diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,5 +1,5 @@
 {
-  "best_metric": 3.3056087493896484,
+  "best_metric": 3.3006131649017334,
   "best_model_checkpoint": "/scratch/cl5625/exceptions/models/100M_495/checkpoint-90000",
   "epoch": 10.0,
   "eval_steps": 1000,
@@ -10,13846 +10,13846 @@
   "log_history": [
     {
       "epoch": 0.005381552039608223,
-      "grad_norm": 1.0683914422988892,
+      "grad_norm": 6.547796249389648,
       "learning_rate": 0.0003,
-      "loss": 8.6798,
+      "loss": 8.6491,
       "step": 50
     },
     {
       "epoch": 0.010763104079216447,
-      "grad_norm": 2.859095573425293,
+      "grad_norm": 5.357585906982422,
       "learning_rate": 0.0006,
-      "loss": 6.9757,
+      "loss": 6.9125,
       "step": 100
     },
     {
       "epoch": 0.01614465611882467,
-      "grad_norm": 1.489780068397522,
+      "grad_norm": 2.4043188095092773,
       "learning_rate": 0.0005996767589699385,
-      "loss": 6.479,
+      "loss": 6.5144,
       "step": 150
     },
     {
       "epoch": 0.021526208158432893,
-      "grad_norm": 1.3649728298187256,
+      "grad_norm": 1.361128330230713,
       "learning_rate": 0.0005993535179398771,
-      "loss": 6.1969,
+      "loss": 6.2176,
       "step": 200
     },
     {
       "epoch": 0.026907760198041114,
-      "grad_norm": 1.613872766494751,
+      "grad_norm": 1.5913020372390747,
       "learning_rate": 0.0005990302769098158,
-      "loss": 6.0747,
+      "loss": 6.0702,
       "step": 250
     },
     {
       "epoch": 0.03228931223764934,
-      "grad_norm": 3.6575379371643066,
+      "grad_norm": 1.3549662828445435,
       "learning_rate": 0.0005987070358797543,
-      "loss": 5.9726,
+      "loss": 5.983,
       "step": 300
     },
     {
       "epoch": 0.03767086427725756,
-      "grad_norm": 1.7698675394058228,
+      "grad_norm": 1.7182697057724,
       "learning_rate": 0.0005983837948496929,
-      "loss": 5.8947,
+      "loss": 5.9004,
       "step": 350
     },
     {
       "epoch": 0.04305241631686579,
-      "grad_norm": 1.8536632061004639,
+      "grad_norm": 0.9793596863746643,
       "learning_rate": 0.0005980605538196314,
-      "loss": 5.7945,
+      "loss": 5.8284,
       "step": 400
     },
     {
       "epoch": 0.048433968356474004,
-      "grad_norm": 1.3679734468460083,
+      "grad_norm": 1.1408363580703735,
       "learning_rate": 0.0005977373127895701,
-      "loss": 5.721,
+      "loss": 5.7264,
       "step": 450
     },
     {
       "epoch": 0.05381552039608223,
-      "grad_norm": 1.759971022605896,
+      "grad_norm": 1.4277925491333008,
       "learning_rate": 0.0005974140717595086,
-      "loss": 5.6392,
+      "loss": 5.6398,
       "step": 500
     },
     {
       "epoch": 0.05919707243569045,
-      "grad_norm": 1.350098729133606,
+      "grad_norm": 2.2396180629730225,
       "learning_rate": 0.0005970908307294472,
-      "loss": 5.5834,
+      "loss": 5.574,
       "step": 550
     },
     {
       "epoch": 0.06457862447529868,
-      "grad_norm": 1.1349716186523438,
+      "grad_norm": 1.143998622894287,
       "learning_rate": 0.0005967675896993858,
-      "loss": 5.5375,
+      "loss": 5.5309,
       "step": 600
     },
     {
       "epoch": 0.0699601765149069,
-      "grad_norm": 1.276620626449585,
+      "grad_norm": 1.3793718814849854,
       "learning_rate": 0.0005964443486693243,
-      "loss": 5.4386,
+      "loss": 5.4287,
       "step": 650
     },
     {
       "epoch": 0.07534172855451512,
-      "grad_norm": 0.9349568486213684,
+      "grad_norm": 0.9985382556915283,
       "learning_rate": 0.000596121107639263,
-      "loss": 5.389,
+      "loss": 5.3816,
       "step": 700
     },
     {
       "epoch": 0.08072328059412334,
-      "grad_norm": 1.2208337783813477,
+      "grad_norm": 1.0711445808410645,
       "learning_rate": 0.0005957978666092015,
-      "loss": 5.347,
+      "loss": 5.3184,
       "step": 750
     },
     {
       "epoch": 0.08610483263373157,
-      "grad_norm": 0.9988003969192505,
+      "grad_norm": 1.4245736598968506,
       "learning_rate": 0.0005954746255791401,
-      "loss": 5.2746,
+      "loss": 5.2668,
       "step": 800
     },
     {
       "epoch": 0.09148638467333979,
-      "grad_norm": 1.0252668857574463,
+      "grad_norm": 1.3065918684005737,
       "learning_rate": 0.0005951513845490787,
-      "loss": 5.2421,
+      "loss": 5.2295,
       "step": 850
     },
     {
       "epoch": 0.09686793671294801,
-      "grad_norm": 1.0475060939788818,
+      "grad_norm": 1.2478379011154175,
       "learning_rate": 0.0005948281435190174,
-      "loss": 5.1758,
+      "loss": 5.1634,
       "step": 900
     },
     {
       "epoch": 0.10224948875255624,
-      "grad_norm": 1.1841458082199097,
+      "grad_norm": 0.9259421229362488,
       "learning_rate": 0.0005945049024889559,
-      "loss": 5.1527,
+      "loss": 5.1204,
       "step": 950
     },
     {
       "epoch": 0.10763104079216446,
-      "grad_norm": 1.3012635707855225,
+      "grad_norm": 1.41254460811615,
       "learning_rate": 0.0005941816614588944,
-      "loss": 5.1178,
+      "loss": 5.0846,
       "step": 1000
     },
     {
       "epoch": 0.10763104079216446,
-      "eval_accuracy": 0.2259316197396655,
-      "eval_loss": 5.038772106170654,
-      "eval_runtime": 185.0,
-      "eval_samples_per_second": 97.357,
-      "eval_steps_per_second": 6.086,
+      "eval_accuracy": 0.22689776121811187,
+      "eval_loss": 5.023209095001221,
+      "eval_runtime": 183.3261,
+      "eval_samples_per_second": 98.246,
+      "eval_steps_per_second": 6.142,
       "step": 1000
     },
     {
       "epoch": 0.11301259283177269,
-      "grad_norm": 1.3463271856307983,
+      "grad_norm": 1.1088929176330566,
       "learning_rate": 0.000593858420428833,
-      "loss": 5.0694,
+      "loss": 5.0586,
       "step": 1050
     },
     {
       "epoch": 0.1183941448713809,
-      "grad_norm": 1.3660123348236084,
+      "grad_norm": 1.2266565561294556,
       "learning_rate": 0.0005935351793987716,
-      "loss": 5.0413,
+      "loss": 5.0344,
       "step": 1100
     },
     {
       "epoch": 0.12377569691098914,
-      "grad_norm": 1.0985352993011475,
+      "grad_norm": 1.1834657192230225,
       "learning_rate": 0.0005932119383687103,
-      "loss": 5.0437,
+      "loss": 5.0128,
       "step": 1150
     },
     {
       "epoch": 0.12915724895059735,
-      "grad_norm": 1.175125241279602,
+      "grad_norm": 1.5358068943023682,
       "learning_rate": 0.0005928886973386488,
-      "loss": 4.9567,
+      "loss": 4.9492,
       "step": 1200
     },
     {
       "epoch": 0.13453880099020557,
-      "grad_norm": 1.1065387725830078,
+      "grad_norm": 0.8692229390144348,
       "learning_rate": 0.0005925654563085874,
-      "loss": 4.9404,
+      "loss": 4.9173,
       "step": 1250
     },
     {
       "epoch": 0.1399203530298138,
-      "grad_norm": 2.134150505065918,
+      "grad_norm": 0.8401411175727844,
       "learning_rate": 0.000592242215278526,
-      "loss": 4.8978,
+      "loss": 4.8852,
       "step": 1300
     },
     {
       "epoch": 0.14530190506942203,
-      "grad_norm": 0.826282262802124,
+      "grad_norm": 1.0361953973770142,
       "learning_rate": 0.0005919189742484645,
-      "loss": 4.8955,
+      "loss": 4.8926,
       "step": 1350
     },
     {
       "epoch": 0.15068345710903025,
-      "grad_norm": 1.1777522563934326,
+      "grad_norm": 0.9142393469810486,
       "learning_rate": 0.0005915957332184032,
-      "loss": 4.8565,
+      "loss": 4.8369,
       "step": 1400
     },
     {
       "epoch": 0.15606500914863847,
-      "grad_norm": 1.0407218933105469,
+      "grad_norm": 1.1336075067520142,
       "learning_rate": 0.0005912724921883417,
-      "loss": 4.8394,
+      "loss": 4.8169,
       "step": 1450
     },
     {
       "epoch": 0.16144656118824668,
-      "grad_norm": 1.1135185956954956,
+      "grad_norm": 1.0692909955978394,
       "learning_rate": 0.0005909492511582803,
-      "loss": 4.8083,
+      "loss": 4.8005,
       "step": 1500
     },
     {
       "epoch": 0.1668281132278549,
-      "grad_norm": 0.9728663563728333,
+      "grad_norm": 0.807711124420166,
       "learning_rate": 0.0005906260101282189,
-      "loss": 4.7959,
+      "loss": 4.7786,
       "step": 1550
     },
     {
       "epoch": 0.17220966526746315,
-      "grad_norm": 0.8725541234016418,
+      "grad_norm": 1.2913539409637451,
       "learning_rate": 0.0005903027690981575,
-      "loss": 4.7704,
+      "loss": 4.7333,
       "step": 1600
     },
     {
       "epoch": 0.17759121730707136,
-      "grad_norm": 0.8491561412811279,
+      "grad_norm": 1.1676956415176392,
       "learning_rate": 0.000589979528068096,
-      "loss": 4.7693,
+      "loss": 4.7357,
       "step": 1650
     },
     {
       "epoch": 0.18297276934667958,
-      "grad_norm": 1.0023645162582397,
+      "grad_norm": 0.9552448987960815,
       "learning_rate": 0.0005896562870380347,
-      "loss": 4.7243,
+      "loss": 4.7186,
       "step": 1700
     },
     {
       "epoch": 0.1883543213862878,
-      "grad_norm": 1.2298816442489624,
+      "grad_norm": 0.9495003819465637,
       "learning_rate": 0.0005893330460079732,
-      "loss": 4.7076,
+      "loss": 4.7037,
       "step": 1750
     },
     {
       "epoch": 0.19373587342589602,
-      "grad_norm": 1.063509225845337,
+      "grad_norm": 0.7337987422943115,
       "learning_rate": 0.0005890098049779118,
-      "loss": 4.6721,
+      "loss": 4.6579,
       "step": 1800
     },
     {
       "epoch": 0.19911742546550426,
-      "grad_norm": 1.0815796852111816,
+      "grad_norm": 1.1409964561462402,
       "learning_rate": 0.0005886865639478504,
-      "loss": 4.6689,
+      "loss": 4.6561,
       "step": 1850
     },
     {
       "epoch": 0.20449897750511248,
-      "grad_norm": 1.0041590929031372,
+      "grad_norm": 0.9477244019508362,
       "learning_rate": 0.0005883633229177889,
-      "loss": 4.664,
+      "loss": 4.6466,
       "step": 1900
     },
     {
       "epoch": 0.2098805295447207,
-      "grad_norm": 0.741370439529419,
+      "grad_norm": 0.7901616096496582,
       "learning_rate": 0.0005880400818877276,
-      "loss": 4.647,
+      "loss": 4.6147,
       "step": 1950
     },
     {
       "epoch": 0.2152620815843289,
-      "grad_norm": 0.9399243593215942,
+      "grad_norm": 0.9213060736656189,
       "learning_rate": 0.0005877168408576662,
-      "loss": 4.5945,
+      "loss": 4.5929,
       "step": 2000
     },
     {
       "epoch": 0.2152620815843289,
-      "eval_accuracy": 0.2670903115197812,
-      "eval_loss": 4.538747310638428,
-      "eval_runtime": 182.7156,
-      "eval_samples_per_second": 98.574,
-      "eval_steps_per_second": 6.163,
+      "eval_accuracy": 0.2689594671488537,
+      "eval_loss": 4.525809288024902,
+      "eval_runtime": 181.8026,
+      "eval_samples_per_second": 99.069,
+      "eval_steps_per_second": 6.194,
       "step": 2000
     },
     {
       "epoch": 0.22064363362393713,
-      "grad_norm": 0.8416916131973267,
+      "grad_norm": 0.9790673851966858,
       "learning_rate": 0.0005873935998276048,
-      "loss": 4.5951,
+      "loss": 4.5608,
       "step": 2050
     },
     {
       "epoch": 0.22602518566354537,
-      "grad_norm": 1.0621873140335083,
+      "grad_norm": 0.6953191161155701,
       "learning_rate": 0.0005870703587975433,
-      "loss": 4.5883,
+      "loss": 4.5888,
       "step": 2100
     },
     {
       "epoch": 0.2314067377031536,
-      "grad_norm": 0.958995521068573,
+      "grad_norm": 0.8427320718765259,
       "learning_rate": 0.0005867471177674818,
-      "loss": 4.564,
+      "loss": 4.548,
       "step": 2150
     },
     {
       "epoch": 0.2367882897427618,
-      "grad_norm": 1.0263909101486206,
+      "grad_norm": 1.2038559913635254,
       "learning_rate": 0.0005864238767374205,
-      "loss": 4.5361,
+      "loss": 4.5136,
       "step": 2200
     },
     {
       "epoch": 0.24216984178237003,
-      "grad_norm": 0.865015983581543,
+      "grad_norm": 0.8505734205245972,
       "learning_rate": 0.0005861006357073591,
-      "loss": 4.5177,
+      "loss": 4.5178,
       "step": 2250
     },
     {
       "epoch": 0.24755139382197827,
-      "grad_norm": 0.8319127559661865,
+      "grad_norm": 0.7332435846328735,
       "learning_rate": 0.0005857773946772977,
-      "loss": 4.5171,
+      "loss": 4.5166,
       "step": 2300
     },
     {
       "epoch": 0.2529329458615865,
-      "grad_norm": 0.8533033132553101,
+      "grad_norm": 0.9289398193359375,
       "learning_rate": 0.0005854541536472362,
-      "loss": 4.4716,
+      "loss": 4.4692,
       "step": 2350
     },
     {
       "epoch": 0.2583144979011947,
-      "grad_norm": 0.9105812907218933,
+      "grad_norm": 0.7615149617195129,
       "learning_rate": 0.0005851309126171749,
-      "loss": 4.4832,
+      "loss": 4.4647,
       "step": 2400
     },
     {
       "epoch": 0.2636960499408029,
-      "grad_norm": 0.9672813415527344,
+      "grad_norm": 1.141018033027649,
       "learning_rate": 0.0005848076715871134,
-      "loss": 4.4371,
+      "loss": 4.4292,
       "step": 2450
     },
     {
       "epoch": 0.26907760198041114,
-      "grad_norm": 0.9679467082023621,
+      "grad_norm": 0.8744039535522461,
       "learning_rate": 0.000584484430557052,
-      "loss": 4.4541,
+      "loss": 4.4525,
       "step": 2500
     },
     {
       "epoch": 0.27445915402001936,
-      "grad_norm": 0.8314814567565918,
+      "grad_norm": 0.7360879778862,
       "learning_rate": 0.0005841611895269906,
-      "loss": 4.4395,
+      "loss": 4.4223,
       "step": 2550
     },
     {
       "epoch": 0.2798407060596276,
-      "grad_norm": 0.7759853601455688,
+      "grad_norm": 0.6995299458503723,
       "learning_rate": 0.0005838379484969291,
-      "loss": 4.4153,
+      "loss": 4.4034,
       "step": 2600
     },
     {
       "epoch": 0.2852222580992358,
-      "grad_norm": 0.7185021638870239,
+      "grad_norm": 0.8300929665565491,
       "learning_rate": 0.0005835147074668678,
-      "loss": 4.3998,
+      "loss": 4.4013,
       "step": 2650
     },
     {
       "epoch": 0.29060381013884407,
-      "grad_norm": 0.7233301997184753,
+      "grad_norm": 0.8035259246826172,
       "learning_rate": 0.0005831914664368063,
-      "loss": 4.3864,
+      "loss": 4.3722,
       "step": 2700
     },
     {
       "epoch": 0.2959853621784523,
-      "grad_norm": 0.768659234046936,
+      "grad_norm": 0.8101586699485779,
       "learning_rate": 0.0005828682254067449,
-      "loss": 4.3805,
+      "loss": 4.3752,
       "step": 2750
     },
     {
       "epoch": 0.3013669142180605,
-      "grad_norm": 0.7125174403190613,
+      "grad_norm": 0.9427844285964966,
       "learning_rate": 0.0005825449843766835,
-      "loss": 4.3937,
+      "loss": 4.3694,
       "step": 2800
     },
     {
       "epoch": 0.3067484662576687,
-      "grad_norm": 0.8199774026870728,
+      "grad_norm": 0.7412756085395813,
       "learning_rate": 0.0005822217433466221,
-      "loss": 4.366,
+      "loss": 4.3536,
       "step": 2850
     },
     {
       "epoch": 0.31213001829727693,
-      "grad_norm": 0.7131398916244507,
+      "grad_norm": 0.7558133602142334,
       "learning_rate": 0.0005818985023165607,
-      "loss": 4.3393,
+      "loss": 4.3376,
       "step": 2900
     },
     {
       "epoch": 0.31751157033688515,
-      "grad_norm": 0.8093417286872864,
+      "grad_norm": 0.8248081803321838,
       "learning_rate": 0.0005815752612864992,
-      "loss": 4.3397,
+      "loss": 4.3552,
       "step": 2950
     },
     {
       "epoch": 0.32289312237649337,
-      "grad_norm": 0.9575026631355286,
+      "grad_norm": 0.7637624740600586,
       "learning_rate": 0.0005812520202564378,
