6.6415094339622645,
+      "grad_norm": 0.6935857534408569,
+      "learning_rate": 0.00020188235294117646,
+      "loss": 3.2764,
       "step": 61600
     },
     {
-      "epoch": 6.635453664836939,
-      "grad_norm": 0.8145275712013245,
-      "learning_rate": 0.00020223251804762416,
-      "loss": 3.2715,
+      "epoch": 6.646900269541779,
+      "grad_norm": 0.740175187587738,
+      "learning_rate": 0.00020155855369670804,
+      "loss": 3.275,
       "step": 61650
     },
     {
-      "epoch": 6.640835216876547,
-      "grad_norm": 0.8078879714012146,
-      "learning_rate": 0.00020190927701756273,
-      "loss": 3.2752,
+      "epoch": 6.652291105121294,
+      "grad_norm": 0.778984010219574,
+      "learning_rate": 0.00020123475445223956,
+      "loss": 3.3037,
       "step": 61700
     },
     {
-      "epoch": 6.6462167689161555,
-      "grad_norm": 1.0516327619552612,
-      "learning_rate": 0.00020158603598750133,
-      "loss": 3.2702,
+      "epoch": 6.657681940700809,
+      "grad_norm": 0.7185824513435364,
+      "learning_rate": 0.00020091743119266052,
+      "loss": 3.2928,
       "step": 61750
     },
     {
-      "epoch": 6.651598320955763,
-      "grad_norm": 0.8313896059989929,
-      "learning_rate": 0.00020126279495743992,
-      "loss": 3.2767,
+      "epoch": 6.663072776280323,
+      "grad_norm": 0.7591161727905273,
+      "learning_rate": 0.0002005936319481921,
+      "loss": 3.2821,
       "step": 61800
     },
     {
-      "epoch": 6.656979872995372,
-      "grad_norm": 0.812926709651947,
-      "learning_rate": 0.0002009395539273785,
-      "loss": 3.2784,
+      "epoch": 6.668463611859838,
+      "grad_norm": 0.7795547246932983,
+      "learning_rate": 0.00020026983270372367,
+      "loss": 3.2903,
       "step": 61850
     },
     {
-      "epoch": 6.66236142503498,
-      "grad_norm": 0.7894045114517212,
-      "learning_rate": 0.00020061631289731706,
-      "loss": 3.2954,
+      "epoch": 6.6738544474393535,
+      "grad_norm": 0.7105275392532349,
+      "learning_rate": 0.00019994603345925525,
+      "loss": 3.2844,
       "step": 61900
     },
     {
-      "epoch": 6.667742977074588,
-      "grad_norm": 0.7732223272323608,
-      "learning_rate": 0.00020029307186725568,
-      "loss": 3.2946,
+      "epoch": 6.679245283018868,
+      "grad_norm": 0.7548275589942932,
+      "learning_rate": 0.00019962223421478683,
+      "loss": 3.2831,
       "step": 61950
     },
     {
-      "epoch": 6.6731245291141965,
-      "grad_norm": 0.7833398580551147,
-      "learning_rate": 0.00019996983083719425,
-      "loss": 3.2802,
+      "epoch": 6.684636118598383,
+      "grad_norm": 0.7622220516204834,
+      "learning_rate": 0.0001992984349703184,
+      "loss": 3.2973,
       "step": 62000
     },
     {
-      "epoch": 6.6731245291141965,
-      "eval_accuracy": 0.38636662678743505,
-      "eval_loss": 3.3642148971557617,
-      "eval_runtime": 180.6181,
-      "eval_samples_per_second": 99.719,
-      "eval_steps_per_second": 6.234,
+      "epoch": 6.684636118598383,
+      "eval_accuracy": 0.3861206366494231,
+      "eval_loss": 3.364516258239746,
+      "eval_runtime": 185.0556,
+      "eval_samples_per_second": 97.328,
+      "eval_steps_per_second": 6.085,
       "step": 62000
     },
     {
-      "epoch": 6.678506081153805,
-      "grad_norm": 0.7945963740348816,
-      "learning_rate": 0.00019964658980713284,
-      "loss": 3.306,
+      "epoch": 6.690026954177897,
+      "grad_norm": 0.7730100750923157,
+      "learning_rate": 0.00019897463572584993,
+      "loss": 3.2831,
       "step": 62050
     },
     {
-      "epoch": 6.683887633193413,
-      "grad_norm": 0.7843768000602722,
-      "learning_rate": 0.00019932334877707144,
-      "loss": 3.2765,
+      "epoch": 6.695417789757412,
+      "grad_norm": 0.7311808466911316,
+      "learning_rate": 0.0001986508364813815,
+      "loss": 3.2819,
       "step": 62100
     },
     {
-      "epoch": 6.689269185233021,
-      "grad_norm": 0.8199366927146912,
-      "learning_rate": 0.00019900657256761124,
-      "loss": 3.2743,
+      "epoch": 6.7008086253369274,
+      "grad_norm": 0.749595046043396,
+      "learning_rate": 0.00019832703723691308,
+      "loss": 3.2862,
       "step": 62150
     },
     {
-      "epoch": 6.69465073727263,
-      "grad_norm": 0.7921463251113892,
-      "learning_rate": 0.0001986833315375498,
-      "loss": 3.284,
+      "epoch": 6.706199460916442,
+      "grad_norm": 0.7545285224914551,
+      "learning_rate": 0.00019800323799244466,
+      "loss": 3.2901,
       "step": 62200
     },
     {
-      "epoch": 6.7000322893122375,
-      "grad_norm": 0.7658042311668396,
-      "learning_rate": 0.00019836009050748838,
-      "loss": 3.2779,
+      "epoch": 6.711590296495957,
+      "grad_norm": 0.7023864984512329,
+      "learning_rate": 0.00019767943874797624,
+      "loss": 3.2911,
       "step": 62250
     },
     {
-      "epoch": 6.705413841351846,
-      "grad_norm": 0.7855052947998047,
-      "learning_rate": 0.000198036849477427,
-      "loss": 3.2939,
+      "epoch": 6.716981132075472,
+      "grad_norm": 0.7340058088302612,
+      "learning_rate": 0.00019735563950350782,
+      "loss": 3.2823,
       "step": 62300
     },
     {
-      "epoch": 6.710795393391454,
-      "grad_norm": 0.778864860534668,
-      "learning_rate": 0.00019771360844736557,
-      "loss": 3.2865,
+      "epoch": 6.722371967654986,
+      "grad_norm": 0.7416219115257263,
+      "learning_rate": 0.0001970318402590394,
+      "loss": 3.286,
       "step": 62350
     },
     {
-      "epoch": 6.716176945431062,
-      "grad_norm": 0.7535867691040039,
-      "learning_rate": 0.00019739036741730413,
-      "loss": 3.281,
+      "epoch": 6.727762803234501,
+      "grad_norm": 0.7272533774375916,
+      "learning_rate": 0.00019670804101457097,
+      "loss": 3.2881,
       "step": 62400
     },
     {
-      "epoch": 6.721558497470671,
-      "grad_norm": 0.8026406168937683,
-      "learning_rate": 0.00019706712638724276,
-      "loss": 3.2898,
+      "epoch": 6.7331536388140165,
+      "grad_norm": 0.729224681854248,
+      "learning_rate": 0.00019638424177010252,
+      "loss": 3.2863,
       "step": 62450
     },
     {
-      "epoch": 6.7269400495102785,
-      "grad_norm": 0.826941192150116,
-      "learning_rate": 0.00019674388535718132,
-      "loss": 3.2813,
+      "epoch": 6.738544474393531,
+      "grad_norm": 0.7703849673271179,
+      "learning_rate": 0.00019606044252563407,
+      "loss": 3.2826,
       "step": 62500
     },
     {
-      "epoch": 6.732321601549887,
-      "grad_norm": 0.8191684484481812,
-      "learning_rate": 0.0001964206443271199,
-      "loss": 3.279,
+      "epoch": 6.743935309973046,
+      "grad_norm": 0.7224081754684448,
+      "learning_rate": 0.00019573664328116565,
+      "loss": 3.2981,
       "step": 62550
     },
     {
-      "epoch": 6.737703153589496,
-      "grad_norm": 0.7996468544006348,
-      "learning_rate": 0.00019609740329705849,
-      "loss": 3.2748,
+      "epoch": 6.74932614555256,
+      "grad_norm": 0.7727299332618713,
+      "learning_rate": 0.00019541284403669723,
+      "loss": 3.2744,
       "step": 62600
     },
     {
-      "epoch": 6.743084705629103,
-      "grad_norm": 0.8293940424919128,
-      "learning_rate": 0.00019577416226699708,
-      "loss": 3.2804,
+      "epoch": 6.754716981132075,
+      "grad_norm": 0.7562722563743591,
+      "learning_rate": 0.0001950890447922288,
+      "loss": 3.2836,