-      "loss": 4.3222,
+      "loss": 4.3137,
       "step": 3000
     },
     {
       "epoch": 0.32289312237649337,
-      "eval_accuracy": 0.297169233717903,
-      "eval_loss": 4.248993873596191,
-      "eval_runtime": 182.6748,
-      "eval_samples_per_second": 98.596,
-      "eval_steps_per_second": 6.164,
+      "eval_accuracy": 0.29744379956541017,
+      "eval_loss": 4.242817401885986,
+      "eval_runtime": 181.6642,
+      "eval_samples_per_second": 99.144,
+      "eval_steps_per_second": 6.198,
       "step": 3000
     },
     {
       "epoch": 0.3282746744161016,
-      "grad_norm": 0.8685798645019531,
+      "grad_norm": 0.7341495156288147,
       "learning_rate": 0.0005809287792263764,
-      "loss": 4.3126,
+      "loss": 4.3128,
       "step": 3050
     },
     {
       "epoch": 0.3336562264557098,
-      "grad_norm": 0.8055946826934814,
+      "grad_norm": 0.7144924998283386,
       "learning_rate": 0.0005806055381963151,
-      "loss": 4.303,
+      "loss": 4.293,
       "step": 3100
     },
     {
       "epoch": 0.3390377784953181,
-      "grad_norm": 0.8136595487594604,
+      "grad_norm": 0.7501051425933838,
       "learning_rate": 0.0005802822971662536,
-      "loss": 4.2861,
+      "loss": 4.2715,
       "step": 3150
     },
     {
       "epoch": 0.3444193305349263,
-      "grad_norm": 0.6919870972633362,
+      "grad_norm": 0.6868528723716736,
       "learning_rate": 0.0005799590561361922,
-      "loss": 4.3128,
+      "loss": 4.2802,
       "step": 3200
     },
     {
       "epoch": 0.3498008825745345,
-      "grad_norm": 1.4158786535263062,
+      "grad_norm": 0.8597863912582397,
       "learning_rate": 0.0005796358151061307,
-      "loss": 4.3008,
+      "loss": 4.2787,
       "step": 3250
     },
     {
       "epoch": 0.35518243461414273,
-      "grad_norm": 0.8841201663017273,
+      "grad_norm": 0.7811806201934814,
       "learning_rate": 0.0005793125740760694,
-      "loss": 4.2924,
+      "loss": 4.2557,
       "step": 3300
     },
     {
       "epoch": 0.36056398665375095,
-      "grad_norm": 0.6999953985214233,
+      "grad_norm": 0.641840398311615,
       "learning_rate": 0.0005789893330460079,
-      "loss": 4.2491,
+      "loss": 4.2406,
       "step": 3350
     },
     {
       "epoch": 0.36594553869335916,
-      "grad_norm": 0.8567237257957458,
+      "grad_norm": 0.860731840133667,
       "learning_rate": 0.0005786660920159465,
-      "loss": 4.2506,
+      "loss": 4.2436,
       "step": 3400
     },
     {
       "epoch": 0.3713270907329674,
-      "grad_norm": 0.779323160648346,
+      "grad_norm": 0.8146591186523438,
       "learning_rate": 0.0005783428509858851,
-      "loss": 4.2543,
+      "loss": 4.2515,
       "step": 3450
     },
     {
       "epoch": 0.3767086427725756,
-      "grad_norm": 0.7086902260780334,
+      "grad_norm": 0.7766924500465393,
       "learning_rate": 0.0005780196099558237,
-      "loss": 4.235,
+      "loss": 4.2245,
       "step": 3500
     },
     {
       "epoch": 0.3820901948121838,
-      "grad_norm": 0.5936145186424255,
+      "grad_norm": 0.6820563077926636,
       "learning_rate": 0.0005776963689257623,
-      "loss": 4.2331,
+      "loss": 4.2161,
       "step": 3550
     },
     {
       "epoch": 0.38747174685179203,
-      "grad_norm": 0.632709264755249,
+      "grad_norm": 0.7251272797584534,
       "learning_rate": 0.0005773731278957008,
-      "loss": 4.2431,
+      "loss": 4.2113,
       "step": 3600
     },
     {
       "epoch": 0.3928532988914003,
-      "grad_norm": 0.9441436529159546,
+      "grad_norm": 0.7628197073936462,
       "learning_rate": 0.0005770498868656394,
-      "loss": 4.2204,
+      "loss": 4.2133,
       "step": 3650
     },
     {
       "epoch": 0.3982348509310085,
-      "grad_norm": 0.8455289602279663,
+      "grad_norm": 0.7888264656066895,
       "learning_rate": 0.000576726645835578,
-      "loss": 4.2151,
+      "loss": 4.1951,
       "step": 3700
     },
     {
       "epoch": 0.40361640297061674,
-      "grad_norm": 0.6691498756408691,
+      "grad_norm": 0.7610321640968323,
       "learning_rate": 0.0005764034048055167,
-      "loss": 4.1907,
+      "loss": 4.2038,
       "step": 3750
     },
     {
       "epoch": 0.40899795501022496,
-      "grad_norm": 0.764274001121521,
+      "grad_norm": 0.7846929430961609,
       "learning_rate": 0.0005760801637754552,
-      "loss": 4.1993,
+      "loss": 4.1858,
       "step": 3800
     },
     {
       "epoch": 0.4143795070498332,
-      "grad_norm": 0.8627861142158508,
+      "grad_norm": 0.7305687665939331,
       "learning_rate": 0.0005757569227453937,
-      "loss": 4.2007,
+      "loss": 4.1908,
       "step": 3850
     },
     {
       "epoch": 0.4197610590894414,
-      "grad_norm": 0.7309780120849609,
+      "grad_norm": 0.8204297423362732,
       "learning_rate": 0.0005754336817153324,
-      "loss": 4.1748,
+      "loss": 4.1887,
       "step": 3900
     },
     {
       "epoch": 0.4251426111290496,
-      "grad_norm": 1.0626062154769897,
+      "grad_norm": 1.0344388484954834,
       "learning_rate": 0.0005751104406852709,
-      "loss": 4.1841,
+      "loss": 4.1589,
       "step": 3950
     },
     {
       "epoch": 0.4305241631686578,
-      "grad_norm": 0.7397871017456055,
+      "grad_norm": 0.7527312636375427,
       "learning_rate": 0.0005747871996552096,
-      "loss": 4.1793,
+      "loss": 4.1807,
       "step": 4000
     },
     {
       "epoch": 0.4305241631686578,
-      "eval_accuracy": 0.31092153838146963,
-      "eval_loss": 4.1014580726623535,
-      "eval_runtime": 183.3521,
-      "eval_samples_per_second": 98.232,
-      "eval_steps_per_second": 6.141,
+      "eval_accuracy": 0.31194559184912113,
+      "eval_loss": 4.095859050750732,
+      "eval_runtime": 181.6682,
+      "eval_samples_per_second": 99.142,
+      "eval_steps_per_second": 6.198,
       "step": 4000
     },
     {
       "epoch": 0.43590571520826604,
-      "grad_norm": 0.7458042502403259,
+      "grad_norm": 0.7800742387771606,
       "learning_rate": 0.0005744639586251481,
-      "loss": 4.1751,
+      "loss": 4.1596,
       "step": 4050
     },
     {
       "epoch": 0.44128726724787426,
-      "grad_norm": 0.6242780685424805,
+      "grad_norm": 0.7501474618911743,
       "learning_rate": 0.0005741407175950867,
-      "loss": 4.1522,
+      "loss": 4.1408,
       "step": 4100
     },
     {
       "epoch": 0.44666881928748253,
-      "grad_norm": 0.8113278746604919,
+      "grad_norm": 0.7201575040817261,
       "learning_rate": 0.0005738174765650253,
-      "loss": 4.1662,
+      "loss": 4.1604,
       "step": 4150
     },
     {
       "epoch": 0.45205037132709075,
-      "grad_norm": 0.7726860046386719,
+      "grad_norm": 0.5777507424354553,
       "learning_rate": 0.0005734942355349638,
-      "loss": 4.1576,
+      "loss": 4.1428,
       "step": 4200
     },
     {
       "epoch": 0.45743192336669897,
-      "grad_norm": 0.5663923621177673,
+      "grad_norm": 0.7474133372306824,
       "learning_rate": 0.0005731709945049025,
-      "loss": 4.1365,
+      "loss": 4.1435,
       "step": 4250
     },
     {
       "epoch": 0.4628134754063072,
-      "grad_norm": 0.9014385938644409,
+      "grad_norm": 0.665733814239502,
       "learning_rate": 0.000572847753474841,
-      "loss": 4.1356,
+      "loss": 4.1553,
       "step": 4300
     },
     {
       "epoch": 0.4681950274459154,
-      "grad_norm": 0.7335566282272339,
+      "grad_norm": 0.7177342176437378,
       "learning_rate": 0.0005725245124447796,
-      "loss": 4.1453,
+      "loss": 4.1161,
       "step": 4350
     },
     {
       "epoch": 0.4735765794855236,
-      "grad_norm": 0.779979944229126,
+      "grad_norm": 0.7497685551643372,
       "learning_rate": 0.0005722012714147182,
-      "loss": 4.1228,
+      "loss": 4.1077,
       "step": 4400
     },
     {
       "epoch": 0.47895813152513184,
-      "grad_norm": 0.8542835116386414,
+      "grad_norm": 0.7238962054252625,
       "learning_rate": 0.0005718780303846568,
-      "loss": 4.1002,
+      "loss": 4.0956,
       "step": 4450
     },
     {
       "epoch": 0.48433968356474005,
-      "grad_norm": 0.6748495697975159,
+      "grad_norm": 0.754342794418335,
       "learning_rate": 0.0005715547893545953,
-      "loss": 4.1213,
+      "loss": 4.1189,
       "step": 4500
     },
     {
       "epoch": 0.48972123560434827,
-      "grad_norm": 0.7007385492324829,
+      "grad_norm": 0.641211986541748,
       "learning_rate": 0.000571231548324534,
-      "loss": 4.1179,
+      "loss": 4.1009,
       "step": 4550
     },
     {
       "epoch": 0.49510278764395654,
-      "grad_norm": 0.6953123807907104,
+      "grad_norm": 0.902698278427124,
       "learning_rate": 0.0005709083072944725,
-      "loss": 4.0866,
+      "loss": 4.0739,
       "step": 4600
     },
     {
       "epoch": 0.5004843396835648,
-      "grad_norm": 0.6920002102851868,
+      "grad_norm": 0.6036228537559509,
       "learning_rate": 0.0005705850662644111,
-      "loss": 4.1128,
+      "loss": 4.0997,
       "step": 4650
     },
     {
       "epoch": 0.505865891723173,
-      "grad_norm": 0.6578840613365173,
+      "grad_norm": 0.6570672988891602,
       "learning_rate": 0.0005702618252343497,
-      "loss": 4.1059,
+      "loss": 4.0842,
       "step": 4700
     },
     {
       "epoch": 0.5112474437627812,
-      "grad_norm": 0.7448212504386902,
+      "grad_norm": 0.733204185962677,
       "learning_rate": 0.0005699385842042882,
-      "loss": 4.1112,
+      "loss": 4.0988,
       "step": 4750
     },
     {
       "epoch": 0.5166289958023894,
-      "grad_norm": 0.6418218612670898,
+      "grad_norm": 0.6296361088752747,
       "learning_rate": 0.0005696153431742269,
-      "loss": 4.083,
+      "loss": 4.0862,
       "step": 4800
     },
     {
       "epoch": 0.5220105478419976,
-      "grad_norm": 0.5668521523475647,
+      "grad_norm": 0.6667340993881226,
       "learning_rate": 0.0005692921021441655,
-      "loss": 4.092,
+      "loss": 4.0714,
       "step": 4850
     },
     {
       "epoch": 0.5273920998816058,
-      "grad_norm": 0.6103580594062805,
+      "grad_norm": 0.6166004538536072,
       "learning_rate": 0.0005689688611141041,
-      "loss": 4.0851,
+      "loss": 4.073,
       "step": 4900
     },
     {
       "epoch": 0.5327736519212141,
-      "grad_norm": 0.691785454750061,
+      "grad_norm": 0.6546381711959839,
       "learning_rate": 0.0005686456200840426,
-      "loss": 4.0865,
+      "loss": 4.0827,
       "step": 4950
     },
     {
       "epoch": 0.5381552039608223,
-      "grad_norm": 0.725322425365448,
+      "grad_norm": 0.6323681473731995,
       "learning_rate": 0.0005683223790539811,
-      "loss": 4.0646,
+      "loss": 4.0535,
       "step": 5000
     },
     {
       "epoch": 0.5381552039608223,
-      "eval_accuracy": 0.3200998824267101,
-      "eval_loss": 4.004289627075195,
-      "eval_runtime": 183.8369,
-      "eval_samples_per_second": 97.973,
-      "eval_steps_per_second": 6.125,
+      "eval_accuracy": 0.3216463389789736,
+      "eval_loss": 3.9916508197784424,
+      "eval_runtime": 181.8071,
+      "eval_samples_per_second": 99.067,
+      "eval_steps_per_second": 6.193,
       "step": 5000
     },
     {
       "epoch": 0.5435367560004305,
-      "grad_norm": 0.7255427837371826,
+      "grad_norm": 0.6683503985404968,
       "learning_rate": 0.0005679991380239198,
-      "loss": 4.0629,
+      "loss": 4.0387,
       "step": 5050
     },
     {
       "epoch": 0.5489183080400387,
-      "grad_norm": 0.8185265064239502,
+      "grad_norm": 0.6250083446502686,
       "learning_rate": 0.0005676758969938584,
-      "loss": 4.0571,
+      "loss": 4.0595,
       "step": 5100
     },
     {
       "epoch": 0.5542998600796469,
-      "grad_norm": 0.6164122223854065,
+      "grad_norm": 0.6272826194763184,
       "learning_rate": 0.000567352655963797,
-      "loss": 4.0833,
+      "loss": 4.0677,
       "step": 5150
     },
     {
       "epoch": 0.5596814121192552,
-      "grad_norm": 0.7035121917724609,
+      "grad_norm": 0.6267666220664978,
       "learning_rate": 0.0005670294149337355,
-      "loss": 4.0478,
+      "loss": 4.0499,
       "step": 5200
     },
     {
       "epoch": 0.5650629641588634,
-      "grad_norm": 0.6536286473274231,
+      "grad_norm": 0.6394859552383423,
       "learning_rate": 0.0005667061739036742,
-      "loss": 4.0478,
+      "loss": 4.0364,
       "step": 5250
     },
     {
       "epoch": 0.5704445161984716,
-      "grad_norm": 0.6543012261390686,
+      "grad_norm": 0.6560896039009094,
       "learning_rate": 0.0005663829328736127,
-      "loss": 4.0573,
+      "loss": 4.0384,
       "step": 5300
     },
     {
       "epoch": 0.5758260682380799,
-      "grad_norm": 0.7880141139030457,
+      "grad_norm": 0.6860178709030151,
       "learning_rate": 0.0005660596918435512,
-      "loss": 4.037,
+      "loss": 4.0365,
       "step": 5350
     },
     {
       "epoch": 0.5812076202776881,
-      "grad_norm": 0.6059675216674805,
+      "grad_norm": 0.6861039400100708,
       "learning_rate": 0.0005657364508134899,
-      "loss": 4.0243,
+      "loss": 4.033,
       "step": 5400
     },
     {
       "epoch": 0.5865891723172963,
-      "grad_norm": 0.5372035503387451,
+      "grad_norm": 0.6428167819976807,
       "learning_rate": 0.0005654132097834284,
-      "loss": 4.044,
+      "loss": 4.0161,
       "step": 5450
     },
     {
       "epoch": 0.5919707243569046,
-      "grad_norm": 0.652932345867157,
+      "grad_norm": 0.6751614212989807,
       "learning_rate": 0.0005650899687533671,
-      "loss": 4.0212,
+      "loss": 4.0278,
       "step": 5500
     },
     {
       "epoch": 0.5973522763965128,
-      "grad_norm": 0.827506422996521,
+      "grad_norm": 0.6935444474220276,
       "learning_rate": 0.0005647667277233056,
-      "loss": 4.0347,
+      "loss": 4.0194,
       "step": 5550
     },
     {
       "epoch": 0.602733828436121,
-      "grad_norm": 0.6992992758750916,
+      "grad_norm": 0.700750470161438,
       "learning_rate": 0.0005644434866932442,
-      "loss": 4.0602,
+      "loss": 4.028,
       "step": 5600
     },
     {
       "epoch": 0.6081153804757292,
-      "grad_norm": 0.6638625860214233,
+      "grad_norm": 0.641048789024353,
       "learning_rate": 0.0005641202456631828,
-      "loss": 4.0149,
+      "loss": 4.0257,
       "step": 5650
     },
     {
       "epoch": 0.6134969325153374,
-      "grad_norm": 0.6855315566062927,
+      "grad_norm": 0.6311193108558655,
       "learning_rate": 0.0005637970046331214,
-      "loss": 3.9967,
+      "loss": 4.005,
       "step": 5700
     },
     {
       "epoch": 0.6188784845549457,
-      "grad_norm": 0.594383955001831,
+      "grad_norm": 0.7258860468864441,
       "learning_rate": 0.00056347376360306,
-      "loss": 4.0143,
+      "loss": 4.0176,
       "step": 5750
     },
     {
       "epoch": 0.6242600365945539,
-      "grad_norm": 1.1790200471878052,
+      "grad_norm": 0.8383513689041138,
       "learning_rate": 0.0005631505225729985,
-      "loss": 4.0077,
+      "loss": 4.0076,
       "step": 5800
     },
     {
       "epoch": 0.6296415886341621,
-      "grad_norm": 0.6079861521720886,
+      "grad_norm": 0.5887354612350464,
       "learning_rate": 0.0005628272815429371,
-      "loss": 4.0057,
+      "loss": 3.9825,
       "step": 5850
     },
     {
       "epoch": 0.6350231406737703,
-      "grad_norm": 0.6142423748970032,
+      "grad_norm": 0.6457569599151611,
       "learning_rate": 0.0005625040405128757,
-      "loss": 4.0097,
+      "loss": 3.9889,
       "step": 5900
     },
     {
       "epoch": 0.6404046927133785,
-      "grad_norm": 0.6306520700454712,
+      "grad_norm": 0.6390018463134766,
       "learning_rate": 0.0005621807994828143,