       "step": 62650
     },
     {
-      "epoch": 6.748466257668712,
-      "grad_norm": 0.7844554781913757,
-      "learning_rate": 0.00019545092123693565,
-      "loss": 3.2701,
+      "epoch": 6.7601078167115904,
+      "grad_norm": 0.7537541389465332,
+      "learning_rate": 0.00019476524554776038,
+      "loss": 3.2814,
       "step": 62700
     },
     {
-      "epoch": 6.75384780970832,
-      "grad_norm": 0.7604687809944153,
-      "learning_rate": 0.00019512768020687424,
-      "loss": 3.2818,
+      "epoch": 6.765498652291106,
+      "grad_norm": 0.7441864013671875,
+      "learning_rate": 0.00019444144630329193,
+      "loss": 3.2798,
       "step": 62750
     },
     {
-      "epoch": 6.759229361747928,
-      "grad_norm": 0.8484272956848145,
-      "learning_rate": 0.0001948044391768128,
-      "loss": 3.2771,
+      "epoch": 6.77088948787062,
+      "grad_norm": 0.7457358241081238,
+      "learning_rate": 0.0001941176470588235,
+      "loss": 3.281,
       "step": 62800
     },
     {
-      "epoch": 6.7646109137875365,
-      "grad_norm": 0.7642599940299988,
-      "learning_rate": 0.00019448119814675143,
-      "loss": 3.2808,
+      "epoch": 6.776280323450135,
+      "grad_norm": 0.7565246224403381,
+      "learning_rate": 0.0001937938478143551,
+      "loss": 3.2755,
       "step": 62850
     },
     {
-      "epoch": 6.769992465827144,
-      "grad_norm": 0.7840580344200134,
-      "learning_rate": 0.00019415795711669,
-      "loss": 3.2957,
+      "epoch": 6.781671159029649,
+      "grad_norm": 0.7899838089942932,
+      "learning_rate": 0.00019347652455477602,
+      "loss": 3.3014,
       "step": 62900
     },
     {
-      "epoch": 6.775374017866753,
-      "grad_norm": 0.8181833028793335,
-      "learning_rate": 0.00019383471608662857,
-      "loss": 3.2691,
+      "epoch": 6.787061994609164,
+      "grad_norm": 0.8076132535934448,
+      "learning_rate": 0.0001931527253103076,
+      "loss": 3.2825,
       "step": 62950
     },
     {
-      "epoch": 6.780755569906361,
-      "grad_norm": 0.8215237259864807,
-      "learning_rate": 0.0001935114750565672,
-      "loss": 3.2696,
+      "epoch": 6.7924528301886795,
+      "grad_norm": 0.7913707494735718,
+      "learning_rate": 0.00019282892606583917,
+      "loss": 3.272,
       "step": 63000
     },
     {
-      "epoch": 6.780755569906361,
-      "eval_accuracy": 0.38650168232698845,
-      "eval_loss": 3.359816312789917,
-      "eval_runtime": 180.6996,
-      "eval_samples_per_second": 99.674,
-      "eval_steps_per_second": 6.231,
+      "epoch": 6.7924528301886795,
+      "eval_accuracy": 0.38650135636832506,
+      "eval_loss": 3.3610804080963135,
+      "eval_runtime": 184.6468,
+      "eval_samples_per_second": 97.543,
+      "eval_steps_per_second": 6.098,
       "step": 63000
     },
     {
-      "epoch": 6.786137121945969,
-      "grad_norm": 0.7587153911590576,
-      "learning_rate": 0.00019318823402650576,
-      "loss": 3.2695,
+      "epoch": 6.797843665768194,
+      "grad_norm": 0.7771248817443848,
+      "learning_rate": 0.00019250512682137075,
+      "loss": 3.2959,
       "step": 63050
     },
     {
-      "epoch": 6.7915186739855775,
-      "grad_norm": 0.7708876729011536,
-      "learning_rate": 0.00019286499299644432,
-      "loss": 3.2752,
+      "epoch": 6.803234501347709,
+      "grad_norm": 0.7436697483062744,
+      "learning_rate": 0.0001921813275769023,
+      "loss": 3.2888,
       "step": 63100
     },
     {
-      "epoch": 6.796900226025185,
-      "grad_norm": 0.8366973400115967,
-      "learning_rate": 0.00019254175196638292,
-      "loss": 3.293,
+      "epoch": 6.808625336927224,
+      "grad_norm": 0.7928493022918701,
+      "learning_rate": 0.00019185752833243388,
+      "loss": 3.2828,
       "step": 63150
     },
     {
-      "epoch": 6.802281778064794,
-      "grad_norm": 0.8045485019683838,
-      "learning_rate": 0.0001922185109363215,
-      "loss": 3.2756,
+      "epoch": 6.814016172506738,
+      "grad_norm": 0.7770071029663086,
+      "learning_rate": 0.00019153372908796545,
+      "loss": 3.3013,
       "step": 63200
     },
     {
-      "epoch": 6.807663330104402,
-      "grad_norm": 0.7623797059059143,
-      "learning_rate": 0.00019189526990626008,
-      "loss": 3.2853,
+      "epoch": 6.819407008086253,
+      "grad_norm": 1.0295462608337402,
+      "learning_rate": 0.000191209929843497,
+      "loss": 3.2822,
       "step": 63250
     },
     {
-      "epoch": 6.813044882144011,
-      "grad_norm": 0.7878138422966003,
-      "learning_rate": 0.00019157202887619867,
-      "loss": 3.2875,
+      "epoch": 6.824797843665769,
+      "grad_norm": 0.7837042808532715,
+      "learning_rate": 0.00019088613059902858,
+      "loss": 3.2862,
       "step": 63300
     },
     {
-      "epoch": 6.8184264341836185,
-      "grad_norm": 0.7722912430763245,
-      "learning_rate": 0.00019124878784613724,
-      "loss": 3.2808,
+      "epoch": 6.830188679245283,
+      "grad_norm": 0.7821605205535889,
+      "learning_rate": 0.00019056233135456016,
+      "loss": 3.2704,
       "step": 63350
     },
     {
-      "epoch": 6.823807986223227,
-      "grad_norm": 0.7824761271476746,
-      "learning_rate": 0.00019092554681607584,
-      "loss": 3.2744,
+      "epoch": 6.835579514824798,
+      "grad_norm": 0.7281560301780701,
+      "learning_rate": 0.0001902385321100917,
+      "loss": 3.2843,
       "step": 63400
     },
     {
-      "epoch": 6.829189538262835,
-      "grad_norm": 0.8032478094100952,
-      "learning_rate": 0.00019060230578601443,
-      "loss": 3.2716,
+      "epoch": 6.840970350404312,
+      "grad_norm": 0.7526901960372925,
+      "learning_rate": 0.0001899147328656233,
+      "loss": 3.2753,
       "step": 63450
     },
     {
-      "epoch": 6.834571090302443,
-      "grad_norm": 0.7636817097663879,
-      "learning_rate": 0.000190279064755953,
-      "loss": 3.2809,
+      "epoch": 6.846361185983827,
+      "grad_norm": 0.7778500914573669,
+      "learning_rate": 0.00018959093362115487,
+      "loss": 3.3013,
       "step": 63500
     },
     {
-      "epoch": 6.839952642342052,
-      "grad_norm": 0.8288126587867737,
-      "learning_rate": 0.00018995582372589157,
-      "loss": 3.2843,
+      "epoch": 6.8517520215633425,
+      "grad_norm": 0.7681808471679688,
+      "learning_rate": 0.00018926713437668644,
+      "loss": 3.2797,
       "step": 63550
     },
     {
-      "epoch": 6.8453341943816595,
-      "grad_norm": 0.7764488458633423,
-      "learning_rate": 0.0001896325826958302,
-      "loss": 3.2944,
+      "epoch": 6.857142857142857,
+      "grad_norm": 0.7437977194786072,
+      "learning_rate": 0.00018894333513221802,
+      "loss": 3.3045,
       "step": 63600
     },
     {
-      "epoch": 6.850715746421268,
-      "grad_norm": 0.8440558910369873,
-      "learning_rate": 0.00018930934166576876,
-      "loss": 3.282,
+      "epoch": 6.862533692722372,
+      "grad_norm": 0.7671492695808411,
+      "learning_rate": 0.00018861953588774957,
+      "loss": 3.2904,
       "step": 63650
     },
     {
-      "epoch": 6.856097298460876,
-      "grad_norm": 0.8410781025886536,
-      "learning_rate": 0.00018898610063570732,
-      "loss": 3.2914,
+      "epoch": 6.867924528301887,
+      "grad_norm": 0.843818187713623,
+      "learning_rate": 0.00018829573664328115,
+      "loss": 3.2881,
       "step": 63700
     },
     {
-      "epoch": 6.861478850500484,
-      "grad_norm": 0.7732300758361816,