-      "loss": 4.0023,
+      "loss": 3.9978,
       "step": 5950
     },
     {
       "epoch": 0.6457862447529867,
-      "grad_norm": 0.5871993899345398,
+      "grad_norm": 0.6403166651725769,
       "learning_rate": 0.0005618575584527529,
-      "loss": 4.0013,
+      "loss": 3.9888,
       "step": 6000
     },
     {
       "epoch": 0.6457862447529867,
-      "eval_accuracy": 0.327245113635166,
-      "eval_loss": 3.9299139976501465,
-      "eval_runtime": 183.1308,
-      "eval_samples_per_second": 98.35,
-      "eval_steps_per_second": 6.149,
+      "eval_accuracy": 0.3272947680048972,
+      "eval_loss": 3.9194672107696533,
+      "eval_runtime": 181.325,
+      "eval_samples_per_second": 99.33,
+      "eval_steps_per_second": 6.21,
       "step": 6000
     },
     {
       "epoch": 0.651167796792595,
-      "grad_norm": 0.6969402432441711,
-      "learning_rate": 0.0005615407822432927,
-      "loss": 4.0079,
+      "grad_norm": 0.6269046664237976,
+      "learning_rate": 0.0005615343174226915,
+      "loss": 3.9825,
       "step": 6050
     },
     {
       "epoch": 0.6565493488322032,
-      "grad_norm": 0.6455370187759399,
-      "learning_rate": 0.0005612175412132313,
-      "loss": 3.976,
+      "grad_norm": 0.7226353883743286,
+      "learning_rate": 0.00056121107639263,
+      "loss": 3.9832,
       "step": 6100
     },
     {
       "epoch": 0.6619309008718114,
-      "grad_norm": 0.696642279624939,
-      "learning_rate": 0.0005608943001831699,
-      "loss": 3.9809,
+      "grad_norm": 0.5846033692359924,
+      "learning_rate": 0.0005608878353625687,
+      "loss": 3.9839,
       "step": 6150
     },
     {
       "epoch": 0.6673124529114196,
-      "grad_norm": 0.6556864380836487,
-      "learning_rate": 0.0005605710591531085,
-      "loss": 3.9824,
+      "grad_norm": 0.6530674695968628,
+      "learning_rate": 0.0005605645943325072,
+      "loss": 3.9759,
       "step": 6200
     },
     {
       "epoch": 0.6726940049510278,
-      "grad_norm": 0.5885508060455322,
-      "learning_rate": 0.000560247818123047,
-      "loss": 3.9573,
+      "grad_norm": 0.6534646153450012,
+      "learning_rate": 0.0005602413533024458,
+      "loss": 3.9751,
       "step": 6250
     },
     {
       "epoch": 0.6780755569906362,
-      "grad_norm": 0.6368624567985535,
-      "learning_rate": 0.0005599245770929855,
-      "loss": 3.9602,
+      "grad_norm": 0.6664760112762451,
+      "learning_rate": 0.0005599181122723844,
+      "loss": 3.9406,
       "step": 6300
     },
     {
       "epoch": 0.6834571090302444,
-      "grad_norm": 0.6564461588859558,
-      "learning_rate": 0.0005596013360629242,
-      "loss": 3.9735,
+      "grad_norm": 0.6543918251991272,
+      "learning_rate": 0.000559594871242323,
+      "loss": 3.9741,
       "step": 6350
     },
     {
       "epoch": 0.6888386610698526,
-      "grad_norm": 0.5797594785690308,
-      "learning_rate": 0.0005592780950328628,
-      "loss": 3.9884,
+      "grad_norm": 0.6636673808097839,
+      "learning_rate": 0.0005592716302122616,
+      "loss": 3.975,
       "step": 6400
     },
     {
       "epoch": 0.6942202131094608,
-      "grad_norm": 0.5808694958686829,
-      "learning_rate": 0.0005589548540028014,
-      "loss": 3.981,
+      "grad_norm": 0.7524754405021667,
+      "learning_rate": 0.0005589483891822001,
+      "loss": 3.9804,
       "step": 6450
     },
     {
       "epoch": 0.699601765149069,
-      "grad_norm": 0.6396571397781372,
-      "learning_rate": 0.0005586316129727399,
-      "loss": 3.9754,
+      "grad_norm": 0.6153088808059692,
+      "learning_rate": 0.0005586251481521387,
+      "loss": 3.9682,
       "step": 6500
     },
     {
       "epoch": 0.7049833171886772,
-      "grad_norm": 0.5532586574554443,
-      "learning_rate": 0.0005583083719426786,
-      "loss": 3.9448,
+      "grad_norm": 0.62467360496521,
+      "learning_rate": 0.0005583019071220773,
+      "loss": 3.9476,
       "step": 6550
     },
     {
       "epoch": 0.7103648692282855,
-      "grad_norm": 0.6465068459510803,
-      "learning_rate": 0.0005579851309126171,
-      "loss": 3.9362,
+      "grad_norm": 0.6360256671905518,
+      "learning_rate": 0.000557978666092016,
+      "loss": 3.9448,
       "step": 6600
     },
     {
       "epoch": 0.7157464212678937,
-      "grad_norm": 0.6378588080406189,
-      "learning_rate": 0.0005576618898825558,
-      "loss": 3.9604,
+      "grad_norm": 0.5992259979248047,
+      "learning_rate": 0.0005576554250619545,
+      "loss": 3.9452,
       "step": 6650
     },
     {
       "epoch": 0.7211279733075019,
-      "grad_norm": 0.5902277231216431,
-      "learning_rate": 0.0005573386488524943,
-      "loss": 3.9683,
+      "grad_norm": 0.6049088835716248,
+      "learning_rate": 0.000557332184031893,
+      "loss": 3.9372,
       "step": 6700
     },
     {
       "epoch": 0.7265095253471101,
-      "grad_norm": 0.6843133568763733,
-      "learning_rate": 0.0005570154078224328,
-      "loss": 3.97,
+      "grad_norm": 0.6222875118255615,
+      "learning_rate": 0.0005570089430018317,
+      "loss": 3.95,
       "step": 6750
     },
     {
       "epoch": 0.7318910773867183,
-      "grad_norm": 0.5807896256446838,
-      "learning_rate": 0.0005566921667923715,
-      "loss": 3.9434,
+      "grad_norm": 0.5336934924125671,
+      "learning_rate": 0.0005566857019717702,
+      "loss": 3.9299,
       "step": 6800
     },
     {
       "epoch": 0.7372726294263265,
-      "grad_norm": 0.5920153856277466,
-      "learning_rate": 0.00055636892576231,
-      "loss": 3.9583,
+      "grad_norm": 0.6668237447738647,
+      "learning_rate": 0.0005563624609417089,
+      "loss": 3.9269,
       "step": 6850
     },
     {
       "epoch": 0.7426541814659348,
-      "grad_norm": 0.6305286288261414,
-      "learning_rate": 0.0005560456847322487,
-      "loss": 3.9369,
+      "grad_norm": 0.5827189087867737,
+      "learning_rate": 0.0005560392199116474,
+      "loss": 3.934,
       "step": 6900
     },
     {
       "epoch": 0.748035733505543,
-      "grad_norm": 0.6596742868423462,
-      "learning_rate": 0.0005557224437021872,
-      "loss": 3.9349,
+      "grad_norm": 0.6519905924797058,
+      "learning_rate": 0.000555715978881586,
+      "loss": 3.9101,
       "step": 6950
     },
     {
       "epoch": 0.7534172855451512,
-      "grad_norm": 0.5761475563049316,
-      "learning_rate": 0.0005553992026721258,
-      "loss": 3.9392,
+      "grad_norm": 0.6796578764915466,
+      "learning_rate": 0.0005553927378515246,
+      "loss": 3.9305,
       "step": 7000
     },
     {
       "epoch": 0.7534172855451512,
-      "eval_accuracy": 0.33236201273390115,
-      "eval_loss": 3.871555805206299,
-      "eval_runtime": 182.3203,
-      "eval_samples_per_second": 98.788,
-      "eval_steps_per_second": 6.176,
+      "eval_accuracy": 0.33403450663602946,
+      "eval_loss": 3.864957571029663,
+      "eval_runtime": 181.9746,
+      "eval_samples_per_second": 98.975,
+      "eval_steps_per_second": 6.188,
       "step": 7000
     },
     {
       "epoch": 0.7587988375847594,
-      "grad_norm": 0.7002006769180298,
-      "learning_rate": 0.0005550759616420644,
-      "loss": 3.9563,
+      "grad_norm": 0.561911940574646,
+      "learning_rate": 0.0005550694968214631,
+      "loss": 3.9368,
       "step": 7050
     },
     {
       "epoch": 0.7641803896243676,
-      "grad_norm": 0.6454929113388062,
-      "learning_rate": 0.000554752720612003,
-      "loss": 3.9207,
+      "grad_norm": 0.6652414798736572,
+      "learning_rate": 0.0005547462557914018,
+      "loss": 3.9173,
       "step": 7100
     },
     {
       "epoch": 0.7695619416639758,
-      "grad_norm": 0.5613568425178528,
-      "learning_rate": 0.0005544294795819415,
-      "loss": 3.9345,
+      "grad_norm": 0.6089035868644714,
+      "learning_rate": 0.0005544230147613403,
+      "loss": 3.9405,
       "step": 7150
     },
     {
       "epoch": 0.7749434937035841,
-      "grad_norm": 0.575527012348175,
-      "learning_rate": 0.0005541062385518801,
-      "loss": 3.9269,
+      "grad_norm": 0.6177041530609131,
+      "learning_rate": 0.0005540997737312789,
+      "loss": 3.9035,
       "step": 7200
     },
     {
       "epoch": 0.7803250457431924,
-      "grad_norm": 0.5978026986122131,
-      "learning_rate": 0.0005537829975218188,
-      "loss": 3.9484,
+      "grad_norm": 0.679482638835907,
+      "learning_rate": 0.0005537765327012175,
+      "loss": 3.9234,
       "step": 7250
     },
     {
       "epoch": 0.7857065977828006,
-      "grad_norm": 0.5839754939079285,
-      "learning_rate": 0.0005534597564917573,
-      "loss": 3.9268,
+      "grad_norm": 0.6637169122695923,
+      "learning_rate": 0.0005534532916711561,
+      "loss": 3.9177,
       "step": 7300
     },
     {
       "epoch": 0.7910881498224088,
-      "grad_norm": 0.6187883019447327,
-      "learning_rate": 0.0005531365154616959,
-      "loss": 3.9256,
+      "grad_norm": 0.6377207636833191,
+      "learning_rate": 0.0005531300506410946,
+      "loss": 3.9106,
       "step": 7350
     },
     {
       "epoch": 0.796469701862017,
-      "grad_norm": 0.6549730896949768,
-      "learning_rate": 0.0005528132744316344,
-      "loss": 3.92,
+      "grad_norm": 0.6485190987586975,
+      "learning_rate": 0.0005528068096110332,
+      "loss": 3.9204,
       "step": 7400
     },
     {
       "epoch": 0.8018512539016253,
-      "grad_norm": 0.5937124490737915,
-      "learning_rate": 0.0005524900334015731,
-      "loss": 3.8838,
+      "grad_norm": 0.571081817150116,
+      "learning_rate": 0.0005524835685809719,
+      "loss": 3.8924,
       "step": 7450
     },
     {
       "epoch": 0.8072328059412335,
-      "grad_norm": 0.5783231854438782,
-      "learning_rate": 0.0005521667923715117,
-      "loss": 3.9049,
+      "grad_norm": 0.677717387676239,
+      "learning_rate": 0.0005521603275509104,
+      "loss": 3.9122,
       "step": 7500
     },
     {
       "epoch": 0.8126143579808417,
-      "grad_norm": 0.5541702508926392,
-      "learning_rate": 0.0005518435513414502,
-      "loss": 3.9001,
+      "grad_norm": 0.5621068477630615,
+      "learning_rate": 0.000551837086520849,
+      "loss": 3.8824,
       "step": 7550
     },
     {
       "epoch": 0.8179959100204499,
-      "grad_norm": 0.5935503244400024,
-      "learning_rate": 0.0005515203103113888,
-      "loss": 3.9101,
+      "grad_norm": 0.5969014167785645,
+      "learning_rate": 0.0005515138454907875,
+      "loss": 3.8977,
       "step": 7600
     },
     {
       "epoch": 0.8233774620600581,
-      "grad_norm": 0.6102548241615295,
-      "learning_rate": 0.0005511970692813274,
-      "loss": 3.9069,
+      "grad_norm": 0.6785698533058167,
+      "learning_rate": 0.0005511906044607262,
+      "loss": 3.8885,
       "step": 7650
     },
     {
       "epoch": 0.8287590140996663,
-      "grad_norm": 0.5769250988960266,
-      "learning_rate": 0.000550873828251266,
-      "loss": 3.9094,
+      "grad_norm": 0.8015792369842529,
+      "learning_rate": 0.0005508673634306648,
+      "loss": 3.9005,
       "step": 7700
     },
     {
       "epoch": 0.8341405661392746,
-      "grad_norm": 0.5763354301452637,
-      "learning_rate": 0.0005505505872212045,
-      "loss": 3.9138,
+      "grad_norm": 0.6184539198875427,
+      "learning_rate": 0.0005505441224006034,
+      "loss": 3.9204,
       "step": 7750
     },
     {
       "epoch": 0.8395221181788828,
-      "grad_norm": 0.6475393176078796,
-      "learning_rate": 0.0005502273461911432,
-      "loss": 3.901,
+      "grad_norm": 0.6219672560691833,
+      "learning_rate": 0.0005502208813705419,
+      "loss": 3.8908,
       "step": 7800
     },
     {
       "epoch": 0.844903670218491,
-      "grad_norm": 0.5787845849990845,
-      "learning_rate": 0.0005499041051610817,
-      "loss": 3.9031,
+      "grad_norm": 0.5572973489761353,
+      "learning_rate": 0.0005498976403404805,
+      "loss": 3.8861,
       "step": 7850
     },
     {
       "epoch": 0.8502852222580992,
-      "grad_norm": 0.5721645355224609,
-      "learning_rate": 0.0005495808641310204,
-      "loss": 3.8975,
+      "grad_norm": 0.6005538105964661,
+      "learning_rate": 0.0005495743993104191,
+      "loss": 3.8814,
       "step": 7900
     },
     {
       "epoch": 0.8556667742977074,
-      "grad_norm": 0.6674014925956726,
-      "learning_rate": 0.0005492576231009589,
-      "loss": 3.8725,
+      "grad_norm": 0.6196568012237549,
+      "learning_rate": 0.0005492511582803576,
+      "loss": 3.8699,
       "step": 7950
     },
     {
       "epoch": 0.8610483263373157,
-      "grad_norm": 0.7338675260543823,
-      "learning_rate": 0.0005489343820708974,
-      "loss": 3.8996,
+      "grad_norm": 0.5852003693580627,
+      "learning_rate": 0.0005489279172502963,
+      "loss": 3.8504,
       "step": 8000
     },
     {
       "epoch": 0.8610483263373157,
-      "eval_accuracy": 0.3371352427484791,
-      "eval_loss": 3.8227832317352295,
-      "eval_runtime": 184.0783,
-      "eval_samples_per_second": 97.844,
-      "eval_steps_per_second": 6.117,
+      "eval_accuracy": 0.33771957797914537,
+      "eval_loss": 3.815176248550415,
+      "eval_runtime": 181.7741,
+      "eval_samples_per_second": 99.085,
+      "eval_steps_per_second": 6.195,
       "step": 8000
     },
     {
       "epoch": 0.8664298783769239,
-      "grad_norm": 0.5568352341651917,
-      "learning_rate": 0.0005486111410408361,
-      "loss": 3.8899,
+      "grad_norm": 0.6063625812530518,
+      "learning_rate": 0.0005486176058614372,
+      "loss": 3.8921,
       "step": 8050
     },
     {
       "epoch": 0.8718114304165321,
-      "grad_norm": 0.5996348261833191,
-      "learning_rate": 0.0005482879000107746,
-      "loss": 3.8833,
+      "grad_norm": 0.5820379257202148,
+      "learning_rate": 0.0005482943648313759,
+      "loss": 3.8736,
       "step": 8100
     },
     {
       "epoch": 0.8771929824561403,
-      "grad_norm": 0.6273235082626343,
-      "learning_rate": 0.0005479646589807133,
-      "loss": 3.8805,
+      "grad_norm": 0.6044704914093018,
+      "learning_rate": 0.0005479711238013145,
+      "loss": 3.8832,
       "step": 8150
     },
     {
       "epoch": 0.8825745344957485,
-      "grad_norm": 0.6420896649360657,
-      "learning_rate": 0.0005476414179506518,
-      "loss": 3.8839,
+      "grad_norm": 0.5778374671936035,
+      "learning_rate": 0.0005476478827712531,
+      "loss": 3.8815,
       "step": 8200
     },
     {
       "epoch": 0.8879560865353568,
-      "grad_norm": 0.5380816459655762,
+      "grad_norm": 0.5325333476066589,
       "learning_rate": 0.0005473246417411916,
-      "loss": 3.8628,
+      "loss": 3.8749,
       "step": 8250
     },
     {
       "epoch": 0.8933376385749651,
-      "grad_norm": 0.6761050224304199,
+      "grad_norm": 0.618618905544281,
       "learning_rate": 0.0005470014007111302,
-      "loss": 3.8861,
+      "loss": 3.8779,
       "step": 8300
     },
     {
       "epoch": 0.8987191906145733,
-      "grad_norm": 0.6020174622535706,
+      "grad_norm": 0.7097580432891846,
       "learning_rate": 0.0005466781596810688,
-      "loss": 3.8795,
+      "loss": 3.8664,
       "step": 8350
     },
     {
       "epoch": 0.9041007426541815,
-      "grad_norm": 0.6668645143508911,
+      "grad_norm": 0.6304284334182739,
       "learning_rate": 0.0005463549186510073,
-      "loss": 3.8844,
+      "loss": 3.859,
       "step": 8400
     },
     {
       "epoch": 0.9094822946937897,
-      "grad_norm": 0.5670798420906067,
+      "grad_norm": 0.5668020248413086,
       "learning_rate": 0.000546031677620946,
-      "loss": 3.8835,
+      "loss": 3.8624,
       "step": 8450
     },
     {
       "epoch": 0.9148638467333979,
-      "grad_norm": 0.5120532512664795,