-      "learning_rate": 0.00018866285960564595,
-      "loss": 3.2623,
+      "epoch": 6.873315363881401,
+      "grad_norm": 0.7092686295509338,
+      "learning_rate": 0.0001879719373988127,
+      "loss": 3.3074,
       "step": 63750
     },
     {
-      "epoch": 6.866860402540093,
-      "grad_norm": 0.7925925850868225,
-      "learning_rate": 0.0001883396185755845,
-      "loss": 3.2792,
+      "epoch": 6.878706199460916,
+      "grad_norm": 0.7547516226768494,
+      "learning_rate": 0.00018764813815434428,
+      "loss": 3.2979,
       "step": 63800
     },
     {
-      "epoch": 6.8722419545797,
-      "grad_norm": 0.7949431538581848,
-      "learning_rate": 0.0001880163775455231,
-      "loss": 3.2922,
+      "epoch": 6.884097035040432,
+      "grad_norm": 0.7473245859146118,
+      "learning_rate": 0.00018732433890987586,
+      "loss": 3.2871,
       "step": 63850
     },
     {
-      "epoch": 6.877623506619309,
-      "grad_norm": 0.7886269092559814,
-      "learning_rate": 0.00018769313651546168,
-      "loss": 3.28,
+      "epoch": 6.889487870619946,
+      "grad_norm": 0.7430441379547119,
+      "learning_rate": 0.00018700053966540743,
+      "loss": 3.282,
       "step": 63900
     },
     {
-      "epoch": 6.8830050586589175,
-      "grad_norm": 0.8459096550941467,
-      "learning_rate": 0.00018736989548540027,
-      "loss": 3.2812,
+      "epoch": 6.894878706199461,
+      "grad_norm": 0.7878106832504272,
+      "learning_rate": 0.000186676740420939,
+      "loss": 3.2906,
       "step": 63950
     },
     {
-      "epoch": 6.888386610698525,
-      "grad_norm": 0.798220694065094,
-      "learning_rate": 0.00018704665445533886,
-      "loss": 3.2895,
+      "epoch": 6.900269541778976,
+      "grad_norm": 0.7469275593757629,
+      "learning_rate": 0.0001863529411764706,
+      "loss": 3.275,
       "step": 64000
     },
     {
-      "epoch": 6.888386610698525,
-      "eval_accuracy": 0.3871914108588348,
-      "eval_loss": 3.3550844192504883,
-      "eval_runtime": 180.8293,
-      "eval_samples_per_second": 99.602,
-      "eval_steps_per_second": 6.227,
+      "epoch": 6.900269541778976,
+      "eval_accuracy": 0.38682112181716305,
+      "eval_loss": 3.3576762676239014,
+      "eval_runtime": 184.785,
+      "eval_samples_per_second": 97.47,
+      "eval_steps_per_second": 6.094,
       "step": 64000
     },
     {
-      "epoch": 6.893768162738134,
-      "grad_norm": 0.788962185382843,
-      "learning_rate": 0.00018672341342527743,
-      "loss": 3.2834,
+      "epoch": 6.90566037735849,
+      "grad_norm": 0.7187466621398926,
+      "learning_rate": 0.00018602914193200217,
+      "loss": 3.2847,
       "step": 64050
     },
     {
-      "epoch": 6.899149714777742,
-      "grad_norm": 0.7914458513259888,
-      "learning_rate": 0.000186400172395216,
-      "loss": 3.2871,
+      "epoch": 6.9110512129380055,
+      "grad_norm": 0.7366333603858948,
+      "learning_rate": 0.0001857053426875337,
+      "loss": 3.2846,
       "step": 64100
     },
     {
-      "epoch": 6.90453126681735,
-      "grad_norm": 0.8202913403511047,
-      "learning_rate": 0.00018607693136515462,
-      "loss": 3.2844,
+      "epoch": 6.916442048517521,
+      "grad_norm": 0.7928003072738647,
+      "learning_rate": 0.00018538154344306527,
+      "loss": 3.2964,
       "step": 64150
     },
     {
-      "epoch": 6.9099128188569585,
-      "grad_norm": 0.8197557330131531,
-      "learning_rate": 0.0001857536903350932,
-      "loss": 3.2812,
+      "epoch": 6.921832884097035,
+      "grad_norm": 0.7863099575042725,
+      "learning_rate": 0.00018505774419859684,
+      "loss": 3.2825,
       "step": 64200
     },
     {
-      "epoch": 6.915294370896566,
-      "grad_norm": 0.8909661769866943,
-      "learning_rate": 0.00018543044930503176,
-      "loss": 3.2823,
+      "epoch": 6.92722371967655,
+      "grad_norm": 0.7603685855865479,
+      "learning_rate": 0.00018473394495412842,
+      "loss": 3.3005,
       "step": 64250
     },
     {
-      "epoch": 6.920675922936175,
-      "grad_norm": 0.8012292385101318,
-      "learning_rate": 0.00018510720827497035,
-      "loss": 3.2831,
+      "epoch": 6.932614555256064,
+      "grad_norm": 0.7410401701927185,
+      "learning_rate": 0.00018441014570966,
+      "loss": 3.2867,
       "step": 64300
     },
     {
-      "epoch": 6.926057474975783,
-      "grad_norm": 0.8419049978256226,
-      "learning_rate": 0.00018478396724490895,
-      "loss": 3.2819,
+      "epoch": 6.938005390835579,
+      "grad_norm": 0.7407622933387756,
+      "learning_rate": 0.00018408634646519158,
+      "loss": 3.2811,
       "step": 64350
     },
     {
-      "epoch": 6.931439027015391,
-      "grad_norm": 0.8665248155593872,
-      "learning_rate": 0.0001844607262148475,
-      "loss": 3.2778,
+      "epoch": 6.943396226415095,
+      "grad_norm": 0.7628289461135864,
+      "learning_rate": 0.00018376254722072316,
+      "loss": 3.2923,
       "step": 64400
     },
     {
-      "epoch": 6.9368205790549995,
-      "grad_norm": 0.8091945052146912,
-      "learning_rate": 0.0001841374851847861,
-      "loss": 3.2969,
+      "epoch": 6.948787061994609,
+      "grad_norm": 0.747056245803833,
+      "learning_rate": 0.0001834387479762547,
+      "loss": 3.2672,
       "step": 64450
     },
     {
-      "epoch": 6.942202131094608,
-      "grad_norm": 0.7917461395263672,
-      "learning_rate": 0.00018381424415472468,
-      "loss": 3.2777,
+      "epoch": 6.954177897574124,
+      "grad_norm": 0.7753750681877136,
+      "learning_rate": 0.00018311494873178628,
+      "loss": 3.2983,
       "step": 64500
     },
     {
-      "epoch": 6.947583683134216,
-      "grad_norm": 0.8483121395111084,
-      "learning_rate": 0.00018349100312466327,
-      "loss": 3.2774,
+      "epoch": 6.959568733153639,
+      "grad_norm": 0.7228161692619324,
+      "learning_rate": 0.00018279114948731783,
+      "loss": 3.2901,
       "step": 64550
     },
     {
-      "epoch": 6.952965235173824,
-      "grad_norm": 0.8231863379478455,
-      "learning_rate": 0.00018316776209460186,
-      "loss": 3.2817,
+      "epoch": 6.964959568733153,
+      "grad_norm": 0.736755907535553,
+      "learning_rate": 0.0001824673502428494,
+      "loss": 3.2753,
       "step": 64600
     },
     {
-      "epoch": 6.958346787213433,
-      "grad_norm": 0.7838043570518494,
-      "learning_rate": 0.00018284452106454043,
-      "loss": 3.2871,
+      "epoch": 6.9703504043126685,
+      "grad_norm": 0.7366510033607483,
+      "learning_rate": 0.000182143550998381,
+      "loss": 3.2889,
       "step": 64650
     },
     {
-      "epoch": 6.9637283392530405,
-      "grad_norm": 0.8054148554801941,
-      "learning_rate": 0.000182521280034479,
-      "loss": 3.2853,
+      "epoch": 6.975741239892184,
+      "grad_norm": 0.7212927937507629,
+      "learning_rate": 0.00018181975175391257,
+      "loss": 3.2938,
       "step": 64700
     },
     {
-      "epoch": 6.969109891292649,
-      "grad_norm": 0.8586058020591736,
-      "learning_rate": 0.00018219803900441762,
-      "loss": 3.2725,
+      "epoch": 6.981132075471698,
+      "grad_norm": 0.8152526021003723,
+      "learning_rate": 0.00018149595250944414,
+      "loss": 3.2776,
       "step": 64750
     },
     {
-      "epoch": 6.974491443332257,
-      "grad_norm": 0.8905388712882996,
-      "learning_rate": 0.0001818747979743562,
-      "loss": 3.2941,
+      "epoch": 6.986522911051213,
+      "grad_norm": 0.8053279519081116,