+      "grad_norm": 0.5498570203781128,
       "learning_rate": 0.0005457084365908845,
-      "loss": 3.8562,
+      "loss": 3.8593,
       "step": 8500
     },
     {
       "epoch": 0.9202453987730062,
-      "grad_norm": 0.5259217023849487,
+      "grad_norm": 0.5575296878814697,
       "learning_rate": 0.0005453851955608232,
-      "loss": 3.8672,
+      "loss": 3.8623,
       "step": 8550
     },
     {
       "epoch": 0.9256269508126144,
-      "grad_norm": 0.583608865737915,
+      "grad_norm": 0.5803061127662659,
       "learning_rate": 0.0005450619545307617,
-      "loss": 3.8806,
+      "loss": 3.8653,
       "step": 8600
     },
     {
       "epoch": 0.9310085028522226,
-      "grad_norm": 0.64499431848526,
+      "grad_norm": 0.6923871636390686,
       "learning_rate": 0.0005447387135007003,
-      "loss": 3.8743,
+      "loss": 3.8628,
       "step": 8650
     },
     {
       "epoch": 0.9363900548918308,
-      "grad_norm": 0.5465984344482422,
+      "grad_norm": 0.6441589593887329,
       "learning_rate": 0.0005444154724706389,
-      "loss": 3.8684,
+      "loss": 3.8624,
       "step": 8700
     },
     {
       "epoch": 0.941771606931439,
-      "grad_norm": 0.5706291198730469,
+      "grad_norm": 0.6883953809738159,
       "learning_rate": 0.0005440922314405775,
-      "loss": 3.8527,
+      "loss": 3.8515,
       "step": 8750
     },
     {
       "epoch": 0.9471531589710472,
-      "grad_norm": 0.5907847285270691,
+      "grad_norm": 0.6100162863731384,
       "learning_rate": 0.0005437689904105161,
-      "loss": 3.8579,
+      "loss": 3.8368,
       "step": 8800
     },
     {
       "epoch": 0.9525347110106555,
-      "grad_norm": 0.6252136826515198,
+      "grad_norm": 0.5788195133209229,
       "learning_rate": 0.0005434457493804546,
-      "loss": 3.8496,
+      "loss": 3.8312,
       "step": 8850
     },
     {
       "epoch": 0.9579162630502637,
-      "grad_norm": 0.5910587310791016,
+      "grad_norm": 0.5563709735870361,
       "learning_rate": 0.0005431225083503932,
-      "loss": 3.8492,
+      "loss": 3.8517,
       "step": 8900
     },
     {
       "epoch": 0.9632978150898719,
-      "grad_norm": 0.5492550730705261,
+      "grad_norm": 0.6155171990394592,
       "learning_rate": 0.0005427992673203318,
-      "loss": 3.8486,
+      "loss": 3.8583,
       "step": 8950
     },
     {
       "epoch": 0.9686793671294801,
-      "grad_norm": 0.6138157844543457,
+      "grad_norm": 0.5446226596832275,
       "learning_rate": 0.0005424760262902704,
-      "loss": 3.8404,
+      "loss": 3.832,
       "step": 9000
     },
     {
       "epoch": 0.9686793671294801,
-      "eval_accuracy": 0.3408974576419433,
-      "eval_loss": 3.782958984375,
-      "eval_runtime": 183.7954,
-      "eval_samples_per_second": 97.995,
-      "eval_steps_per_second": 6.126,
+      "eval_accuracy": 0.34148950722764443,
+      "eval_loss": 3.7837822437286377,
+      "eval_runtime": 181.5161,
+      "eval_samples_per_second": 99.225,
+      "eval_steps_per_second": 6.203,
       "step": 9000
     },
     {
       "epoch": 0.9740609191690883,
-      "grad_norm": 0.5365199446678162,
+      "grad_norm": 0.6641152501106262,
       "learning_rate": 0.000542152785260209,
-      "loss": 3.8344,
+      "loss": 3.8364,
       "step": 9050
     },
     {
       "epoch": 0.9794424712086965,
-      "grad_norm": 0.5951078534126282,
+      "grad_norm": 0.5310914516448975,
       "learning_rate": 0.0005418295442301476,
-      "loss": 3.849,
+      "loss": 3.8395,
       "step": 9100
     },
     {
       "epoch": 0.9848240232483048,
-      "grad_norm": 0.586760401725769,
+      "grad_norm": 0.6875292062759399,
       "learning_rate": 0.0005415063032000861,
-      "loss": 3.8422,
+      "loss": 3.837,
       "step": 9150
     },
     {
       "epoch": 0.9902055752879131,
-      "grad_norm": 0.5982652902603149,
+      "grad_norm": 0.6055516600608826,
       "learning_rate": 0.0005411830621700248,
-      "loss": 3.8481,
+      "loss": 3.838,
       "step": 9200
     },
     {
       "epoch": 0.9955871273275213,
-      "grad_norm": 0.5718969106674194,
+      "grad_norm": 0.586495041847229,
       "learning_rate": 0.0005408598211399633,
-      "loss": 3.8469,
+      "loss": 3.8258,
       "step": 9250
     },
     {
       "epoch": 1.0009686793671295,
-      "grad_norm": 0.6520361304283142,
+      "grad_norm": 0.5521350502967834,
       "learning_rate": 0.0005405365801099019,
-      "loss": 3.8171,
+      "loss": 3.8056,
       "step": 9300
     },
     {
       "epoch": 1.0063502314067376,
-      "grad_norm": 0.5989835262298584,
+      "grad_norm": 0.543850302696228,
       "learning_rate": 0.0005402133390798405,
-      "loss": 3.7906,
+      "loss": 3.7712,
       "step": 9350
     },
     {
       "epoch": 1.011731783446346,
-      "grad_norm": 0.5910343527793884,
+      "grad_norm": 0.6066914200782776,
       "learning_rate": 0.000539890098049779,
-      "loss": 3.7743,
+      "loss": 3.7626,
       "step": 9400
     },
     {
       "epoch": 1.017113335485954,
-      "grad_norm": 0.531580924987793,
+      "grad_norm": 0.5474905967712402,
       "learning_rate": 0.0005395668570197177,
-      "loss": 3.7769,
+      "loss": 3.7664,
       "step": 9450
     },
     {
       "epoch": 1.0224948875255624,
-      "grad_norm": 0.5643661022186279,
+      "grad_norm": 0.6281277537345886,
       "learning_rate": 0.0005392436159896562,
-      "loss": 3.7784,
+      "loss": 3.7591,
       "step": 9500
     },
     {
       "epoch": 1.0278764395651705,
-      "grad_norm": 0.5594536662101746,
+      "grad_norm": 0.5506016612052917,
       "learning_rate": 0.0005389203749595948,
-      "loss": 3.7859,
+      "loss": 3.7604,
       "step": 9550
     },
     {
       "epoch": 1.0332579916047788,
-      "grad_norm": 0.5723885297775269,
+      "grad_norm": 0.5733294486999512,
       "learning_rate": 0.0005385971339295334,
-      "loss": 3.7723,
+      "loss": 3.7693,
       "step": 9600
     },
     {
       "epoch": 1.0386395436443872,
-      "grad_norm": 0.5776793956756592,
+      "grad_norm": 0.6508961915969849,
       "learning_rate": 0.000538273892899472,
-      "loss": 3.7622,
+      "loss": 3.7602,
       "step": 9650
     },
     {
       "epoch": 1.0440210956839953,
-      "grad_norm": 1.0429363250732422,
+      "grad_norm": 0.8109516501426697,
       "learning_rate": 0.0005379506518694106,
-      "loss": 3.7716,
+      "loss": 3.7589,
       "step": 9700
     },
     {
       "epoch": 1.0494026477236036,
-      "grad_norm": 0.6570566296577454,
+      "grad_norm": 0.5636641979217529,
       "learning_rate": 0.0005376274108393491,
-      "loss": 3.7688,
+      "loss": 3.7685,
       "step": 9750
     },
     {
       "epoch": 1.0547841997632117,
-      "grad_norm": 0.5422590374946594,
+      "grad_norm": 0.5453770160675049,
       "learning_rate": 0.0005373041698092877,
-      "loss": 3.775,
+      "loss": 3.748,
       "step": 9800
     },
     {
       "epoch": 1.06016575180282,
-      "grad_norm": 0.5695186853408813,
+      "grad_norm": 0.5409913063049316,
       "learning_rate": 0.0005369809287792263,
-      "loss": 3.7836,
+      "loss": 3.7585,
       "step": 9850
     },
     {
       "epoch": 1.0655473038424281,
-      "grad_norm": 0.5725084543228149,
+      "grad_norm": 0.5672662854194641,
       "learning_rate": 0.000536657687749165,
-      "loss": 3.7777,
+      "loss": 3.7782,
       "step": 9900
     },
     {
       "epoch": 1.0709288558820365,
-      "grad_norm": 0.5916622281074524,
+      "grad_norm": 0.6657660007476807,
       "learning_rate": 0.0005363344467191035,
-      "loss": 3.7616,
+      "loss": 3.7581,
       "step": 9950
     },
     {
       "epoch": 1.0763104079216446,
-      "grad_norm": 0.6363723278045654,
+      "grad_norm": 0.6501979231834412,
       "learning_rate": 0.000536011205689042,
-      "loss": 3.7704,
+      "loss": 3.7729,
       "step": 10000
     },
     {
       "epoch": 1.0763104079216446,
-      "eval_accuracy": 0.34435577040819043,
-      "eval_loss": 3.753837823867798,
-      "eval_runtime": 183.8836,
-      "eval_samples_per_second": 97.948,
-      "eval_steps_per_second": 6.123,
+      "eval_accuracy": 0.3445393937885969,
+      "eval_loss": 3.748054027557373,
+      "eval_runtime": 180.5408,
+      "eval_samples_per_second": 99.761,
+      "eval_steps_per_second": 6.237,
       "step": 10000
     },
     {
       "epoch": 1.081691959961253,
-      "grad_norm": 0.6295488476753235,
-      "learning_rate": 0.0005356879646589807,
-      "loss": 3.7654,
+      "grad_norm": 0.5782188177108765,
+      "learning_rate": 0.0005356944294795819,
+      "loss": 3.7595,
       "step": 10050
     },
     {
       "epoch": 1.087073512000861,
-      "grad_norm": 0.5673714280128479,
-      "learning_rate": 0.0005353647236289192,
-      "loss": 3.7797,
+      "grad_norm": 0.6113789677619934,
+      "learning_rate": 0.0005353711884495205,
+      "loss": 3.7733,
       "step": 10100
     },
     {
       "epoch": 1.0924550640404693,
-      "grad_norm": 0.5380173325538635,
-      "learning_rate": 0.0005350414825988579,
-      "loss": 3.767,
+      "grad_norm": 0.6462923288345337,
+      "learning_rate": 0.000535047947419459,
+      "loss": 3.7539,
       "step": 10150
     },
     {
       "epoch": 1.0978366160800774,
-      "grad_norm": 0.5471936464309692,
-      "learning_rate": 0.0005347182415687964,
-      "loss": 3.7705,
+      "grad_norm": 0.6201062798500061,
+      "learning_rate": 0.0005347247063893976,
+      "loss": 3.7409,
       "step": 10200
     },
     {
       "epoch": 1.1032181681196858,
-      "grad_norm": 0.5841044187545776,
-      "learning_rate": 0.000534395000538735,
-      "loss": 3.7676,
+      "grad_norm": 0.5814409255981445,
+      "learning_rate": 0.0005344014653593362,
+      "loss": 3.7534,
       "step": 10250
     },
     {
       "epoch": 1.1085997201592939,
-      "grad_norm": 0.5935708284378052,
-      "learning_rate": 0.0005340717595086736,
-      "loss": 3.7552,
+      "grad_norm": 0.5960645079612732,
+      "learning_rate": 0.0005340782243292748,
+      "loss": 3.7567,
       "step": 10300
     },
     {
       "epoch": 1.1139812721989022,
-      "grad_norm": 0.5655228495597839,
-      "learning_rate": 0.0005337485184786122,
-      "loss": 3.7727,
+      "grad_norm": 0.6369213461875916,
+      "learning_rate": 0.0005337549832992134,
+      "loss": 3.771,
       "step": 10350
     },
     {
       "epoch": 1.1193628242385103,
-      "grad_norm": 0.5181326270103455,
-      "learning_rate": 0.0005334252774485507,
-      "loss": 3.7737,
+      "grad_norm": 0.585033118724823,
+      "learning_rate": 0.000533431742269152,
+      "loss": 3.7652,
       "step": 10400
     },
     {
       "epoch": 1.1247443762781186,
-      "grad_norm": 0.5479722619056702,
+      "grad_norm": 0.5668421983718872,
       "learning_rate": 0.0005331085012390905,
-      "loss": 3.7639,
+      "loss": 3.7632,
       "step": 10450
     },
     {
       "epoch": 1.1301259283177267,
-      "grad_norm": 0.6414276361465454,
+      "grad_norm": 0.5859597325325012,
       "learning_rate": 0.0005327852602090292,
-      "loss": 3.7678,
+      "loss": 3.7572,
       "step": 10500
     },
     {
       "epoch": 1.135507480357335,
-      "grad_norm": 0.5645982623100281,
+      "grad_norm": 0.5835729241371155,
       "learning_rate": 0.0005324620191789678,
-      "loss": 3.781,
+      "loss": 3.7649,
       "step": 10550
     },
     {
       "epoch": 1.1408890323969434,
-      "grad_norm": 0.5996174812316895,
+      "grad_norm": 0.5598398447036743,
       "learning_rate": 0.0005321387781489063,
-      "loss": 3.7616,
+      "loss": 3.7717,
       "step": 10600
     },
     {
       "epoch": 1.1462705844365515,
-      "grad_norm": 0.5798478126525879,
+      "grad_norm": 0.6151158809661865,
       "learning_rate": 0.0005318155371188449,
-      "loss": 3.7509,
+      "loss": 3.7641,
       "step": 10650
     },
     {
       "epoch": 1.1516521364761596,
-      "grad_norm": 0.6145731210708618,
+      "grad_norm": 0.5782050490379333,
       "learning_rate": 0.0005314922960887834,
-      "loss": 3.742,
+      "loss": 3.7378,
       "step": 10700
     },
     {
       "epoch": 1.157033688515768,
-      "grad_norm": 0.5426759123802185,
+      "grad_norm": 0.5831411480903625,
       "learning_rate": 0.0005311690550587221,
-      "loss": 3.7573,
+      "loss": 3.7568,
       "step": 10750
     },
     {
       "epoch": 1.1624152405553763,
-      "grad_norm": 0.6367711424827576,
+      "grad_norm": 0.5914080739021301,
       "learning_rate": 0.0005308458140286607,
-      "loss": 3.7447,
+      "loss": 3.7639,
       "step": 10800
     },
     {
       "epoch": 1.1677967925949844,
-      "grad_norm": 0.5617762804031372,
+      "grad_norm": 0.5431932210922241,
       "learning_rate": 0.0005305225729985993,
-      "loss": 3.7552,
+      "loss": 3.7397,
       "step": 10850
     },
     {
       "epoch": 1.1731783446345927,
-      "grad_norm": 0.5642740726470947,
+      "grad_norm": 0.5925922393798828,
       "learning_rate": 0.0005301993319685378,
-      "loss": 3.7717,
+      "loss": 3.7445,
       "step": 10900
     },
     {
       "epoch": 1.1785598966742008,
-      "grad_norm": 0.5389478206634521,
+      "grad_norm": 0.6344436407089233,
       "learning_rate": 0.0005298760909384765,
-      "loss": 3.7591,
+      "loss": 3.741,
       "step": 10950
     },
     {
       "epoch": 1.1839414487138091,
-      "grad_norm": 0.5747219920158386,
-      "learning_rate": 0.0005295593147290163,
-      "loss": 3.7569,
+      "grad_norm": 0.5130102038383484,
+      "learning_rate": 0.000529552849908415,
+      "loss": 3.7631,
       "step": 11000
     },
     {
       "epoch": 1.1839414487138091,
-      "eval_accuracy": 0.34640528983103497,
-      "eval_loss": 3.7258360385894775,
-      "eval_runtime": 181.9752,
-      "eval_samples_per_second": 98.975,
-      "eval_steps_per_second": 6.188,
+      "eval_accuracy": 0.3475153963858355,
+      "eval_loss": 3.7219784259796143,
+      "eval_runtime": 180.2627,
+      "eval_samples_per_second": 99.915,
+      "eval_steps_per_second": 6.246,
       "step": 11000
     },
     {
       "epoch": 1.1893230007534172,
-      "grad_norm": 0.535572350025177,
-      "learning_rate": 0.0005292360736989548,
-      "loss": 3.7503,
+      "grad_norm": 0.5697294473648071,
+      "learning_rate": 0.0005292296088783535,
+      "loss": 3.7476,
       "step": 11050
     },
     {
       "epoch": 1.1947045527930256,
-      "grad_norm": 0.5539134740829468,
-      "learning_rate": 0.0005289128326688933,
-      "loss": 3.7425,
+      "grad_norm": 0.5963912010192871,
+      "learning_rate": 0.0005289063678482922,
+      "loss": 3.7428,
       "step": 11100
     },
     {
       "epoch": 1.2000861048326337,
-      "grad_norm": 0.5157566070556641,
-      "learning_rate": 0.000528589591638832,
-      "loss": 3.7573,
+      "grad_norm": 0.5590356588363647,
+      "learning_rate": 0.0005285831268182307,
+      "loss": 3.7333,
       "step": 11150
     },
     {
       "epoch": 1.205467656872242,
-      "grad_norm": 0.6375812888145447,
-      "learning_rate": 0.0005282663506087705,
-      "loss": 3.747,
+      "grad_norm": 0.5846792459487915,
+      "learning_rate": 0.0005282598857881694,
+      "loss": 3.7406,
       "step": 11200
     },
     {
       "epoch": 1.21084920891185,
-      "grad_norm": 0.5762513279914856,
-      "learning_rate": 0.0005279431095787092,
-      "loss": 3.7664,
+      "grad_norm": 0.5689064860343933,
+      "learning_rate": 0.0005279366447581079,
+      "loss": 3.7412,
       "step": 11250
     },
     {
       "epoch": 1.2162307609514584,
-      "grad_norm": 0.5227396488189697,
-      "learning_rate": 0.0005276198685486477,