+      "learning_rate": 0.0001811721532649757,
+      "loss": 3.2878,
       "step": 64800
     },
     {
-      "epoch": 6.979872995371865,
-      "grad_norm": 0.8023561835289001,
-      "learning_rate": 0.00018155155694429478,
-      "loss": 3.2772,
+      "epoch": 6.991913746630727,
+      "grad_norm": 0.7522391676902771,
+      "learning_rate": 0.00018084835402050727,
+      "loss": 3.2619,
       "step": 64850
     },
     {
-      "epoch": 6.985254547411474,
-      "grad_norm": 0.8267327547073364,
-      "learning_rate": 0.00018122831591423338,
-      "loss": 3.3077,
+      "epoch": 6.997304582210242,
+      "grad_norm": 0.741847574710846,
+      "learning_rate": 0.00018052455477603885,
+      "loss": 3.3074,
       "step": 64900
     },
     {
-      "epoch": 6.990636099451081,
-      "grad_norm": 0.8111370801925659,
-      "learning_rate": 0.00018090507488417195,
-      "loss": 3.2822,
+      "epoch": 7.002695417789758,
+      "grad_norm": 0.754662036895752,
+      "learning_rate": 0.0001802007555315704,
+      "loss": 3.2549,
       "step": 64950
     },
     {
-      "epoch": 6.99601765149069,
-      "grad_norm": 0.7818148732185364,
-      "learning_rate": 0.00018058829867471175,
-      "loss": 3.2834,
+      "epoch": 7.008086253369272,
+      "grad_norm": 0.7811526656150818,
+      "learning_rate": 0.00017987695628710198,
+      "loss": 3.206,
       "step": 65000
     },
     {
-      "epoch": 6.99601765149069,
-      "eval_accuracy": 0.38743957405460305,
-      "eval_loss": 3.350639581680298,
-      "eval_runtime": 180.6606,
-      "eval_samples_per_second": 99.695,
-      "eval_steps_per_second": 6.233,
+      "epoch": 7.008086253369272,
+      "eval_accuracy": 0.38698475306621166,
+      "eval_loss": 3.359518051147461,
+      "eval_runtime": 184.6567,
+      "eval_samples_per_second": 97.538,
+      "eval_steps_per_second": 6.098,
       "step": 65000
     },
     {
-      "epoch": 7.0013992035302985,
-      "grad_norm": 0.8154913783073425,
-      "learning_rate": 0.00018026505764465035,
-      "loss": 3.2506,
+      "epoch": 7.013477088948787,
+      "grad_norm": 0.7693818807601929,
+      "learning_rate": 0.00017955315704263356,
+      "loss": 3.2067,
       "step": 65050
     },
     {
-      "epoch": 7.006780755569906,
-      "grad_norm": 0.776719868183136,
-      "learning_rate": 0.00017994181661458894,
-      "loss": 3.1834,
+      "epoch": 7.018867924528302,
+      "grad_norm": 0.7586972117424011,
+      "learning_rate": 0.00017923583378305448,
+      "loss": 3.2073,
       "step": 65100
     },
     {
-      "epoch": 7.012162307609515,
-      "grad_norm": 0.7705662846565247,
-      "learning_rate": 0.0001796185755845275,
-      "loss": 3.2001,
+      "epoch": 7.024258760107816,
+      "grad_norm": 0.8107732534408569,
+      "learning_rate": 0.00017891203453858606,
+      "loss": 3.2127,
       "step": 65150
     },
     {
-      "epoch": 7.017543859649122,
-      "grad_norm": 0.8426011800765991,
-      "learning_rate": 0.00017929533455446608,
-      "loss": 3.1784,
+      "epoch": 7.0296495956873315,
+      "grad_norm": 0.7354645729064941,
+      "learning_rate": 0.00017858823529411764,
+      "loss": 3.2202,
       "step": 65200
     },
     {
-      "epoch": 7.022925411688731,
-      "grad_norm": 0.8200896978378296,
-      "learning_rate": 0.0001789720935244047,
-      "loss": 3.1967,
+      "epoch": 7.035040431266847,
+      "grad_norm": 0.7854849100112915,
+      "learning_rate": 0.00017826443604964921,
+      "loss": 3.1969,
       "step": 65250
     },
     {
-      "epoch": 7.0283069637283395,
-      "grad_norm": 0.8123102188110352,
-      "learning_rate": 0.00017864885249434327,
-      "loss": 3.201,
+      "epoch": 7.040431266846361,
+      "grad_norm": 0.7866175174713135,
+      "learning_rate": 0.00017794063680518077,
+      "loss": 3.2094,
       "step": 65300
     },
     {
-      "epoch": 7.033688515767947,
-      "grad_norm": 0.8369794487953186,
-      "learning_rate": 0.00017832561146428183,
-      "loss": 3.1928,
+      "epoch": 7.045822102425876,
+      "grad_norm": 0.744255006313324,
+      "learning_rate": 0.00017761683756071234,
+      "loss": 3.2052,
       "step": 65350
     },
     {
-      "epoch": 7.039070067807556,
-      "grad_norm": 0.7779251933097839,
-      "learning_rate": 0.00017800237043422046,
-      "loss": 3.1874,
+      "epoch": 7.051212938005391,
+      "grad_norm": 0.7620016932487488,
+      "learning_rate": 0.00017729303831624392,
+      "loss": 3.2066,
       "step": 65400
     },
     {
-      "epoch": 7.044451619847164,
-      "grad_norm": 0.8082521557807922,
-      "learning_rate": 0.00017767912940415902,
-      "loss": 3.1967,
+      "epoch": 7.056603773584905,
+      "grad_norm": 0.757378101348877,
+      "learning_rate": 0.00017696923907177547,
+      "loss": 3.1946,
       "step": 65450
     },
     {
-      "epoch": 7.049833171886772,
-      "grad_norm": 0.8156419992446899,
-      "learning_rate": 0.0001773558883740976,
-      "loss": 3.1956,
+      "epoch": 7.061994609164421,
+      "grad_norm": 0.7628493309020996,
+      "learning_rate": 0.00017664543982730705,
+      "loss": 3.2097,
       "step": 65500
     },
     {
-      "epoch": 7.0552147239263805,
-      "grad_norm": 0.8252975344657898,
-      "learning_rate": 0.00017703264734403619,
-      "loss": 3.2095,
+      "epoch": 7.067385444743936,
+      "grad_norm": 0.7638753056526184,
+      "learning_rate": 0.00017632164058283863,
+      "loss": 3.1945,
       "step": 65550
     },
     {
-      "epoch": 7.060596275965988,
-      "grad_norm": 0.7841750383377075,
-      "learning_rate": 0.00017670940631397475,
-      "loss": 3.1895,
+      "epoch": 7.07277628032345,
+      "grad_norm": 0.7687901258468628,
+      "learning_rate": 0.0001759978413383702,
+      "loss": 3.1878,
       "step": 65600
     },
     {
-      "epoch": 7.065977828005597,
-      "grad_norm": 0.8294128179550171,
-      "learning_rate": 0.00017638616528391337,
-      "loss": 3.212,
+      "epoch": 7.078167115902965,
+      "grad_norm": 0.7791463136672974,
+      "learning_rate": 0.00017567404209390178,
+      "loss": 3.2067,
       "step": 65650
     },
     {
-      "epoch": 7.071359380045205,
-      "grad_norm": 0.8671795129776001,
-      "learning_rate": 0.00017606292425385194,
-      "loss": 3.2094,
+      "epoch": 7.083557951482479,
+      "grad_norm": 0.7624858617782593,
+      "learning_rate": 0.00017535024284943333,
+      "loss": 3.2304,
       "step": 65700
     },
     {
-      "epoch": 7.076740932084813,
-      "grad_norm": 0.7949682474136353,
-      "learning_rate": 0.0001757396832237905,
-      "loss": 3.2189,
+      "epoch": 7.0889487870619945,
+      "grad_norm": 0.7333658933639526,
+      "learning_rate": 0.00017502644360496488,
+      "loss": 3.2159,
       "step": 65750
     },
     {
-      "epoch": 7.0821224841244215,
-      "grad_norm": 0.8465704917907715,
-      "learning_rate": 0.00017541644219372913,
-      "loss": 3.2139,
+      "epoch": 7.09433962264151,
+      "grad_norm": 0.7534026503562927,
+      "learning_rate": 0.00017470264436049646,
+      "loss": 3.2068,
       "step": 65800
     },
     {
-      "epoch": 7.08750403616403,
-      "grad_norm": 0.8210230469703674,
-      "learning_rate": 0.0001750932011636677,
-      "loss": 3.2113,
+      "epoch": 7.099730458221024,
+      "grad_norm": 0.7843572497367859,
+      "learning_rate": 0.00017437884511602804,
+      "loss": 3.2185,
       "step": 65850
     },
     {
-      "epoch": 7.092885588203638,
-      "grad_norm": 0.8009158372879028,
-      "learning_rate": 0.00017476996013360627,
-      "loss": 3.2013,
+      "epoch": 7.105121293800539,
+      "grad_norm": 0.7714036107063293,
+      "learning_rate": 0.00017405504587155962,
+      "loss": 3.2306,
       "step": 65900
     },
     {
-      "epoch": 7.098267140243246,
-      "grad_norm": 0.8548141717910767,
-      "learning_rate": 0.00017444671910354486,
-      "loss": 3.2163,
+      "epoch": 7.110512129380054,
+      "grad_norm": 0.7692075967788696,
+      "learning_rate": 0.0001737312466270912,
+      "loss": 3.2265,
       "step": 65950
     },
     {
-      "epoch": 7.103648692282855,
-      "grad_norm": 0.8232263922691345,
-      "learning_rate": 0.00017412347807348346,
-      "loss": 3.2163,
+      "epoch": 7.115902964959568,
+      "grad_norm": 0.7614142298698425,
+      "learning_rate": 0.00017340744738262277,
+      "loss": 3.2005,
       "step": 66000
     },
     {
-      "epoch": 7.103648692282855,
-      "eval_accuracy": 0.3872667073100902,
-      "eval_loss": 3.3580427169799805,
-      "eval_runtime": 180.5751,
-      "eval_samples_per_second": 99.742,
-      "eval_steps_per_second": 6.236,
+      "epoch": 7.115902964959568,
+      "eval_accuracy": 0.38687142810422115,
+      "eval_loss": 3.361088275909424,
+      "eval_runtime": 184.9846,
+      "eval_samples_per_second": 97.365,
+      "eval_steps_per_second": 6.087,
       "step": 66000
     },
     {
-      "epoch": 7.109030244322462,
-      "grad_norm": 0.7871417999267578,
-      "learning_rate": 0.00017380023704342202,
-      "loss": 3.217,
+      "epoch": 7.121293800539084,
+      "grad_norm": 0.7808488011360168,
+      "learning_rate": 0.00017308364813815435,
+      "loss": 3.2523,
       "step": 66050
     },
     {
-      "epoch": 7.114411796362071,
-      "grad_norm": 0.8225372433662415,
-      "learning_rate": 0.00017347699601336062,
-      "loss": 3.2348,
+      "epoch": 7.126684636118599,
+      "grad_norm": 0.7770368456840515,
+      "learning_rate": 0.00017275984889368593,
+      "loss": 3.2059,
       "step": 66100
     },
     {
-      "epoch": 7.119793348401679,
-      "grad_norm": 0.8037530779838562,
-      "learning_rate": 0.00017315375498329919,
-      "loss": 3.2072,
+      "epoch": 7.132075471698113,
+      "grad_norm": 0.8269479274749756,
+      "learning_rate": 0.00017243604964921745,
+      "loss": 3.2103,
       "step": 66150
     },
     {
-      "epoch": 7.125174900441287,
-      "grad_norm": 1.7943295240402222,
-      "learning_rate": 0.00017283051395323778,
-      "loss": 3.2096,
+      "epoch": 7.137466307277628,
+      "grad_norm": 0.8035179972648621,
+      "learning_rate": 0.00017211225040474903,
+      "loss": 3.2091,
       "step": 66200
     },
     {
-      "epoch": 7.130556452480896,
-      "grad_norm": 0.8043410181999207,
-      "learning_rate": 0.00017250727292317638,
-      "loss": 3.2008,
+      "epoch": 7.142857142857143,
+      "grad_norm": 0.8429258465766907,
+      "learning_rate": 0.0001717884511602806,
+      "loss": 3.2233,
       "step": 66250
     },
     {
-      "epoch": 7.135938004520503,
-      "grad_norm": 0.8701448440551758,
-      "learning_rate": 0.00017218403189311494,
-      "loss": 3.2097,
+      "epoch": 7.1482479784366575,
+      "grad_norm": 0.7574069499969482,
+      "learning_rate": 0.00017146465191581218,
+      "loss": 3.2282,
       "step": 66300
     },
     {
-      "epoch": 7.141319556560112,
-      "grad_norm": 0.8197288513183594,
-      "learning_rate": 0.0001718607908630535,
-      "loss": 3.1995,
+      "epoch": 7.153638814016173,
+      "grad_norm": 0.7994669675827026,
+      "learning_rate": 0.00017114085267134376,
+      "loss": 3.2247,
       "step": 66350
     },
     {
-      "epoch": 7.1467011085997205,
-      "grad_norm": 0.8066554665565491,
-      "learning_rate": 0.00017153754983299213,
-      "loss": 3.2064,
+      "epoch": 7.159029649595688,
+      "grad_norm": 0.7484458684921265,
+      "learning_rate": 0.00017081705342687534,
+      "loss": 3.2324,
       "step": 66400
     },
     {
-      "epoch": 7.152082660639328,
-      "grad_norm": 0.7715657353401184,
-      "learning_rate": 0.0001712143088029307,
-      "loss": 3.2109,
+      "epoch": 7.164420485175202,
+      "grad_norm": 0.7799018025398254,
+      "learning_rate": 0.0001704932541824069,
+      "loss": 3.2306,
       "step": 66450
     },
     {
-      "epoch": 7.157464212678937,
-      "grad_norm": 0.8049308061599731,
-      "learning_rate": 0.00017089106777286927,
-      "loss": 3.2056,
+      "epoch": 7.169811320754717,
+      "grad_norm": 0.7528384327888489,
+      "learning_rate": 0.00017016945493793847,
+      "loss": 3.2158,
       "step": 66500
     },
     {
-      "epoch": 7.162845764718545,
-      "grad_norm": 0.8044945001602173,
-      "learning_rate": 0.0001705678267428079,
-      "loss": 3.2161,
+      "epoch": 7.175202156334231,
+      "grad_norm": 0.7923641204833984,
+      "learning_rate": 0.00016984565569347002,
+      "loss": 3.2163,
       "step": 66550
     },
     {
-      "epoch": 7.168227316758153,
-      "grad_norm": 0.8120958805084229,
-      "learning_rate": 0.00017024458571274646,
-      "loss": 3.2157,
+      "epoch": 7.180592991913747,
+      "grad_norm": 0.7415347099304199,
+      "learning_rate": 0.0001695218564490016,
+      "loss": 3.2183,
       "step": 66600
     },
     {
-      "epoch": 7.1736088687977615,
-      "grad_norm": 0.8076058030128479,
-      "learning_rate": 0.00016992134468268505,
-      "loss": 3.2389,
+      "epoch": 7.185983827493262,
+      "grad_norm": 0.7724004983901978,
+      "learning_rate": 0.00016919805720453317,
+      "loss": 3.2205,
       "step": 66650
     },
     {
-      "epoch": 7.178990420837369,
-      "grad_norm": 0.8148431181907654,
-      "learning_rate": 0.00016959810365262362,
-      "loss": 3.2187,
+      "epoch": 7.191374663072776,
+      "grad_norm": 0.813212513923645,
+      "learning_rate": 0.00016887425796006475,
+      "loss": 3.2178,
       "step": 66700
     },
     {
-      "epoch": 7.184371972876978,
-      "grad_norm": 0.8567858338356018,
-      "learning_rate": 0.0001692748626225622,
-      "loss": 3.2188,
+      "epoch": 7.196765498652291,
+      "grad_norm": 0.7927876114845276,
+      "learning_rate": 0.00016855045871559633,
+      "loss": 3.2103,
       "step": 66750
     },
     {
-      "epoch": 7.189753524916586,
-      "grad_norm": 0.8196995854377747,
-      "learning_rate": 0.0001689516215925008,
-      "loss": 3.2203,
+      "epoch": 7.202156334231806,
+      "grad_norm": 0.8068609833717346,
+      "learning_rate": 0.00016822665947112788,
+      "loss": 3.2318,
       "step": 66800
     },
     {
-      "epoch": 7.195135076956194,
-      "grad_norm": 0.8134042620658875,
-      "learning_rate": 0.00016862838056243938,
-      "loss": 3.2023,
+      "epoch": 7.2075471698113205,
+      "grad_norm": 0.8054190278053284,
+      "learning_rate": 0.00016790286022665946,
+      "loss": 3.2371,
       "step": 66850
     },
     {
-      "epoch": 7.2005166289958025,
-      "grad_norm": 0.8222635984420776,
-      "learning_rate": 0.00016830513953237794,
-      "loss": 3.2136,
+      "epoch": 7.212938005390836,
+      "grad_norm": 0.7957893013954163,
+      "learning_rate": 0.00016757906098219103,
+      "loss": 3.2164,
       "step": 66900
     },
     {
-      "epoch": 7.205898181035411,
-      "grad_norm": 0.8523157835006714,
-      "learning_rate": 0.00016798189850231657,