-      "loss": 3.7466,
+      "grad_norm": 0.5604305863380432,
+      "learning_rate": 0.0005276134037280465,
+      "loss": 3.7377,
       "step": 11300
     },
     {
       "epoch": 1.2216123129910665,
-      "grad_norm": 0.5795957446098328,
-      "learning_rate": 0.0005272966275185863,
-      "loss": 3.7608,
+      "grad_norm": 0.5481633543968201,
+      "learning_rate": 0.0005272901626979851,
+      "loss": 3.7537,
       "step": 11350
     },
     {
       "epoch": 1.2269938650306749,
-      "grad_norm": 0.6271036863327026,
-      "learning_rate": 0.0005269733864885249,
-      "loss": 3.7467,
+      "grad_norm": 0.5674983263015747,
+      "learning_rate": 0.0005269669216679236,
+      "loss": 3.7466,
       "step": 11400
     },
     {
       "epoch": 1.232375417070283,
-      "grad_norm": 0.6098865866661072,
-      "learning_rate": 0.0005266501454584636,
-      "loss": 3.7546,
+      "grad_norm": 0.590938150882721,
+      "learning_rate": 0.0005266436806378623,
+      "loss": 3.7321,
       "step": 11450
     },
     {
       "epoch": 1.2377569691098913,
-      "grad_norm": 0.5986586809158325,
-      "learning_rate": 0.0005263269044284021,
-      "loss": 3.7606,
+      "grad_norm": 0.620957612991333,
+      "learning_rate": 0.0005263204396078008,
+      "loss": 3.7495,
       "step": 11500
     },
     {
       "epoch": 1.2431385211494996,
-      "grad_norm": 0.564387857913971,
-      "learning_rate": 0.0005260036633983406,
-      "loss": 3.7539,
+      "grad_norm": 0.5767254829406738,
+      "learning_rate": 0.0005259971985777394,
+      "loss": 3.7523,
       "step": 11550
     },
     {
       "epoch": 1.2485200731891077,
-      "grad_norm": 0.5610516667366028,
-      "learning_rate": 0.0005256804223682792,
-      "loss": 3.7399,
+      "grad_norm": 0.6098749041557312,
+      "learning_rate": 0.000525673957547678,
+      "loss": 3.7457,
       "step": 11600
     },
     {
       "epoch": 1.2539016252287158,
-      "grad_norm": 0.535050094127655,
-      "learning_rate": 0.0005253571813382178,
-      "loss": 3.7442,
+      "grad_norm": 0.5407803058624268,
+      "learning_rate": 0.0005253507165176167,
+      "loss": 3.729,
       "step": 11650
     },
     {
       "epoch": 1.2592831772683242,
-      "grad_norm": 0.600563645362854,
-      "learning_rate": 0.0005250339403081564,
-      "loss": 3.7306,
+      "grad_norm": 0.5610188245773315,
+      "learning_rate": 0.0005250274754875552,
+      "loss": 3.7268,
       "step": 11700
     },
     {
       "epoch": 1.2646647293079325,
-      "grad_norm": 0.6017090678215027,
-      "learning_rate": 0.000524710699278095,
-      "loss": 3.7414,
+      "grad_norm": 0.5667799115180969,
+      "learning_rate": 0.0005247042344574938,
+      "loss": 3.7441,
       "step": 11750
     },
     {
       "epoch": 1.2700462813475406,
-      "grad_norm": 0.7243601679801941,
-      "learning_rate": 0.0005243874582480336,
-      "loss": 3.748,
+      "grad_norm": 0.5885990262031555,
+      "learning_rate": 0.0005243809934274323,
+      "loss": 3.736,
       "step": 11800
     },
     {
       "epoch": 1.275427833387149,
-      "grad_norm": 0.5405592918395996,
-      "learning_rate": 0.0005240642172179722,
-      "loss": 3.7235,
+      "grad_norm": 0.5766186118125916,
+      "learning_rate": 0.0005240577523973709,
+      "loss": 3.73,
       "step": 11850
     },
     {
       "epoch": 1.280809385426757,
-      "grad_norm": 0.5994392037391663,
-      "learning_rate": 0.0005237409761879107,
-      "loss": 3.7396,
+      "grad_norm": 0.5727947354316711,
+      "learning_rate": 0.0005237345113673095,
+      "loss": 3.735,
       "step": 11900
     },
     {
       "epoch": 1.2861909374663654,
-      "grad_norm": 0.568977415561676,
-      "learning_rate": 0.0005234177351578493,
-      "loss": 3.7417,
+      "grad_norm": 0.6027268171310425,
+      "learning_rate": 0.0005234112703372481,
+      "loss": 3.7312,
       "step": 11950
     },
     {
       "epoch": 1.2915724895059735,
-      "grad_norm": 0.5830429196357727,
-      "learning_rate": 0.0005230944941277878,
-      "loss": 3.7236,
+      "grad_norm": 0.5663910508155823,
+      "learning_rate": 0.0005230880293071867,
+      "loss": 3.712,
       "step": 12000
     },
     {
       "epoch": 1.2915724895059735,
-      "eval_accuracy": 0.3490606577563331,
-      "eval_loss": 3.7004106044769287,
-      "eval_runtime": 181.7656,
-      "eval_samples_per_second": 99.089,
-      "eval_steps_per_second": 6.195,
+      "eval_accuracy": 0.34973050280970935,
+      "eval_loss": 3.696995973587036,
+      "eval_runtime": 180.8275,
+      "eval_samples_per_second": 99.603,
+      "eval_steps_per_second": 6.227,
       "step": 12000
     },
     {
       "epoch": 1.2969540415455818,
-      "grad_norm": 0.5280687212944031,
-      "learning_rate": 0.0005227712530977265,
-      "loss": 3.7264,
+      "grad_norm": 0.6671003103256226,
+      "learning_rate": 0.0005227647882771253,
+      "loss": 3.7403,
       "step": 12050
     },
     {
       "epoch": 1.30233559358519,
-      "grad_norm": 0.5419384837150574,
-      "learning_rate": 0.0005224480120676651,
-      "loss": 3.7493,
+      "grad_norm": 0.5946469902992249,
+      "learning_rate": 0.0005224415472470639,
+      "loss": 3.7339,
       "step": 12100
     },
     {
       "epoch": 1.3077171456247982,
-      "grad_norm": 0.5870584845542908,
-      "learning_rate": 0.0005221247710376037,
-      "loss": 3.7321,
+      "grad_norm": 0.6041794419288635,
+      "learning_rate": 0.0005221183062170024,
+      "loss": 3.7389,
       "step": 12150
     },
     {
       "epoch": 1.3130986976644063,
-      "grad_norm": 0.6159378886222839,
-      "learning_rate": 0.0005218015300075422,
-      "loss": 3.7377,
+      "grad_norm": 0.5971732139587402,
+      "learning_rate": 0.0005217950651869409,
+      "loss": 3.7083,
       "step": 12200
     },
     {
       "epoch": 1.3184802497040147,
-      "grad_norm": 0.6157721281051636,
+      "grad_norm": 0.5819820165634155,
       "learning_rate": 0.0005214782889774809,
-      "loss": 3.7259,
+      "loss": 3.7381,
       "step": 12250
     },
     {
       "epoch": 1.3238618017436228,
-      "grad_norm": 0.6117875576019287,
+      "grad_norm": 0.6009936332702637,
       "learning_rate": 0.0005211550479474194,
-      "loss": 3.7355,
+      "loss": 3.7107,
       "step": 12300
     },
     {
       "epoch": 1.329243353783231,
-      "grad_norm": 0.6459168791770935,
+      "grad_norm": 0.5555267930030823,
       "learning_rate": 0.000520831806917358,
-      "loss": 3.7286,
+      "loss": 3.732,
       "step": 12350
     },
     {
       "epoch": 1.3346249058228392,
-      "grad_norm": 0.5380271673202515,
+      "grad_norm": 0.5123233199119568,
       "learning_rate": 0.0005205085658872966,
-      "loss": 3.7315,
+      "loss": 3.7234,
       "step": 12400
     },
     {
       "epoch": 1.3400064578624475,
-      "grad_norm": 0.5702387094497681,
+      "grad_norm": 0.5637345314025879,
       "learning_rate": 0.0005201853248572351,
-      "loss": 3.7275,
+      "loss": 3.7251,
       "step": 12450
     },
     {
       "epoch": 1.3453880099020559,
-      "grad_norm": 0.6522827744483948,
+      "grad_norm": 0.5650758743286133,
       "learning_rate": 0.0005198620838271738,
-      "loss": 3.731,
+      "loss": 3.7111,
       "step": 12500
     },
     {
       "epoch": 1.350769561941664,
-      "grad_norm": 0.6003380417823792,
+      "grad_norm": 0.5396456122398376,
       "learning_rate": 0.0005195388427971123,
-      "loss": 3.7168,
+      "loss": 3.713,
       "step": 12550
     },
     {
       "epoch": 1.356151113981272,
-      "grad_norm": 0.5687087178230286,
+      "grad_norm": 0.6567341685295105,
       "learning_rate": 0.000519215601767051,
-      "loss": 3.7108,
+      "loss": 3.7259,
       "step": 12600
     },
     {
       "epoch": 1.3615326660208804,
-      "grad_norm": 0.602519690990448,
+      "grad_norm": 0.5406765341758728,
       "learning_rate": 0.0005188923607369895,
-      "loss": 3.7322,
+      "loss": 3.7184,
       "step": 12650
     },
     {
       "epoch": 1.3669142180604887,
-      "grad_norm": 0.5567368268966675,
+      "grad_norm": 0.5665279626846313,
       "learning_rate": 0.000518569119706928,
-      "loss": 3.7188,
+      "loss": 3.7163,
       "step": 12700
     },
     {
       "epoch": 1.3722957701000968,
-      "grad_norm": 0.6820892095565796,
+      "grad_norm": 0.5712398290634155,
       "learning_rate": 0.0005182458786768667,
-      "loss": 3.7278,
+      "loss": 3.7218,
       "step": 12750
     },
     {
       "epoch": 1.3776773221397052,
-      "grad_norm": 0.6009151339530945,
+      "grad_norm": 0.5649273991584778,
       "learning_rate": 0.0005179226376468052,
-      "loss": 3.7225,
+      "loss": 3.7087,
       "step": 12800
     },
     {
       "epoch": 1.3830588741793133,
-      "grad_norm": 0.5885601043701172,
+      "grad_norm": 0.6065050959587097,
       "learning_rate": 0.0005175993966167438,
-      "loss": 3.7263,
+      "loss": 3.7095,
       "step": 12850
     },
     {
       "epoch": 1.3884404262189216,
-      "grad_norm": 0.6636022329330444,
+      "grad_norm": 0.7034748196601868,
       "learning_rate": 0.0005172761555866824,
-      "loss": 3.7311,
+      "loss": 3.7115,
       "step": 12900
     },
     {
       "epoch": 1.3938219782585297,
-      "grad_norm": 0.6068388223648071,
+      "grad_norm": 0.5810460448265076,
       "learning_rate": 0.0005169529145566211,
-      "loss": 3.7087,
+      "loss": 3.7196,
       "step": 12950
     },
     {
       "epoch": 1.399203530298138,
-      "grad_norm": 0.534371018409729,
+      "grad_norm": 0.5916039347648621,
       "learning_rate": 0.0005166296735265596,
-      "loss": 3.7062,
+      "loss": 3.7094,
       "step": 13000
     },
     {
       "epoch": 1.399203530298138,
-      "eval_accuracy": 0.35125294707376586,
-      "eval_loss": 3.679492712020874,
-      "eval_runtime": 182.0328,
-      "eval_samples_per_second": 98.944,
-      "eval_steps_per_second": 6.186,
+      "eval_accuracy": 0.3518014268514534,
+      "eval_loss": 3.6733601093292236,
+      "eval_runtime": 180.8077,
+      "eval_samples_per_second": 99.614,
+      "eval_steps_per_second": 6.228,
       "step": 13000
     },
     {
       "epoch": 1.4045850823377461,
-      "grad_norm": 0.5494077801704407,
+      "grad_norm": 0.5988173484802246,
       "learning_rate": 0.0005163064324964982,
-      "loss": 3.7203,
+      "loss": 3.7396,
       "step": 13050
     },
     {
       "epoch": 1.4099666343773545,
-      "grad_norm": 0.573820948600769,
+      "grad_norm": 0.578076183795929,
       "learning_rate": 0.0005159831914664367,
-      "loss": 3.6958,
+      "loss": 3.6889,
       "step": 13100
     },
     {
       "epoch": 1.4153481864169626,
-      "grad_norm": 0.6078840494155884,
+      "grad_norm": 0.5827022194862366,
       "learning_rate": 0.0005156599504363753,
-      "loss": 3.7048,
+      "loss": 3.6909,
       "step": 13150
     },
     {
       "epoch": 1.420729738456571,
-      "grad_norm": 0.6435017585754395,
+      "grad_norm": 0.5574851036071777,
       "learning_rate": 0.000515336709406314,
-      "loss": 3.719,
+      "loss": 3.7278,
       "step": 13200
     },
     {
       "epoch": 1.426111290496179,
-      "grad_norm": 0.5942484736442566,
+      "grad_norm": 0.6071725487709045,
       "learning_rate": 0.0005150134683762525,
-      "loss": 3.724,
+      "loss": 3.7183,
       "step": 13250
     },
     {
       "epoch": 1.4314928425357873,
-      "grad_norm": 0.5379244685173035,
+      "grad_norm": 0.5537685751914978,
       "learning_rate": 0.0005146902273461911,
-      "loss": 3.7028,
+      "loss": 3.7043,
       "step": 13300
     },
     {
       "epoch": 1.4368743945753955,
-      "grad_norm": 0.5623278617858887,
+      "grad_norm": 0.6063131093978882,
       "learning_rate": 0.0005143669863161297,
-      "loss": 3.7117,
+      "loss": 3.708,
       "step": 13350
     },
     {
       "epoch": 1.4422559466150038,
-      "grad_norm": 0.6037217378616333,
+      "grad_norm": 0.5979368686676025,
       "learning_rate": 0.0005140437452860683,
-      "loss": 3.7123,
+      "loss": 3.7116,
       "step": 13400
     },
     {
       "epoch": 1.447637498654612,
-      "grad_norm": 0.6184905767440796,
+      "grad_norm": 0.5357139706611633,
       "learning_rate": 0.0005137205042560069,
-      "loss": 3.7258,
+      "loss": 3.7085,
       "step": 13450
     },
     {
       "epoch": 1.4530190506942202,
-      "grad_norm": 0.5425773859024048,
+      "grad_norm": 0.5596767663955688,
       "learning_rate": 0.0005133972632259455,
-      "loss": 3.6993,
+      "loss": 3.7101,
       "step": 13500
     },
     {
       "epoch": 1.4584006027338283,
-      "grad_norm": 0.6625587344169617,
+      "grad_norm": 0.5640789270401001,
       "learning_rate": 0.000513074022195884,
-      "loss": 3.6997,
+      "loss": 3.6942,
       "step": 13550
     },
     {
       "epoch": 1.4637821547734367,
-      "grad_norm": 0.5868763327598572,
+      "grad_norm": 0.6157506704330444,
       "learning_rate": 0.0005127507811658226,
-      "loss": 3.7111,
+      "loss": 3.6997,
       "step": 13600
     },
     {
       "epoch": 1.469163706813045,
-      "grad_norm": 0.5310501456260681,
+      "grad_norm": 0.533714771270752,
       "learning_rate": 0.0005124275401357612,
-      "loss": 3.7201,
+      "loss": 3.7264,
       "step": 13650
     },
     {
       "epoch": 1.474545258852653,
-      "grad_norm": 0.5664429664611816,
+      "grad_norm": 0.5745089650154114,
       "learning_rate": 0.0005121042991056997,
-      "loss": 3.689,
+      "loss": 3.7102,
       "step": 13700
     },
     {
       "epoch": 1.4799268108922612,
-      "grad_norm": 0.5378715395927429,
+      "grad_norm": 0.5803775787353516,
       "learning_rate": 0.0005117810580756384,
-      "loss": 3.7096,
+      "loss": 3.7017,
       "step": 13750
     },
     {
       "epoch": 1.4853083629318695,
-      "grad_norm": 0.5423511266708374,
+      "grad_norm": 0.5686407685279846,
       "learning_rate": 0.0005114578170455769,
-      "loss": 3.7038,
+      "loss": 3.7144,
       "step": 13800
     },
     {
       "epoch": 1.4906899149714778,
-      "grad_norm": 0.5492138862609863,
+      "grad_norm": 0.5848187208175659,
       "learning_rate": 0.0005111345760155156,
-      "loss": 3.7261,
+      "loss": 3.7044,
       "step": 13850
     },
     {
       "epoch": 1.496071467011086,
-      "grad_norm": 0.5724121928215027,
+      "grad_norm": 0.544100821018219,
       "learning_rate": 0.0005108113349854541,
-      "loss": 3.7067,
+      "loss": 3.7011,
       "step": 13900
     },
     {
       "epoch": 1.501453019050694,
-      "grad_norm": 0.5634517669677734,
+      "grad_norm": 0.6704515814781189,
       "learning_rate": 0.0005104880939553926,
-      "loss": 3.6915,
+      "loss": 3.7058,
       "step": 13950
     },
     {
       "epoch": 1.5068345710903024,
-      "grad_norm": 0.6407066583633423,
+      "grad_norm": 0.5616099238395691,
       "learning_rate": 0.0005101648529253313,
-      "loss": 3.7181,
+      "loss": 3.7,
       "step": 14000
     },
     {
       "epoch": 1.5068345710903024,
-      "eval_accuracy": 0.35367851414133633,
-      "eval_loss": 3.6570589542388916,
-      "eval_runtime": 182.1403,
-      "eval_samples_per_second": 98.885,
-      "eval_steps_per_second": 6.182,
+      "eval_accuracy": 0.35376793546800767,
+      "eval_loss": 3.655121326446533,
+      "eval_runtime": 180.3147,
+      "eval_samples_per_second": 99.887,
+      "eval_steps_per_second": 6.245,
       "step": 14000
     },
     {
       "epoch": 1.5122161231299107,
-      "grad_norm": 0.5960928797721863,
+      "grad_norm": 0.5651384592056274,
       "learning_rate": 0.0005098416118952699,
-      "loss": 3.6972,
+      "loss": 3.6862,
       "step": 14050
     },
     {
       "epoch": 1.5175976751695188,
-      "grad_norm": 0.5764510035514832,
+      "grad_norm": 0.5399876236915588,
       "learning_rate": 0.0005095183708652085,
-      "loss": 3.7012,
+      "loss": 3.6925,
       "step": 14100
     },
     {
       "epoch": 1.5229792272091272,
-      "grad_norm": 0.5632539391517639,
+      "grad_norm": 0.5576087236404419,
       "learning_rate": 0.000509195129835147,
-      "loss": 3.7139,
+      "loss": 3.7065,
       "step": 14150
     },
     {
       "epoch": 1.5283607792487355,
-      "grad_norm": 0.5743679404258728,
+      "grad_norm": 0.5209526419639587,
       "learning_rate": 0.0005088718888050856,
-      "loss": 3.6973,
+      "loss": 3.7018,
       "step": 14200
     },
     {
       "epoch": 1.5337423312883436,
-      "grad_norm": 0.5791763663291931,
+      "grad_norm": 0.5912092328071594,
       "learning_rate": 0.0005085486477750242,
-      "loss": 3.7127,
+      "loss": 3.7012,
       "step": 14250
     },
     {
       "epoch": 1.5391238833279517,
-      "grad_norm": 0.5912649631500244,
+      "grad_norm": 0.5570158362388611,
       "learning_rate": 0.0005082254067449629,
-      "loss": 3.6992,
+      "loss": 3.6776,
       "step": 14300
     },
     {
       "epoch": 1.54450543536756,
-      "grad_norm": 0.5650004744529724,
+      "grad_norm": 0.6079233884811401,
       "learning_rate": 0.0005079021657149014,
-      "loss": 3.7003,
+      "loss": 3.698,
       "step": 14350
     },
     {
       "epoch": 1.5498869874071683,
-      "grad_norm": 0.5783352851867676,
+      "grad_norm": 0.580946683883667,
       "learning_rate": 0.0005075789246848399,
-      "loss": 3.7108,
+      "loss": 3.6949,
       "step": 14400
     },
     {
       "epoch": 1.5552685394467765,
-      "grad_norm": 0.6116908192634583,
+      "grad_norm": 0.5581622123718262,
       "learning_rate": 0.0005072556836547785,
-      "loss": 3.6884,
+      "loss": 3.6902,
       "step": 14450
     },
     {
       "epoch": 1.5606500914863846,
-      "grad_norm": 0.5558695197105408,
+      "grad_norm": 0.5859177708625793,
       "learning_rate": 0.0005069324426247171,
-      "loss": 3.6849,
+      "loss": 3.6929,
       "step": 14500
     },
     {
       "epoch": 1.566031643525993,
-      "grad_norm": 0.6027122735977173,
-      "learning_rate": 0.0005066092015946557,
-      "loss": 3.6764,
+      "grad_norm": 0.5541911721229553,
+      "learning_rate": 0.0005066156664152569,
+      "loss": 3.695,
       "step": 14550
     },
     {
       "epoch": 1.5714131955656012,
-      "grad_norm": 0.6431997418403625,
-      "learning_rate": 0.0005062859605645943,
-      "loss": 3.7126,
+      "grad_norm": 0.5521026253700256,
+      "learning_rate": 0.0005062924253851955,
+      "loss": 3.6869,
       "step": 14600
     },
     {
       "epoch": 1.5767947476052093,
-      "grad_norm": 0.548612117767334,
-      "learning_rate": 0.0005059627195345329,
-      "loss": 3.6901,
+      "grad_norm": 0.5752954483032227,
+      "learning_rate": 0.0005059691843551341,
+      "loss": 3.6773,
       "step": 14650
     },
     {
       "epoch": 1.5821762996448174,
-      "grad_norm": 0.6005812883377075,
-      "learning_rate": 0.0005056394785044715,
-      "loss": 3.695,
+      "grad_norm": 0.5865405201911926,
+      "learning_rate": 0.0005056459433250727,
+      "loss": 3.685,
       "step": 14700
     },
     {
       "epoch": 1.5875578516844258,
-      "grad_norm": 0.6675664782524109,
-      "learning_rate": 0.00050531623747441,
-      "loss": 3.6855,
+      "grad_norm": 0.5415293574333191,
+      "learning_rate": 0.0005053227022950113,
+      "loss": 3.662,
       "step": 14750
     },
     {
       "epoch": 1.592939403724034,
-      "grad_norm": 0.6004554033279419,
-      "learning_rate": 0.0005049929964443486,
-      "loss": 3.6797,
+      "grad_norm": 0.644170343875885,
+      "learning_rate": 0.0005049994612649499,
+      "loss": 3.6704,
       "step": 14800
     },
     {
       "epoch": 1.5983209557636422,
-      "grad_norm": 0.578523576259613,
-      "learning_rate": 0.0005046697554142871,
-      "loss": 3.6766,
+      "grad_norm": 0.5317102670669556,
+      "learning_rate": 0.0005046762202348884,
+      "loss": 3.6806,
       "step": 14850
     },
     {
       "epoch": 1.6037025078032503,
-      "grad_norm": 0.5477756857872009,
-      "learning_rate": 0.0005043465143842258,
-      "loss": 3.6804,
+      "grad_norm": 0.6086379289627075,
+      "learning_rate": 0.000504352979204827,
+      "loss": 3.6748,
       "step": 14900
     },
     {
       "epoch": 1.6090840598428586,
-      "grad_norm": 0.5516514778137207,
-      "learning_rate": 0.0005040232733541644,
-      "loss": 3.6764,
+      "grad_norm": 0.5908480882644653,
+      "learning_rate": 0.0005040297381747656,
+      "loss": 3.6994,
       "step": 14950
     },
     {
       "epoch": 1.614465611882467,
-      "grad_norm": 0.5752219557762146,
-      "learning_rate": 0.000503700032324103,
-      "loss": 3.6986,
+      "grad_norm": 0.5562351942062378,
+      "learning_rate": 0.0005037064971447042,
+      "loss": 3.6918,
       "step": 15000
     },
     {
       "epoch": 1.614465611882467,
-      "eval_accuracy": 0.3556933732929681,
-      "eval_loss": 3.6386351585388184,
-      "eval_runtime": 181.8961,
-      "eval_samples_per_second": 99.018,
-      "eval_steps_per_second": 6.19,
+      "eval_accuracy": 0.3556150345608538,
+      "eval_loss": 3.6376163959503174,
+      "eval_runtime": 180.4248,
+      "eval_samples_per_second": 99.825,
+      "eval_steps_per_second": 6.241,
       "step": 15000
     },
     {
       "epoch": 1.619847163922075,
-      "grad_norm": 0.5473290085792542,
+      "grad_norm": 0.5730331540107727,
       "learning_rate": 0.0005033832561146428,
-      "loss": 3.6819,
+      "loss": 3.6764,
       "step": 15050
     },
     {
       "epoch": 1.6252287159616834,
-      "grad_norm": 0.5905124545097351,
+      "grad_norm": 0.6177064180374146,
       "learning_rate": 0.0005030600150845813,
-      "loss": 3.672,
+      "loss": 3.6803,
       "step": 15100
     },
     {
       "epoch": 1.6306102680012917,
-      "grad_norm": 0.6535428762435913,
+      "grad_norm": 0.5296208262443542,
       "learning_rate": 0.00050273677405452,
-      "loss": 3.6776,
+      "loss": 3.6811,
       "step": 15150
     },
     {
       "epoch": 1.6359918200408998,
-      "grad_norm": 0.5464186072349548,
+      "grad_norm": 0.5068338513374329,
       "learning_rate": 0.0005024135330244585,
-      "loss": 3.6604,
+      "loss": 3.6784,
       "step": 15200
     },
     {
       "epoch": 1.641373372080508,
-      "grad_norm": 0.5981823801994324,
+      "grad_norm": 0.5747875571250916,
       "learning_rate": 0.0005020902919943972,
-      "loss": 3.6632,
+      "loss": 3.6882,
       "step": 15250
     },
     {
       "epoch": 1.6467549241201163,
-      "grad_norm": 0.5628587603569031,
+      "grad_norm": 0.5960535407066345,
       "learning_rate": 0.0005017670509643357,
-      "loss": 3.6775,
+      "loss": 3.6624,
       "step": 15300
     },
     {
       "epoch": 1.6521364761597246,
-      "grad_norm": 0.5168464183807373,
+      "grad_norm": 0.5554191470146179,
       "learning_rate": 0.0005014438099342743,
-      "loss": 3.6883,
+      "loss": 3.6785,
       "step": 15350
     },
     {
       "epoch": 1.6575180281993327,
-      "grad_norm": 0.5790534019470215,
+      "grad_norm": 0.5345931053161621,
       "learning_rate": 0.0005011205689042129,
-      "loss": 3.6784,
+      "loss": 3.6854,
       "step": 15400
     },
     {
       "epoch": 1.6628995802389408,
-      "grad_norm": 0.5653476119041443,
+      "grad_norm": 0.5266675353050232,
       "learning_rate": 0.0005007973278741514,
-      "loss": 3.6495,
+      "loss": 3.6526,
       "step": 15450
     },
     {
       "epoch": 1.6682811322785491,
-      "grad_norm": 0.5761940479278564,
+      "grad_norm": 0.595730185508728,
       "learning_rate": 0.00050047408684409,
-      "loss": 3.6852,
+      "loss": 3.6677,
       "step": 15500
     },
     {
       "epoch": 1.6736626843181575,
-      "grad_norm": 0.5976137518882751,
+      "grad_norm": 0.6001737117767334,
       "learning_rate": 0.0005001508458140286,
-      "loss": 3.6591,
+      "loss": 3.6658,
       "step": 15550
     },
     {
       "epoch": 1.6790442363577656,
-      "grad_norm": 0.5772688388824463,
+      "grad_norm": 0.6037628054618835,
       "learning_rate": 0.0004998276047839673,
-      "loss": 3.654,
+      "loss": 3.6563,
       "step": 15600
     },
     {
       "epoch": 1.6844257883973737,
-      "grad_norm": 0.5581541061401367,
+      "grad_norm": 0.6878479719161987,
       "learning_rate": 0.0004995043637539058,
-      "loss": 3.6748,
+      "loss": 3.6749,
       "step": 15650
     },
     {
       "epoch": 1.689807340436982,
-      "grad_norm": 0.5584582090377808,
+      "grad_norm": 0.5797104239463806,
       "learning_rate": 0.0004991811227238443,
-      "loss": 3.666,
+      "loss": 3.658,
       "step": 15700
     },
     {
       "epoch": 1.6951888924765903,
-      "grad_norm": 0.6056560277938843,
+      "grad_norm": 0.5414761304855347,
       "learning_rate": 0.0004988578816937829,
-      "loss": 3.656,
+      "loss": 3.6646,
       "step": 15750
     },
     {
       "epoch": 1.7005704445161984,
-      "grad_norm": 0.5387526750564575,
+      "grad_norm": 0.6197190880775452,
       "learning_rate": 0.0004985346406637215,
-      "loss": 3.6844,
+      "loss": 3.6749,
       "step": 15800
     },
     {
       "epoch": 1.7059519965558065,
-      "grad_norm": 0.6060013175010681,
+      "grad_norm": 0.5700779557228088,
       "learning_rate": 0.0004982113996336602,
-      "loss": 3.6704,
+      "loss": 3.6855,
       "step": 15850
     },
     {
       "epoch": 1.7113335485954149,
-      "grad_norm": 0.5558328628540039,
+      "grad_norm": 0.548884928226471,
       "learning_rate": 0.0004978881586035987,
-      "loss": 3.6695,
+      "loss": 3.6678,
       "step": 15900
     },
     {
       "epoch": 1.7167151006350232,
-      "grad_norm": 0.6431989073753357,
+      "grad_norm": 0.5491683483123779,
       "learning_rate": 0.0004975649175735373,
-      "loss": 3.6561,
+      "loss": 3.6686,
       "step": 15950
     },
     {
       "epoch": 1.7220966526746313,
-      "grad_norm": 0.5722593069076538,
+      "grad_norm": 0.5902188420295715,
       "learning_rate": 0.0004972416765434759,
-      "loss": 3.6594,
+      "loss": 3.6604,
       "step": 16000
     },
     {
       "epoch": 1.7220966526746313,
-      "eval_accuracy": 0.35697623793939365,
-      "eval_loss": 3.6226398944854736,
-      "eval_runtime": 182.9771,
-      "eval_samples_per_second": 98.433,
-      "eval_steps_per_second": 6.154,
+      "eval_accuracy": 0.35732892521323945,
+      "eval_loss": 3.6174802780151367,
+      "eval_runtime": 181.6566,
+      "eval_samples_per_second": 99.149,
+      "eval_steps_per_second": 6.199,
       "step": 16000
     },
     {
       "epoch": 1.7274782047142396,
-      "grad_norm": 0.5558207035064697,
+      "grad_norm": 0.5962377786636353,
       "learning_rate": 0.0004969184355134145,
-      "loss": 3.6741,
+      "loss": 3.673,
       "step": 16050
     },
     {
       "epoch": 1.732859756753848,
-      "grad_norm": 0.6220034956932068,
+      "grad_norm": 0.6508206725120544,
       "learning_rate": 0.0004965951944833531,
-      "loss": 3.6571,
+      "loss": 3.6609,
       "step": 16100
     },
     {
       "epoch": 1.738241308793456,
-      "grad_norm": 0.5504184365272522,
+      "grad_norm": 0.5311007499694824,
       "learning_rate": 0.0004962719534532916,
-      "loss": 3.6812,
+      "loss": 3.6601,
       "step": 16150
     },
     {
       "epoch": 1.7436228608330642,
-      "grad_norm": 0.6175248622894287,
+      "grad_norm": 0.5204987525939941,
       "learning_rate": 0.0004959487124232302,
-      "loss": 3.6989,
+      "loss": 3.6712,
       "step": 16200
     },
     {
       "epoch": 1.7490044128726725,
-      "grad_norm": 0.6186313629150391,
+      "grad_norm": 0.650790810585022,
       "learning_rate": 0.0004956254713931688,
-      "loss": 3.6716,
+      "loss": 3.6662,
       "step": 16250
     },
     {
       "epoch": 1.7543859649122808,
-      "grad_norm": 0.5942351818084717,
+      "grad_norm": 0.6375752687454224,
       "learning_rate": 0.0004953022303631074,
-      "loss": 3.6773,
+      "loss": 3.6894,
       "step": 16300
     },
     {
       "epoch": 1.759767516951889,
-      "grad_norm": 0.5420483350753784,
+      "grad_norm": 0.5430384874343872,
       "learning_rate": 0.0004949789893330459,
-      "loss": 3.6632,
+      "loss": 3.672,
       "step": 16350
     },
     {
       "epoch": 1.765149068991497,
-      "grad_norm": 0.5678284764289856,
+      "grad_norm": 0.5661519169807434,
       "learning_rate": 0.0004946557483029846,
       "loss": 3.6603,
       "step": 16400
     },
     {
       "epoch": 1.7705306210311054,
-      "grad_norm": 0.5654410719871521,
+      "grad_norm": 0.5399925112724304,
       "learning_rate": 0.0004943325072729231,
-      "loss": 3.6556,
+      "loss": 3.655,
       "step": 16450
     },
     {
       "epoch": 1.7759121730707137,
-      "grad_norm": 0.5799243450164795,
+      "grad_norm": 0.5748807787895203,
       "learning_rate": 0.0004940092662428617,
-      "loss": 3.6594,
+      "loss": 3.6492,
       "step": 16500
     },
     {
       "epoch": 1.7812937251103218,
-      "grad_norm": 0.5611895322799683,
+      "grad_norm": 0.5472862124443054,
       "learning_rate": 0.0004936860252128003,
-      "loss": 3.6558,
+      "loss": 3.6645,
       "step": 16550
     },
     {
       "epoch": 1.78667527714993,
-      "grad_norm": 0.6114226579666138,
+      "grad_norm": 0.5696579813957214,
       "learning_rate": 0.0004933627841827388,
-      "loss": 3.6482,
+      "loss": 3.6233,
       "step": 16600
     },
     {
       "epoch": 1.7920568291895382,
-      "grad_norm": 0.5653012990951538,
+      "grad_norm": 0.5894352197647095,
       "learning_rate": 0.0004930395431526775,
-      "loss": 3.6621,
+      "loss": 3.6491,
       "step": 16650
     },
     {
       "epoch": 1.7974383812291466,
-      "grad_norm": 0.5843732357025146,
-      "learning_rate": 0.0004927163021226161,
-      "loss": 3.6607,
+      "grad_norm": 0.5581018328666687,
+      "learning_rate": 0.0004927227669432173,
+      "loss": 3.6589,
       "step": 16700
     },
     {
       "epoch": 1.8028199332687547,
-      "grad_norm": 0.622482180595398,
-      "learning_rate": 0.0004923930610925547,
-      "loss": 3.6551,
+      "grad_norm": 0.5879452228546143,
+      "learning_rate": 0.0004923995259131558,
+      "loss": 3.6392,
       "step": 16750
     },
     {
       "epoch": 1.8082014853083628,
-      "grad_norm": 0.5526164770126343,
-      "learning_rate": 0.0004920698200624932,
-      "loss": 3.6533,
+      "grad_norm": 0.5551671385765076,
+      "learning_rate": 0.0004920762848830945,
+      "loss": 3.6338,
       "step": 16800
     },
     {
       "epoch": 1.813583037347971,
-      "grad_norm": 0.6223166584968567,
-      "learning_rate": 0.0004917465790324317,
-      "loss": 3.6685,
+      "grad_norm": 0.5505126714706421,
+      "learning_rate": 0.000491753043853033,
+      "loss": 3.6637,
       "step": 16850
     },
     {
       "epoch": 1.8189645893875794,
-      "grad_norm": 0.6319237947463989,
-      "learning_rate": 0.0004914233380023704,
-      "loss": 3.6679,
+      "grad_norm": 0.5661875605583191,
+      "learning_rate": 0.0004914298028229717,
+      "loss": 3.6596,
       "step": 16900
     },
     {
       "epoch": 1.8243461414271875,
-      "grad_norm": 0.5310279726982117,
-      "learning_rate": 0.0004911000969723089,
-      "loss": 3.6702,