-      "loss": 3.2292,
+      "epoch": 7.218328840970351,
+      "grad_norm": 0.7549981474876404,
+      "learning_rate": 0.0001672552617377226,
+      "loss": 3.2372,
       "step": 66950
     },
     {
-      "epoch": 7.211279733075019,
-      "grad_norm": 0.8117284178733826,
-      "learning_rate": 0.00016765865747225513,
-      "loss": 3.2243,
+      "epoch": 7.223719676549865,
+      "grad_norm": 0.731869101524353,
+      "learning_rate": 0.00016693146249325416,
+      "loss": 3.2249,
       "step": 67000
     },
     {
-      "epoch": 7.211279733075019,
-      "eval_accuracy": 0.38720662226312885,
-      "eval_loss": 3.3582334518432617,
-      "eval_runtime": 180.7195,
-      "eval_samples_per_second": 99.663,
-      "eval_steps_per_second": 6.231,
+      "epoch": 7.223719676549865,
+      "eval_accuracy": 0.38759386115529965,
+      "eval_loss": 3.36012864112854,
+      "eval_runtime": 184.9495,
+      "eval_samples_per_second": 97.383,
+      "eval_steps_per_second": 6.088,
       "step": 67000
     },
     {
-      "epoch": 7.216661285114627,
-      "grad_norm": 0.849255383014679,
-      "learning_rate": 0.0001673354164421937,
-      "loss": 3.2088,
+      "epoch": 7.22911051212938,
+      "grad_norm": 0.9114392399787903,
+      "learning_rate": 0.00016660766324878574,
+      "loss": 3.2494,
       "step": 67050
     },
     {
-      "epoch": 7.222042837154235,
-      "grad_norm": 0.8112828135490417,
-      "learning_rate": 0.00016701217541213232,
-      "loss": 3.2221,
+      "epoch": 7.234501347708895,
+      "grad_norm": 0.7874780893325806,
+      "learning_rate": 0.0001662838640043173,
+      "loss": 3.2246,
       "step": 67100
     },
     {
-      "epoch": 7.2274243891938434,
-      "grad_norm": 0.8703665733337402,
-      "learning_rate": 0.0001666889343820709,
-      "loss": 3.2114,
+      "epoch": 7.2398921832884096,
+      "grad_norm": 0.833976686000824,
+      "learning_rate": 0.00016596006475984887,
+      "loss": 3.2383,
       "step": 67150
     },
     {
-      "epoch": 7.232805941233452,
-      "grad_norm": 0.7933388352394104,
-      "learning_rate": 0.00016636569335200946,
+      "epoch": 7.245283018867925,
+      "grad_norm": 0.8088477849960327,
+      "learning_rate": 0.00016563626551538045,
       "loss": 3.2238,
       "step": 67200
     },
     {
-      "epoch": 7.23818749327306,
-      "grad_norm": 0.8398449420928955,
-      "learning_rate": 0.00016604245232194805,
-      "loss": 3.2391,
+      "epoch": 7.250673854447439,
+      "grad_norm": 0.7701541781425476,
+      "learning_rate": 0.00016531246627091202,
+      "loss": 3.2332,
       "step": 67250
     },
     {
-      "epoch": 7.243569045312668,
-      "grad_norm": 0.8231996297836304,
-      "learning_rate": 0.00016571921129188665,
-      "loss": 3.2372,
+      "epoch": 7.256064690026954,
+      "grad_norm": 0.7686187624931335,
+      "learning_rate": 0.0001649886670264436,
+      "loss": 3.2307,
       "step": 67300
     },
     {
-      "epoch": 7.248950597352277,
-      "grad_norm": 0.8363311886787415,
-      "learning_rate": 0.00016539597026182521,
-      "loss": 3.2078,
+      "epoch": 7.261455525606469,
+      "grad_norm": 0.7967571020126343,
+      "learning_rate": 0.00016466486778197518,
+      "loss": 3.2253,
       "step": 67350
     },
     {
-      "epoch": 7.254332149391884,
-      "grad_norm": 0.8070518970489502,
-      "learning_rate": 0.0001650727292317638,
-      "loss": 3.2155,
+      "epoch": 7.2668463611859835,
+      "grad_norm": 0.8138226866722107,
+      "learning_rate": 0.0001643410685375067,
+      "loss": 3.2298,
       "step": 67400
     },
     {
-      "epoch": 7.259713701431493,
-      "grad_norm": 0.8043374419212341,
-      "learning_rate": 0.00016474948820170238,
-      "loss": 3.229,
+      "epoch": 7.272237196765499,
+      "grad_norm": 0.7992923855781555,
+      "learning_rate": 0.00016401726929303828,
+      "loss": 3.2251,
       "step": 67450
     },
     {
-      "epoch": 7.265095253471101,
-      "grad_norm": 0.8224925398826599,
-      "learning_rate": 0.00016442624717164094,
-      "loss": 3.2222,
+      "epoch": 7.277628032345014,
+      "grad_norm": 0.7716299891471863,
+      "learning_rate": 0.00016369347004856986,
+      "loss": 3.2362,
       "step": 67500
     },
     {
-      "epoch": 7.270476805510709,
-      "grad_norm": 0.8394501209259033,
-      "learning_rate": 0.00016410300614157957,
-      "loss": 3.2113,
+      "epoch": 7.283018867924528,
+      "grad_norm": 0.8016715049743652,
+      "learning_rate": 0.00016336967080410143,
+      "loss": 3.2404,
       "step": 67550
     },
     {
-      "epoch": 7.275858357550318,
-      "grad_norm": 0.8679465055465698,
-      "learning_rate": 0.00016377976511151813,
-      "loss": 3.2274,
+      "epoch": 7.288409703504043,
+      "grad_norm": 0.7413058876991272,
+      "learning_rate": 0.000163045871559633,
+      "loss": 3.2129,
       "step": 67600
     },
     {
-      "epoch": 7.281239909589925,
-      "grad_norm": 0.8839657306671143,
-      "learning_rate": 0.00016345652408145675,
-      "loss": 3.2073,
+      "epoch": 7.293800539083558,
+      "grad_norm": 0.7324450612068176,
+      "learning_rate": 0.0001627220723151646,
+      "loss": 3.2346,
       "step": 67650
     },
     {
-      "epoch": 7.286621461629534,
-      "grad_norm": 0.9003349542617798,
-      "learning_rate": 0.00016313328305139532,
-      "loss": 3.2276,
+      "epoch": 7.2991913746630726,
+      "grad_norm": 0.7739547491073608,
+      "learning_rate": 0.00016239827307069617,
+      "loss": 3.2217,
       "step": 67700
     },
     {
-      "epoch": 7.2920030136691425,
-      "grad_norm": 0.8355953097343445,
-      "learning_rate": 0.0001628100420213339,
-      "loss": 3.229,
+      "epoch": 7.304582210242588,
+      "grad_norm": 0.7586296796798706,
+      "learning_rate": 0.00016207447382622775,
+      "loss": 3.2313,
       "step": 67750
     },
     {
-      "epoch": 7.29738456570875,
-      "grad_norm": 0.8046534657478333,
-      "learning_rate": 0.00016248680099127248,
-      "loss": 3.2386,
+      "epoch": 7.309973045822103,
+      "grad_norm": 0.7944618463516235,
+      "learning_rate": 0.00016175067458175932,
+      "loss": 3.2225,
       "step": 67800
     },
     {
-      "epoch": 7.302766117748359,
-      "grad_norm": 0.8797202110290527,
-      "learning_rate": 0.00016216355996121105,
-      "loss": 3.2217,
+      "epoch": 7.315363881401617,
+      "grad_norm": 0.7308440208435059,
+      "learning_rate": 0.00016142687533729085,
+      "loss": 3.2313,
       "step": 67850
     },
     {
-      "epoch": 7.308147669787967,
-      "grad_norm": 0.9474117755889893,
-      "learning_rate": 0.00016184031893114965,
-      "loss": 3.2168,
+      "epoch": 7.320754716981132,
+      "grad_norm": 0.7774273157119751,
+      "learning_rate": 0.00016110307609282242,
+      "loss": 3.2173,
       "step": 67900
     },
     {
-      "epoch": 7.313529221827575,
-      "grad_norm": 0.8707678318023682,
-      "learning_rate": 0.00016151707790108824,
-      "loss": 3.2345,
+      "epoch": 7.3261455525606465,
+      "grad_norm": 0.7571539878845215,
+      "learning_rate": 0.000160779276848354,
+      "loss": 3.22,
       "step": 67950
     },
     {
-      "epoch": 7.3189107738671835,
-      "grad_norm": 0.8038205504417419,
-      "learning_rate": 0.0001611938368710268,
-      "loss": 3.222,
+      "epoch": 7.331536388140162,
+      "grad_norm": 0.8019688725471497,
+      "learning_rate": 0.00016045547760388558,
+      "loss": 3.2262,
       "step": 68000
     },
     {
-      "epoch": 7.3189107738671835,
-      "eval_accuracy": 0.387976102014631,
-      "eval_loss": 3.352640390396118,
-      "eval_runtime": 180.5346,
-      "eval_samples_per_second": 99.765,
-      "eval_steps_per_second": 6.237,
+      "epoch": 7.331536388140162,
+      "eval_accuracy": 0.38772761286019924,
+      "eval_loss": 3.3541133403778076,
+      "eval_runtime": 184.6341,
+      "eval_samples_per_second": 97.55,
+      "eval_steps_per_second": 6.099,
       "step": 68000
     },
     {
-      "epoch": 7.324292325906791,
-      "grad_norm": 0.8763945698738098,
-      "learning_rate": 0.00016087059584096538,
-      "loss": 3.2126,
+      "epoch": 7.336927223719677,
+      "grad_norm": 0.8166679739952087,
+      "learning_rate": 0.00016013167835941716,
+      "loss": 3.2332,
       "step": 68050
     },
     {
-      "epoch": 7.3296738779464,
-      "grad_norm": 0.8443440198898315,
-      "learning_rate": 0.000160547354810904,
-      "loss": 3.2391,
+      "epoch": 7.342318059299191,
+      "grad_norm": 0.7608354687690735,
+      "learning_rate": 0.00015980787911494873,
+      "loss": 3.2595,
       "step": 68100
     },
     {
-      "epoch": 7.335055429986008,
-      "grad_norm": 0.8131680488586426,
-      "learning_rate": 0.00016023057860144378,
-      "loss": 3.2449,
+      "epoch": 7.347708894878706,
+      "grad_norm": 0.8633112907409668,
+      "learning_rate": 0.00015948407987048029,
+      "loss": 3.2232,
       "step": 68150
     },
     {
-      "epoch": 7.340436982025616,
-      "grad_norm": 0.8589426279067993,
-      "learning_rate": 0.0001599073375713824,
-      "loss": 3.2291,
+      "epoch": 7.353099730458221,
+      "grad_norm": 0.7595582008361816,
+      "learning_rate": 0.00015916028062601186,
+      "loss": 3.221,
       "step": 68200
     },
     {
-      "epoch": 7.3458185340652244,
-      "grad_norm": 0.8745540380477905,
-      "learning_rate": 0.00015958409654132097,
-      "loss": 3.2265,
+      "epoch": 7.3584905660377355,
+      "grad_norm": 0.8150188326835632,
+      "learning_rate": 0.00015883648138154344,
+      "loss": 3.25,
       "step": 68250
     },
     {
-      "epoch": 7.351200086104833,
-      "grad_norm": 0.838469386100769,
-      "learning_rate": 0.00015926085551125953,
-      "loss": 3.2466,
+      "epoch": 7.363881401617251,
+      "grad_norm": 0.829832136631012,
+      "learning_rate": 0.000158512682137075,
+      "loss": 3.2272,
       "step": 68300
     },
     {
-      "epoch": 7.356581638144441,
-      "grad_norm": 0.8351563811302185,
-      "learning_rate": 0.00015893761448119813,
-      "loss": 3.2189,
+      "epoch": 7.369272237196766,
+      "grad_norm": 0.7878278493881226,
+      "learning_rate": 0.00015818888289260657,
+      "loss": 3.2386,
       "step": 68350
     },
     {
-      "epoch": 7.361963190184049,
-      "grad_norm": 0.8260790705680847,
-      "learning_rate": 0.00015861437345113672,
-      "loss": 3.2165,
+      "epoch": 7.37466307277628,
+      "grad_norm": 0.7915233373641968,
+      "learning_rate": 0.00015786508364813815,
+      "loss": 3.2403,
       "step": 68400
     },
     {
-      "epoch": 7.367344742223658,
-      "grad_norm": 0.8453501462936401,
-      "learning_rate": 0.00015829759724167653,
-      "loss": 3.2279,
+      "epoch": 7.380053908355795,
+      "grad_norm": 0.7962988615036011,
+      "learning_rate": 0.0001575412844036697,
+      "loss": 3.2236,
       "step": 68450
     },
     {
-      "epoch": 7.372726294263265,
-      "grad_norm": 0.8862749934196472,
-      "learning_rate": 0.0001579743562116151,
-      "loss": 3.2315,
+      "epoch": 7.38544474393531,
+      "grad_norm": 0.7690643072128296,
+      "learning_rate": 0.00015722396114409065,
+      "loss": 3.2106,
       "step": 68500
     },
     {
-      "epoch": 7.378107846302874,
-      "grad_norm": 0.7972759008407593,
-      "learning_rate": 0.00015765111518155372,
-      "loss": 3.2317,
+      "epoch": 7.390835579514825,
+      "grad_norm": 0.7795893549919128,
+      "learning_rate": 0.00015690016189962223,
+      "loss": 3.2353,
       "step": 68550
     },
     {
-      "epoch": 7.383489398342482,
-      "grad_norm": 0.8577588796615601,
-      "learning_rate": 0.0001573278741514923,
-      "loss": 3.2332,
+      "epoch": 7.39622641509434,
+      "grad_norm": 0.8003530502319336,
+      "learning_rate": 0.00015657636265515378,
+      "loss": 3.2169,
       "step": 68600
     },
     {
-      "epoch": 7.38887095038209,
-      "grad_norm": 0.8090537190437317,
-      "learning_rate": 0.00015700463312143085,
-      "loss": 3.2207,
+      "epoch": 7.401617250673855,
+      "grad_norm": 0.8012276291847229,
+      "learning_rate": 0.00015625256341068536,
+      "loss": 3.2393,
       "step": 68650
     },
     {
-      "epoch": 7.394252502421699,
-      "grad_norm": 0.8941483497619629,
-      "learning_rate": 0.00015668139209136945,
-      "loss": 3.2325,
+      "epoch": 7.407008086253369,
+      "grad_norm": 0.768429160118103,
+      "learning_rate": 0.00015592876416621693,
+      "loss": 3.2154,
       "step": 68700
     },
     {
-      "epoch": 7.399634054461306,
-      "grad_norm": 0.8350239396095276,
-      "learning_rate": 0.00015635815106130804,
-      "loss": 3.2347,
+      "epoch": 7.412398921832884,
+      "grad_norm": 0.8270565271377563,
+      "learning_rate": 0.0001556049649217485,
+      "loss": 3.2256,
       "step": 68750
     },
     {
-      "epoch": 7.405015606500915,
-      "grad_norm": 0.834892749786377,
-      "learning_rate": 0.0001560349100312466,
-      "loss": 3.2314,
+      "epoch": 7.4177897574123985,
+      "grad_norm": 0.7837132215499878,
+      "learning_rate": 0.00015528116567728006,
+      "loss": 3.2431,
       "step": 68800
     },
     {
-      "epoch": 7.4103971585405235,
-      "grad_norm": 0.8487468957901001,
-      "learning_rate": 0.0001557116690011852,
-      "loss": 3.2407,
+      "epoch": 7.423180592991914,
+      "grad_norm": 0.7685801386833191,
+      "learning_rate": 0.00015495736643281164,
+      "loss": 3.2363,
       "step": 68850
     },
     {
-      "epoch": 7.415778710580131,
-      "grad_norm": 0.8758741021156311,
-      "learning_rate": 0.00015538842797112377,
-      "loss": 3.2153,
+      "epoch": 7.428571428571429,
+      "grad_norm": 0.8626431822776794,
+      "learning_rate": 0.00015463356718834322,
+      "loss": 3.2411,
       "step": 68900
     },
     {
-      "epoch": 7.42116026261974,
-      "grad_norm": 0.879677951335907,
-      "learning_rate": 0.00015506518694106237,
-      "loss": 3.2249,
+      "epoch": 7.433962264150943,
+      "grad_norm": 0.7880833148956299,
+      "learning_rate": 0.0001543097679438748,
+      "loss": 3.2314,
       "step": 68950
     },
     {
-      "epoch": 7.426541814659347,
-      "grad_norm": 0.8545063138008118,
-      "learning_rate": 0.00015474194591100096,
-      "loss": 3.2349,
+      "epoch": 7.439353099730458,
+      "grad_norm": 0.7299560308456421,
+      "learning_rate": 0.00015398596869940637,
+      "loss": 3.2216,
       "step": 69000
     },
     {
-      "epoch": 7.426541814659347,
-      "eval_accuracy": 0.3882562091594167,
-      "eval_loss": 3.3482072353363037,
-      "eval_runtime": 180.8861,
-      "eval_samples_per_second": 99.571,
-      "eval_steps_per_second": 6.225,
+      "epoch": 7.439353099730458,
+      "eval_accuracy": 0.38783952533464816,
+      "eval_loss": 3.351468324661255,
+      "eval_runtime": 185.0392,