+      "grad_norm": 0.5683614611625671,
+      "learning_rate": 0.0004911065617929102,
+      "loss": 3.6614,
       "step": 16950
     },
     {
       "epoch": 1.8297276934667959,
-      "grad_norm": 0.5771204233169556,
-      "learning_rate": 0.0004907768559422476,
-      "loss": 3.6535,
+      "grad_norm": 0.5729179978370667,
+      "learning_rate": 0.0004907833207628487,
+      "loss": 3.6528,
       "step": 17000
     },
     {
       "epoch": 1.8297276934667959,
-      "eval_accuracy": 0.3584541345194462,
-      "eval_loss": 3.609053373336792,
-      "eval_runtime": 183.1048,
-      "eval_samples_per_second": 98.364,
-      "eval_steps_per_second": 6.149,
+      "eval_accuracy": 0.35874163006060333,
+      "eval_loss": 3.6047439575195312,
+      "eval_runtime": 181.5584,
+      "eval_samples_per_second": 99.202,
+      "eval_steps_per_second": 6.202,
       "step": 17000
     },
     {
       "epoch": 1.8351092455064042,
-      "grad_norm": 0.6009894609451294,
-      "learning_rate": 0.0004904536149121861,
-      "loss": 3.6407,
+      "grad_norm": 0.5084844827651978,
+      "learning_rate": 0.0004904600797327874,
+      "loss": 3.6347,
       "step": 17050
     },
     {
       "epoch": 1.8404907975460123,
-      "grad_norm": 0.5864176750183105,
-      "learning_rate": 0.0004901303738821248,
-      "loss": 3.6577,
+      "grad_norm": 0.5654804706573486,
+      "learning_rate": 0.0004901368387027259,
+      "loss": 3.6496,
       "step": 17100
     },
     {
       "epoch": 1.8458723495856204,
-      "grad_norm": 0.5710988640785217,
+      "grad_norm": 0.5467087626457214,
       "learning_rate": 0.0004898135976726646,
-      "loss": 3.6456,
+      "loss": 3.6625,
       "step": 17150
     },
     {
       "epoch": 1.8512539016252287,
-      "grad_norm": 0.6018273830413818,
+      "grad_norm": 0.5446532368659973,
       "learning_rate": 0.0004894903566426031,
-      "loss": 3.6359,
+      "loss": 3.6412,
       "step": 17200
     },
     {
       "epoch": 1.856635453664837,
-      "grad_norm": 0.613945484161377,
+      "grad_norm": 0.543746292591095,
       "learning_rate": 0.0004891671156125417,
-      "loss": 3.6632,
+      "loss": 3.6485,
       "step": 17250
     },
     {
       "epoch": 1.8620170057044452,
-      "grad_norm": 0.5496900677680969,
+      "grad_norm": 0.5890633463859558,
       "learning_rate": 0.0004888438745824803,
-      "loss": 3.6434,
+      "loss": 3.6472,
       "step": 17300
     },
     {
       "epoch": 1.8673985577440533,
-      "grad_norm": 0.5686019659042358,
+      "grad_norm": 0.6070407032966614,
       "learning_rate": 0.0004885206335524189,
-      "loss": 3.6718,
+      "loss": 3.6509,
       "step": 17350
     },
     {
       "epoch": 1.8727801097836616,
-      "grad_norm": 0.5641106963157654,
+      "grad_norm": 0.601956307888031,
       "learning_rate": 0.0004881973925223575,
-      "loss": 3.6485,
+      "loss": 3.6566,
       "step": 17400
     },
     {
       "epoch": 1.87816166182327,
-      "grad_norm": 0.5751502513885498,
+      "grad_norm": 0.6081631183624268,
       "learning_rate": 0.00048787415149229604,
-      "loss": 3.6619,
+      "loss": 3.6482,
       "step": 17450
     },
     {
       "epoch": 1.883543213862878,
-      "grad_norm": 0.5386393666267395,
+      "grad_norm": 0.5423747301101685,
       "learning_rate": 0.00048755091046223464,
-      "loss": 3.6627,
+      "loss": 3.6429,
       "step": 17500
     },
     {
       "epoch": 1.8889247659024861,
-      "grad_norm": 0.5446299910545349,
+      "grad_norm": 0.5485900044441223,
       "learning_rate": 0.0004872276694321732,
-      "loss": 3.6486,
+      "loss": 3.6501,
       "step": 17550
     },
     {
       "epoch": 1.8943063179420945,
-      "grad_norm": 0.5315663814544678,
+      "grad_norm": 0.5473901629447937,
       "learning_rate": 0.00048690442840211177,
-      "loss": 3.6565,
+      "loss": 3.6573,
       "step": 17600
     },
     {
       "epoch": 1.8996878699817028,
-      "grad_norm": 0.5348063111305237,
+      "grad_norm": 0.580836832523346,
       "learning_rate": 0.0004865811873720504,
-      "loss": 3.6573,
+      "loss": 3.6427,
       "step": 17650
     },
     {
       "epoch": 1.905069422021311,
-      "grad_norm": 0.5544908046722412,
+      "grad_norm": 0.6399887204170227,
       "learning_rate": 0.00048625794634198896,
-      "loss": 3.6473,
+      "loss": 3.6343,
       "step": 17700
     },
     {
       "epoch": 1.910450974060919,
-      "grad_norm": 0.5778284668922424,
+      "grad_norm": 0.5329297184944153,
       "learning_rate": 0.00048593470531192756,
-      "loss": 3.6594,
+      "loss": 3.6444,
       "step": 17750
     },
     {
       "epoch": 1.9158325261005273,
-      "grad_norm": 0.566500723361969,
+      "grad_norm": 0.5967398881912231,
       "learning_rate": 0.00048561146428186615,
-      "loss": 3.6411,
+      "loss": 3.6328,
       "step": 17800
     },
     {
       "epoch": 1.9212140781401357,
-      "grad_norm": 0.5555658340454102,
+      "grad_norm": 0.5948544144630432,
       "learning_rate": 0.0004852882232518047,
-      "loss": 3.6261,
+      "loss": 3.6232,
       "step": 17850
     },
     {
       "epoch": 1.9265956301797438,
-      "grad_norm": 0.5867842435836792,
+      "grad_norm": 0.5533584356307983,
       "learning_rate": 0.00048496498222174334,
-      "loss": 3.6291,
+      "loss": 3.6326,
       "step": 17900
     },
     {
       "epoch": 1.931977182219352,
-      "grad_norm": 0.5909476280212402,
+      "grad_norm": 0.5360528230667114,
       "learning_rate": 0.00048464174119168193,
-      "loss": 3.6282,
+      "loss": 3.6105,
       "step": 17950
     },
     {
       "epoch": 1.9373587342589604,
-      "grad_norm": 0.642209529876709,
+      "grad_norm": 0.5491250157356262,
       "learning_rate": 0.0004843185001616205,
-      "loss": 3.6528,
+      "loss": 3.6527,
       "step": 18000
     },
     {
       "epoch": 1.9373587342589604,
-      "eval_accuracy": 0.3601386888921219,
-      "eval_loss": 3.5923776626586914,
-      "eval_runtime": 182.6986,
-      "eval_samples_per_second": 98.583,
-      "eval_steps_per_second": 6.163,
+      "eval_accuracy": 0.3602193093348803,
+      "eval_loss": 3.5871007442474365,
+      "eval_runtime": 181.7702,
+      "eval_samples_per_second": 99.087,
+      "eval_steps_per_second": 6.195,
       "step": 18000
     },
     {
       "epoch": 1.9427402862985685,
-      "grad_norm": 0.5417137742042542,
+      "grad_norm": 0.5205147862434387,
       "learning_rate": 0.00048399525913155907,
-      "loss": 3.6509,
+      "loss": 3.6512,
       "step": 18050
     },
     {
       "epoch": 1.9481218383381766,
-      "grad_norm": 0.5468534231185913,
+      "grad_norm": 0.5882537364959717,
       "learning_rate": 0.0004836720181014976,
-      "loss": 3.6373,
+      "loss": 3.62,
       "step": 18100
     },
     {
       "epoch": 1.953503390377785,
-      "grad_norm": 0.5604886412620544,
+      "grad_norm": 0.5908693671226501,
       "learning_rate": 0.0004833487770714362,
-      "loss": 3.6485,
+      "loss": 3.6263,
       "step": 18150
     },
     {
       "epoch": 1.9588849424173933,
-      "grad_norm": 0.5776407718658447,
+      "grad_norm": 0.5874225497245789,
       "learning_rate": 0.00048302553604137485,
-      "loss": 3.6288,
+      "loss": 3.6169,
       "step": 18200
     },
     {
       "epoch": 1.9642664944570014,
-      "grad_norm": 0.6095754504203796,
+      "grad_norm": 0.5783534646034241,
       "learning_rate": 0.0004827022950113134,
-      "loss": 3.6571,
+      "loss": 3.6485,
       "step": 18250
     },
     {
       "epoch": 1.9696480464966095,
-      "grad_norm": 0.5740944147109985,
+      "grad_norm": 0.6188249588012695,
       "learning_rate": 0.000482379053981252,
-      "loss": 3.6374,
+      "loss": 3.6409,
       "step": 18300
     },
     {
       "epoch": 1.9750295985362178,
-      "grad_norm": 0.5488967299461365,
+      "grad_norm": 0.5157595276832581,
       "learning_rate": 0.0004820558129511906,
-      "loss": 3.6425,
+      "loss": 3.6357,
       "step": 18350
     },
     {
       "epoch": 1.9804111505758262,
-      "grad_norm": 0.5918818712234497,
+      "grad_norm": 0.6100007891654968,
       "learning_rate": 0.0004817325719211291,
-      "loss": 3.6471,
+      "loss": 3.6362,
       "step": 18400
     },
     {
       "epoch": 1.9857927026154343,
-      "grad_norm": 0.5412439703941345,
+      "grad_norm": 0.5755199193954468,
       "learning_rate": 0.0004814093308910677,
-      "loss": 3.6463,
+      "loss": 3.6305,
       "step": 18450
     },
     {
       "epoch": 1.9911742546550424,
-      "grad_norm": 0.5630549192428589,
+      "grad_norm": 0.5345621109008789,
       "learning_rate": 0.00048108608986100637,
-      "loss": 3.6531,
+      "loss": 3.6497,
       "step": 18500
     },
     {
       "epoch": 1.9965558066946507,
-      "grad_norm": 0.5589700937271118,
+      "grad_norm": 0.5259223580360413,
       "learning_rate": 0.0004807628488309449,
-      "loss": 3.65,
+      "loss": 3.6421,
       "step": 18550
     },
     {
       "epoch": 2.001937358734259,
-      "grad_norm": 0.5413640141487122,
+      "grad_norm": 0.5224030613899231,
       "learning_rate": 0.0004804396078008835,
-      "loss": 3.6016,
+      "loss": 3.6069,
       "step": 18600
     },
     {
       "epoch": 2.007318910773867,
-      "grad_norm": 0.5609384179115295,
+      "grad_norm": 0.5612570643424988,
       "learning_rate": 0.00048011636677082204,
-      "loss": 3.5453,
+      "loss": 3.5469,
       "step": 18650
     },
     {
       "epoch": 2.0127004628134753,
-      "grad_norm": 0.6163759827613831,
+      "grad_norm": 0.6124724745750427,
       "learning_rate": 0.00047979312574076064,
-      "loss": 3.5521,
+      "loss": 3.5346,
       "step": 18700
     },
     {
       "epoch": 2.018082014853084,
-      "grad_norm": 0.5741806626319885,
+      "grad_norm": 0.5609890818595886,
       "learning_rate": 0.0004794698847106992,
-      "loss": 3.5383,
+      "loss": 3.5332,
       "step": 18750
     },
     {
       "epoch": 2.023463566892692,
-      "grad_norm": 0.6155529022216797,
-      "learning_rate": 0.0004791466436806378,
-      "loss": 3.5698,
+      "grad_norm": 0.5898619890213013,
+      "learning_rate": 0.00047915310850123904,
+      "loss": 3.5576,
       "step": 18800
     },
     {
       "epoch": 2.0288451189323,
-      "grad_norm": 0.6185212731361389,
-      "learning_rate": 0.0004788234026505764,
-      "loss": 3.5489,
+      "grad_norm": 0.5770263671875,
+      "learning_rate": 0.0004788298674711777,
+      "loss": 3.5507,
       "step": 18850
     },
     {
       "epoch": 2.034226670971908,
-      "grad_norm": 0.5259607434272766,
-      "learning_rate": 0.00047850016162051496,
-      "loss": 3.5425,
+      "grad_norm": 0.593312680721283,
+      "learning_rate": 0.00047850662644111623,
+      "loss": 3.5316,
       "step": 18900
     },
     {
       "epoch": 2.0396082230115167,
-      "grad_norm": 0.5763450264930725,
-      "learning_rate": 0.00047817692059045356,
-      "loss": 3.5663,
+      "grad_norm": 0.6398000121116638,
+      "learning_rate": 0.0004781833854110548,
+      "loss": 3.5527,
       "step": 18950
     },
     {
       "epoch": 2.044989775051125,
-      "grad_norm": 0.5624204277992249,
-      "learning_rate": 0.00047785367956039215,
-      "loss": 3.5646,
+      "grad_norm": 0.5566432476043701,
+      "learning_rate": 0.00047786014438099336,
+      "loss": 3.5538,
       "step": 19000
     },
     {
       "epoch": 2.044989775051125,
-      "eval_accuracy": 0.36152814202149347,
-      "eval_loss": 3.5838420391082764,
-      "eval_runtime": 183.3409,
-      "eval_samples_per_second": 98.238,
-      "eval_steps_per_second": 6.142,
+      "eval_accuracy": 0.3617640274409387,
+      "eval_loss": 3.5782222747802734,
+      "eval_runtime": 181.5033,
+      "eval_samples_per_second": 99.232,
+      "eval_steps_per_second": 6.204,
       "step": 19000
     },
     {
       "epoch": 2.050371327090733,
-      "grad_norm": 0.5626105666160583,
-      "learning_rate": 0.00047753043853033075,
-      "loss": 3.5464,
+      "grad_norm": 0.5329643487930298,
+      "learning_rate": 0.00047753690335093196,
+      "loss": 3.5432,
       "step": 19050
     },
     {
       "epoch": 2.055752879130341,
-      "grad_norm": 0.5418869853019714,
-      "learning_rate": 0.00047720719750026934,
-      "loss": 3.5534,
+      "grad_norm": 0.6505568027496338,
+      "learning_rate": 0.00047721366232087055,
+      "loss": 3.5421,
       "step": 19100
     },
     {
       "epoch": 2.0611344311699495,
-      "grad_norm": 0.6087026000022888,
+      "grad_norm": 0.5331689715385437,
       "learning_rate": 0.00047689042129080915,
-      "loss": 3.5573,
+      "loss": 3.5491,
       "step": 19150
     },
     {
       "epoch": 2.0665159832095576,
-      "grad_norm": 0.5950330495834351,
+      "grad_norm": 0.544685423374176,
       "learning_rate": 0.00047656718026074774,
-      "loss": 3.5528,
+      "loss": 3.5569,
       "step": 19200
     },
     {
       "epoch": 2.0718975352491658,
-      "grad_norm": 0.5562347769737244,
+      "grad_norm": 0.5895106196403503,
       "learning_rate": 0.00047624393923068634,
-      "loss": 3.5703,
+      "loss": 3.5533,
       "step": 19250
     },
     {
       "epoch": 2.0772790872887743,
-      "grad_norm": 0.5437284111976624,
+      "grad_norm": 0.5611186623573303,
       "learning_rate": 0.0004759206982006249,
-      "loss": 3.5563,
+      "loss": 3.558,
       "step": 19300
     },
     {
       "epoch": 2.0826606393283824,
-      "grad_norm": 0.6348011493682861,
+      "grad_norm": 0.5915340781211853,
       "learning_rate": 0.00047559745717056347,
-      "loss": 3.5651,
+      "loss": 3.5563,
       "step": 19350
     },
     {
       "epoch": 2.0880421913679905,
-      "grad_norm": 0.5337033271789551,
+      "grad_norm": 0.5206013917922974,
       "learning_rate": 0.000475274216140502,
-      "loss": 3.556,
+      "loss": 3.5542,
       "step": 19400
     },
     {
       "epoch": 2.0934237434075986,
-      "grad_norm": 0.6107613444328308,
+      "grad_norm": 0.5762017369270325,
       "learning_rate": 0.00047495097511044066,
-      "loss": 3.5584,
+      "loss": 3.5514,
       "step": 19450
     },
     {
       "epoch": 2.098805295447207,
-      "grad_norm": 0.6084286570549011,
+      "grad_norm": 0.587688148021698,
       "learning_rate": 0.00047462773408037925,
-      "loss": 3.5613,
+      "loss": 3.5501,
       "step": 19500
     },
     {
       "epoch": 2.1041868474868153,
-      "grad_norm": 0.5908589959144592,
+      "grad_norm": 0.5966100692749023,
       "learning_rate": 0.0004743044930503178,
-      "loss": 3.5732,
+      "loss": 3.5649,
       "step": 19550
     },
     {
       "epoch": 2.1095683995264234,
-      "grad_norm": 0.5735676884651184,
+      "grad_norm": 0.5434450507164001,
       "learning_rate": 0.0004739812520202564,
-      "loss": 3.5512,
+      "loss": 3.5631,
       "step": 19600
     },
     {
       "epoch": 2.1149499515660315,
-      "grad_norm": 0.5397831201553345,
+      "grad_norm": 0.6018627882003784,
       "learning_rate": 0.000473658010990195,
-      "loss": 3.5579,
+      "loss": 3.5618,
       "step": 19650
     },
     {
       "epoch": 2.12033150360564,
-      "grad_norm": 0.6003082394599915,
+      "grad_norm": 0.6211711168289185,
       "learning_rate": 0.0004733347699601336,
-      "loss": 3.5704,
+      "loss": 3.5583,
       "step": 19700
     },
     {
       "epoch": 2.125713055645248,
-      "grad_norm": 0.6014171242713928,
+      "grad_norm": 0.5908129811286926,
       "learning_rate": 0.0004730115289300722,
-      "loss": 3.5688,
+      "loss": 3.5681,
       "step": 19750
     },
     {
       "epoch": 2.1310946076848563,
-      "grad_norm": 0.5529822707176208,