+      "eval_samples_per_second": 97.336,
+      "eval_steps_per_second": 6.085,
       "step": 69000
     },
     {
-      "epoch": 7.431923366698956,
-      "grad_norm": 0.8772327303886414,
-      "learning_rate": 0.00015441870488093953,
-      "loss": 3.2284,
+      "epoch": 7.444743935309973,
+      "grad_norm": 0.7486545443534851,
+      "learning_rate": 0.00015366216945493792,
+      "loss": 3.2187,
       "step": 69050
     },
     {
-      "epoch": 7.4373049187385645,
-      "grad_norm": 0.9024936556816101,
-      "learning_rate": 0.0001540954638508781,
-      "loss": 3.218,
+      "epoch": 7.450134770889488,
+      "grad_norm": 0.7718955278396606,
+      "learning_rate": 0.00015333837021046947,
+      "loss": 3.2249,
       "step": 69100
     },
     {
-      "epoch": 7.442686470778172,
-      "grad_norm": 0.8099319338798523,
-      "learning_rate": 0.00015377222282081672,
-      "loss": 3.2183,
+      "epoch": 7.455525606469003,
+      "grad_norm": 0.860206127166748,
+      "learning_rate": 0.00015301457096600105,
+      "loss": 3.2364,
       "step": 69150
     },
     {
-      "epoch": 7.448068022817781,
-      "grad_norm": 0.793427050113678,
-      "learning_rate": 0.0001534489817907553,
-      "loss": 3.2307,
+      "epoch": 7.460916442048518,
+      "grad_norm": 0.928758978843689,
+      "learning_rate": 0.00015269077172153263,
+      "loss": 3.2494,
       "step": 69200
     },
     {
-      "epoch": 7.453449574857389,
-      "grad_norm": 0.8657283186912537,
-      "learning_rate": 0.00015312574076069388,
-      "loss": 3.2321,
+      "epoch": 7.466307277628032,
+      "grad_norm": 0.8024472594261169,
+      "learning_rate": 0.0001523669724770642,
+      "loss": 3.227,
       "step": 69250
     },
     {
-      "epoch": 7.458831126896997,
-      "grad_norm": 0.8640968799591064,
-      "learning_rate": 0.00015280249973063248,
-      "loss": 3.2242,
+      "epoch": 7.471698113207547,
+      "grad_norm": 0.8709169030189514,
+      "learning_rate": 0.00015204317323259578,
+      "loss": 3.256,
       "step": 69300
     },
     {
-      "epoch": 7.4642126789366054,
-      "grad_norm": 0.8989497423171997,
-      "learning_rate": 0.00015247925870057104,
-      "loss": 3.2395,
+      "epoch": 7.4770889487870615,
+      "grad_norm": 0.8366961479187012,
+      "learning_rate": 0.00015171937398812736,
+      "loss": 3.2388,
       "step": 69350
     },
     {
-      "epoch": 7.469594230976213,
-      "grad_norm": 0.8708150386810303,
-      "learning_rate": 0.00015215601767050964,
-      "loss": 3.2317,
+      "epoch": 7.482479784366577,
+      "grad_norm": 0.8059794306755066,
+      "learning_rate": 0.00015139557474365894,
+      "loss": 3.2311,
       "step": 69400
     },
     {
-      "epoch": 7.474975783015822,
-      "grad_norm": 0.8274281024932861,
-      "learning_rate": 0.0001518327766404482,
-      "loss": 3.2348,
+      "epoch": 7.487870619946092,
+      "grad_norm": 0.7974139451980591,
+      "learning_rate": 0.00015107177549919046,
+      "loss": 3.2324,
       "step": 69450
     },
     {
-      "epoch": 7.48035733505543,
-      "grad_norm": 0.841109573841095,
-      "learning_rate": 0.0001515095356103868,
-      "loss": 3.2263,
+      "epoch": 7.493261455525606,
+      "grad_norm": 0.8359848856925964,
+      "learning_rate": 0.00015074797625472204,
+      "loss": 3.2453,
       "step": 69500
     },
     {
-      "epoch": 7.485738887095038,
-      "grad_norm": 0.8123463988304138,
-      "learning_rate": 0.0001511862945803254,
-      "loss": 3.2116,
+      "epoch": 7.498652291105121,
+      "grad_norm": 0.7960279583930969,
+      "learning_rate": 0.00015042417701025362,
+      "loss": 3.2229,
       "step": 69550
     },
     {
-      "epoch": 7.491120439134646,
-      "grad_norm": 0.8069713115692139,
-      "learning_rate": 0.00015086305355026396,
-      "loss": 3.201,
+      "epoch": 7.504043126684636,
+      "grad_norm": 0.826728105545044,
+      "learning_rate": 0.0001501003777657852,
+      "loss": 3.2342,
       "step": 69600
     },
     {
-      "epoch": 7.496501991174255,
-      "grad_norm": 0.8293700218200684,
-      "learning_rate": 0.00015053981252020253,
-      "loss": 3.2439,
+      "epoch": 7.509433962264151,
+      "grad_norm": 0.8121812343597412,
+      "learning_rate": 0.00014977657852131677,
+      "loss": 3.249,
       "step": 69650
     },
     {
-      "epoch": 7.501883543213863,
-      "grad_norm": 0.8899844884872437,
-      "learning_rate": 0.00015021657149014115,
-      "loss": 3.2287,
+      "epoch": 7.514824797843666,
+      "grad_norm": 0.78941410779953,
+      "learning_rate": 0.00014945277927684835,
+      "loss": 3.2428,
       "step": 69700
     },
     {
-      "epoch": 7.507265095253471,
-      "grad_norm": 0.8449057936668396,
-      "learning_rate": 0.00014989333046007972,
-      "loss": 3.2416,
+      "epoch": 7.520215633423181,
+      "grad_norm": 0.7744514346122742,
+      "learning_rate": 0.00014912898003237993,
+      "loss": 3.2616,
       "step": 69750
     },
     {
-      "epoch": 7.51264664729308,
-      "grad_norm": 0.8326901197433472,
-      "learning_rate": 0.00014957008943001832,
-      "loss": 3.2332,
+      "epoch": 7.525606469002695,
+      "grad_norm": 0.8503783941268921,
+      "learning_rate": 0.00014880518078791148,
+      "loss": 3.2482,
       "step": 69800
     },
     {
-      "epoch": 7.518028199332687,
-      "grad_norm": 0.8048696517944336,
-      "learning_rate": 0.00014924684839995688,
-      "loss": 3.2153,
+      "epoch": 7.53099730458221,
+      "grad_norm": 0.7720679044723511,
+      "learning_rate": 0.00014848138154344306,
+      "loss": 3.2262,
       "step": 69850
     },
     {
-      "epoch": 7.523409751372296,
-      "grad_norm": 0.8521336913108826,
-      "learning_rate": 0.00014892360736989548,
-      "loss": 3.2298,
+      "epoch": 7.536388140161725,
+      "grad_norm": 0.7836202383041382,
+      "learning_rate": 0.00014815758229897463,
+      "loss": 3.216,
       "step": 69900
     },
     {
-      "epoch": 7.528791303411904,
-      "grad_norm": 0.8536204695701599,
-      "learning_rate": 0.00014860036633983407,
-      "loss": 3.2237,
+      "epoch": 7.54177897574124,
+      "grad_norm": 0.8227617144584656,
+      "learning_rate": 0.0001478337830545062,
+      "loss": 3.2513,
       "step": 69950
     },
     {
-      "epoch": 7.534172855451512,
-      "grad_norm": 0.8090736865997314,
-      "learning_rate": 0.00014827712530977264,
-      "loss": 3.2364,
+      "epoch": 7.547169811320755,
+      "grad_norm": 0.8092595934867859,
+      "learning_rate": 0.00014750998381003776,
+      "loss": 3.2305,
       "step": 70000
     },
     {
-      "epoch": 7.534172855451512,
-      "eval_accuracy": 0.38876535659171535,
-      "eval_loss": 3.3461689949035645,
-      "eval_runtime": 180.6754,
-      "eval_samples_per_second": 99.687,
-      "eval_steps_per_second": 6.232,
+      "epoch": 7.547169811320755,
+      "eval_accuracy": 0.38827544072055986,
+      "eval_loss": 3.347477912902832,
+      "eval_runtime": 184.9341,
+      "eval_samples_per_second": 97.391,
+      "eval_steps_per_second": 6.089,
       "step": 70000
     }
   ],
   "logging_steps": 50,
-  "max_steps": 92910,
+  "max_steps": 92750,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 10,
   "save_steps": 10000,
@@ -10456,7 +10456,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 5.85279519326208e+17,
+  "total_flos": 5.852484255744e+17,
   "train_batch_size": 32,
   "trial_name": null,
   "trial_params": null