+      "grad_norm": 0.5816216468811035,
       "learning_rate": 0.00047268828790001077,
-      "loss": 3.5427,
+      "loss": 3.5442,
       "step": 19800
     },
     {
       "epoch": 2.1364761597244644,
-      "grad_norm": 0.6028132438659668,
+      "grad_norm": 0.5796759128570557,
       "learning_rate": 0.0004723650468699493,
-      "loss": 3.5596,
+      "loss": 3.5512,
       "step": 19850
     },
     {
       "epoch": 2.141857711764073,
-      "grad_norm": 0.5742948651313782,
+      "grad_norm": 0.5867259502410889,
       "learning_rate": 0.0004720418058398879,
-      "loss": 3.5591,
+      "loss": 3.566,
       "step": 19900
     },
     {
       "epoch": 2.147239263803681,
-      "grad_norm": 0.6175876259803772,
+      "grad_norm": 0.605074405670166,
       "learning_rate": 0.00047171856480982644,
-      "loss": 3.5649,
+      "loss": 3.5599,
       "step": 19950
     },
     {
       "epoch": 2.152620815843289,
-      "grad_norm": 0.5835720300674438,
+      "grad_norm": 0.6474159955978394,
       "learning_rate": 0.0004713953237797651,
-      "loss": 3.5654,
+      "loss": 3.5657,
       "step": 20000
     },
     {
       "epoch": 2.152620815843289,
-      "eval_accuracy": 0.36219320634780594,
-      "eval_loss": 3.5732574462890625,
-      "eval_runtime": 183.9589,
-      "eval_samples_per_second": 97.908,
-      "eval_steps_per_second": 6.121,
+      "eval_accuracy": 0.36281448356032914,
+      "eval_loss": 3.5691561698913574,
+      "eval_runtime": 182.1501,
+      "eval_samples_per_second": 98.88,
+      "eval_steps_per_second": 6.182,
       "step": 20000
     },
     {
       "epoch": 2.1580023678828972,
-      "grad_norm": 0.5701449513435364,
+      "grad_norm": 0.5862100124359131,
       "learning_rate": 0.0004710720827497037,
-      "loss": 3.5597,
+      "loss": 3.5535,
       "step": 20050
     },
     {
       "epoch": 2.163383919922506,
-      "grad_norm": 0.5607191920280457,
+      "grad_norm": 0.5728433728218079,
       "learning_rate": 0.00047074884171964223,
-      "loss": 3.5617,
+      "loss": 3.5615,
       "step": 20100
     },
     {
       "epoch": 2.168765471962114,
-      "grad_norm": 0.5911417603492737,
+      "grad_norm": 0.5794762969017029,
       "learning_rate": 0.0004704256006895808,
-      "loss": 3.5635,
+      "loss": 3.555,
       "step": 20150
     },
     {
       "epoch": 2.174147024001722,
-      "grad_norm": 0.5466926097869873,
+      "grad_norm": 0.5577318072319031,
       "learning_rate": 0.00047010235965951936,
-      "loss": 3.5751,
+      "loss": 3.5496,
       "step": 20200
     },
     {
       "epoch": 2.1795285760413305,
-      "grad_norm": 0.58399897813797,
+      "grad_norm": 0.5643510222434998,
       "learning_rate": 0.00046977911862945796,
-      "loss": 3.5768,
+      "loss": 3.5498,
       "step": 20250
     },
     {
       "epoch": 2.1849101280809387,
-      "grad_norm": 0.6332233548164368,
+      "grad_norm": 0.5473456382751465,
       "learning_rate": 0.0004694558775993966,
-      "loss": 3.5679,
+      "loss": 3.55,
       "step": 20300
     },
     {
       "epoch": 2.1902916801205468,
-      "grad_norm": 0.5964486598968506,
+      "grad_norm": 0.5973044633865356,
       "learning_rate": 0.00046913263656933515,
-      "loss": 3.5699,
+      "loss": 3.553,
       "step": 20350
     },
     {
       "epoch": 2.195673232160155,
-      "grad_norm": 0.5788668394088745,
+      "grad_norm": 0.5803256630897522,
       "learning_rate": 0.00046880939553927374,
-      "loss": 3.5586,
+      "loss": 3.555,
       "step": 20400
     },
     {
       "epoch": 2.2010547841997634,
-      "grad_norm": 0.5830016732215881,
+      "grad_norm": 0.5904147028923035,
       "learning_rate": 0.00046848615450921234,
-      "loss": 3.5675,
+      "loss": 3.5559,
       "step": 20450
     },
     {
       "epoch": 2.2064363362393715,
-      "grad_norm": 0.6435354948043823,
+      "grad_norm": 0.5470393300056458,
       "learning_rate": 0.0004681629134791509,
-      "loss": 3.572,
+      "loss": 3.558,
       "step": 20500
     },
     {
       "epoch": 2.2118178882789796,
-      "grad_norm": 0.5667049884796143,
+      "grad_norm": 0.5575466156005859,
       "learning_rate": 0.00046783967244908947,
-      "loss": 3.5638,
+      "loss": 3.5567,
       "step": 20550
     },
     {
       "epoch": 2.2171994403185877,
-      "grad_norm": 0.6753138899803162,
+      "grad_norm": 0.5824530720710754,
       "learning_rate": 0.0004675164314190281,
-      "loss": 3.5719,
+      "loss": 3.5689,
       "step": 20600
     },
     {
       "epoch": 2.2225809923581963,
-      "grad_norm": 0.6224473118782043,
+      "grad_norm": 0.5693963170051575,
       "learning_rate": 0.00046719319038896666,
-      "loss": 3.5729,
+      "loss": 3.5585,
       "step": 20650
     },
     {
       "epoch": 2.2279625443978044,
-      "grad_norm": 0.6385303735733032,
+      "grad_norm": 0.6045997142791748,
       "learning_rate": 0.00046686994935890526,
-      "loss": 3.5713,
+      "loss": 3.5741,
       "step": 20700
     },
     {
       "epoch": 2.2333440964374125,
-      "grad_norm": 0.5930754542350769,
+      "grad_norm": 0.5361005067825317,
       "learning_rate": 0.0004665467083288438,
-      "loss": 3.5553,
+      "loss": 3.5738,
       "step": 20750
     },
     {
       "epoch": 2.2387256484770206,
-      "grad_norm": 0.5501213669776917,
+      "grad_norm": 0.5668167471885681,
       "learning_rate": 0.0004662234672987824,
-      "loss": 3.5775,
+      "loss": 3.5644,
       "step": 20800
     },
     {
       "epoch": 2.244107200516629,
-      "grad_norm": 0.5862070322036743,
+      "grad_norm": 0.5758764743804932,
       "learning_rate": 0.00046590022626872104,
-      "loss": 3.5752,
+      "loss": 3.5608,
       "step": 20850
     },
     {
       "epoch": 2.2494887525562373,
-      "grad_norm": 0.5672935247421265,
+      "grad_norm": 0.6167986392974854,
       "learning_rate": 0.0004655769852386596,
-      "loss": 3.5702,
+      "loss": 3.5714,
       "step": 20900
     },
     {
       "epoch": 2.2548703045958454,
-      "grad_norm": 0.564063310623169,
+      "grad_norm": 0.6584091782569885,
       "learning_rate": 0.0004652537442085982,
-      "loss": 3.5571,
+      "loss": 3.5495,
       "step": 20950
     },
     {
       "epoch": 2.2602518566354535,
-      "grad_norm": 0.6813990473747253,
+      "grad_norm": 0.6173479557037354,
       "learning_rate": 0.00046493050317853677,
-      "loss": 3.5613,
+      "loss": 3.573,
       "step": 21000
     },
     {
       "epoch": 2.2602518566354535,
-      "eval_accuracy": 0.36378736151782,
-      "eval_loss": 3.5629472732543945,
-      "eval_runtime": 183.0527,
-      "eval_samples_per_second": 98.392,
-      "eval_steps_per_second": 6.151,
+      "eval_accuracy": 0.36381854489662274,
+      "eval_loss": 3.5603933334350586,
+      "eval_runtime": 181.1814,
+      "eval_samples_per_second": 99.409,
+      "eval_steps_per_second": 6.215,
       "step": 21000
     },
     {
       "epoch": 2.265633408675062,
-      "grad_norm": 0.5663150548934937,
+      "grad_norm": 0.5912428498268127,
       "learning_rate": 0.0004646072621484753,
-      "loss": 3.5268,
+      "loss": 3.5507,
       "step": 21050
     },
     {
       "epoch": 2.27101496071467,
-      "grad_norm": 0.5659851431846619,
-      "learning_rate": 0.00046429048593901517,
-      "loss": 3.5753,
+      "grad_norm": 0.5723901987075806,
+      "learning_rate": 0.0004642840211184139,
+      "loss": 3.5623,
       "step": 21100
     },
     {
       "epoch": 2.2763965127542782,
-      "grad_norm": 0.5690809488296509,
-      "learning_rate": 0.0004639672449089537,
-      "loss": 3.5542,
+      "grad_norm": 0.5158131718635559,
+      "learning_rate": 0.00046396078008835255,
+      "loss": 3.5527,
       "step": 21150
     },
     {
       "epoch": 2.281778064793887,
-      "grad_norm": 0.6252328157424927,
+      "grad_norm": 0.6142178177833557,
       "learning_rate": 0.0004636440038788923,
-      "loss": 3.565,
+      "loss": 3.5571,
       "step": 21200
     },
     {
       "epoch": 2.287159616833495,
-      "grad_norm": 0.5633666515350342,
+      "grad_norm": 0.5338762402534485,
       "learning_rate": 0.00046332076284883095,
-      "loss": 3.5501,
+      "loss": 3.5527,
       "step": 21250
     },
     {
       "epoch": 2.292541168873103,
-      "grad_norm": 0.634242057800293,
+      "grad_norm": 0.5781612992286682,
       "learning_rate": 0.0004629975218187695,
-      "loss": 3.5546,
+      "loss": 3.5531,
       "step": 21300
     },
     {
       "epoch": 2.297922720912711,
-      "grad_norm": 0.5788732767105103,
+      "grad_norm": 0.5228679776191711,
       "learning_rate": 0.0004626742807887081,
-      "loss": 3.5609,
+      "loss": 3.5574,
       "step": 21350
     },
     {
       "epoch": 2.303304272952319,
-      "grad_norm": 0.5648044347763062,
+      "grad_norm": 0.620266854763031,
       "learning_rate": 0.00046235103975864663,
-      "loss": 3.546,
+      "loss": 3.5499,
       "step": 21400
     },
     {
       "epoch": 2.3086858249919278,
-      "grad_norm": 0.6103492379188538,
+      "grad_norm": 0.5789138078689575,
       "learning_rate": 0.0004620277987285852,
-      "loss": 3.5766,
+      "loss": 3.576,
       "step": 21450
     },
     {
       "epoch": 2.314067377031536,
-      "grad_norm": 0.6276610493659973,
+      "grad_norm": 0.5878793001174927,
       "learning_rate": 0.0004617045576985239,
-      "loss": 3.5737,
+      "loss": 3.5756,
       "step": 21500
     },
     {
       "epoch": 2.319448929071144,
-      "grad_norm": 0.5594583749771118,
+      "grad_norm": 0.5757933855056763,
       "learning_rate": 0.0004613813166684624,
-      "loss": 3.5475,
+      "loss": 3.5534,
       "step": 21550
     },
     {
       "epoch": 2.3248304811107525,
-      "grad_norm": 0.5839714407920837,
+      "grad_norm": 0.6496496200561523,
       "learning_rate": 0.000461058075638401,
-      "loss": 3.5706,
+      "loss": 3.5603,
       "step": 21600
     },
     {
       "epoch": 2.3302120331503606,
-      "grad_norm": 0.5914338827133179,
+      "grad_norm": 0.5533931851387024,
       "learning_rate": 0.00046073483460833955,
-      "loss": 3.5553,
+      "loss": 3.5525,
       "step": 21650
     },
     {
       "epoch": 2.3355935851899687,
-      "grad_norm": 0.5473167896270752,
+      "grad_norm": 0.5627120137214661,
       "learning_rate": 0.00046041159357827814,
-      "loss": 3.5412,
+      "loss": 3.5465,
       "step": 21700
     },
     {
       "epoch": 2.340975137229577,
-      "grad_norm": 0.5862297415733337,
+      "grad_norm": 0.55203777551651,
       "learning_rate": 0.00046008835254821674,
-      "loss": 3.5511,
+      "loss": 3.5588,
       "step": 21750
     },
     {
       "epoch": 2.3463566892691854,
-      "grad_norm": 0.5849961042404175,
+      "grad_norm": 0.5614556670188904,
       "learning_rate": 0.0004597651115181554,
-      "loss": 3.5424,
+      "loss": 3.5425,
       "step": 21800
     },
     {
       "epoch": 2.3517382413087935,
-      "grad_norm": 0.6141690015792847,
+      "grad_norm": 0.5911089777946472,
       "learning_rate": 0.00045944187048809393,
-      "loss": 3.5617,
+      "loss": 3.5481,
       "step": 21850
     },
     {
       "epoch": 2.3571197933484016,
-      "grad_norm": 0.5442442893981934,
+      "grad_norm": 0.5486292243003845,
       "learning_rate": 0.0004591186294580325,
-      "loss": 3.5594,
+      "loss": 3.552,
       "step": 21900
     },
     {
       "epoch": 2.3625013453880097,
-      "grad_norm": 0.5921179056167603,
+      "grad_norm": 0.547752320766449,
       "learning_rate": 0.00045879538842797106,
-      "loss": 3.5472,
+      "loss": 3.559,
       "step": 21950
     },
     {
       "epoch": 2.3678828974276183,
-      "grad_norm": 0.5729749798774719,
+      "grad_norm": 0.561741054058075,
       "learning_rate": 0.00045847214739790966,
-      "loss": 3.5588,
+      "loss": 3.5483,
       "step": 22000
     },
     {
       "epoch": 2.3678828974276183,
-      "eval_accuracy": 0.36438647354122905,
-      "eval_loss": 3.5525951385498047,
-      "eval_runtime": 183.627,
-      "eval_samples_per_second": 98.085,
-      "eval_steps_per_second": 6.132,
+      "eval_accuracy": 0.36455325572402425,
+      "eval_loss": 3.550053596496582,
+      "eval_runtime": 181.7456,
+      "eval_samples_per_second": 99.1,
+      "eval_steps_per_second": 6.195,
       "step": 22000
     },
     {
       "epoch": 2.3732644494672264,
-      "grad_norm": 0.5790433287620544,
+      "grad_norm": 0.5793853402137756,
       "learning_rate": 0.0004581489063678482,
-      "loss": 3.5655,
+      "loss": 3.5462,
       "step": 22050
     },
     {
       "epoch": 2.3786460015068345,
-      "grad_norm": 0.5972291827201843,
+      "grad_norm": 0.575937807559967,
       "learning_rate": 0.00045782566533778685,
-      "loss": 3.553,
+      "loss": 3.5618,
       "step": 22100
     },
     {
       "epoch": 2.384027553546443,
-      "grad_norm": 0.6700719594955444,
+      "grad_norm": 0.6419012546539307,
       "learning_rate": 0.00045750242430772544,
-      "loss": 3.5554,
+      "loss": 3.5547,
       "step": 22150
     },
     {
       "epoch": 2.389409105586051,
-      "grad_norm": 0.5736136436462402,
+      "grad_norm": 0.6173324584960938,
       "learning_rate": 0.000457179183277664,
-      "loss": 3.5505,
+      "loss": 3.5747,
       "step": 22200
     },
     {
       "epoch": 2.3947906576256592,
-      "grad_norm": 0.5969544053077698,
+      "grad_norm": 0.5861020684242249,
       "learning_rate": 0.0004568559422476026,
-      "loss": 3.5582,
+      "loss": 3.5506,
       "step": 22250
     },
     {
       "epoch": 2.4001722096652673,
-      "grad_norm": 0.5612258911132812,
+      "grad_norm": 0.6053194403648376,
       "learning_rate": 0.00045653270121754117,
-      "loss": 3.552,
+      "loss": 3.5563,
       "step": 22300
     },
     {
       "epoch": 2.4055537617048754,
-      "grad_norm": 0.5993139743804932,
+      "grad_norm": 0.623187780380249,
       "learning_rate": 0.0004562094601874797,
-      "loss": 3.5753,
+      "loss": 3.5688,
       "step": 22350
     },
     {
       "epoch": 2.410935313744484,
-      "grad_norm": 0.5695291757583618,
+      "grad_norm": 0.6081986427307129,
       "learning_rate": 0.00045588621915741836,
-      "loss": 3.5636,
+      "loss": 3.5464,
       "step": 22400
     },
     {
       "epoch": 2.416316865784092,
-      "grad_norm": 0.6173685193061829,
+      "grad_norm": 0.5826542973518372,
       "learning_rate": 0.00045556297812735696,
-      "loss": 3.5615,
+      "loss": 3.5592,
       "step": 22450
     },
     {
       "epoch": 2.4216984178237,
-      "grad_norm": 0.5470847487449646,
+      "grad_norm": 0.5407750606536865,
       "learning_rate": 0.0004552397370972955,
-      "loss": 3.5524,
+      "loss": 3.5446,
       "step": 22500
     },
     {
       "epoch": 2.4270799698633088,
-      "grad_norm": 0.5483660697937012,
+      "grad_norm": 0.5309058427810669,
       "learning_rate": 0.0004549164960672341,
-      "loss": 3.5562,
+      "loss": 3.5451,
       "step": 22550
     },
     {
       "epoch": 2.432461521902917,
-      "grad_norm": 0.630761444568634,
+      "grad_norm": 0.6191596388816833,
       "learning_rate": 0.00045459325503717263,
-      "loss": 3.564,
+      "loss": 3.5564,
       "step": 22600
     },
     {
       "epoch": 2.437843073942525,
-      "grad_norm": 0.5955401062965393,
+      "grad_norm": 0.6055700182914734,
       "learning_rate": 0.0004542700140071113,
-      "loss": 3.5453,
+      "loss": 3.5403,
       "step": 22650
     },
     {
       "epoch": 2.443224625982133,
-      "grad_norm": 0.5875324606895447,
+      "grad_norm": 0.5493307709693909,
       "learning_rate": 0.0004539467729770499,
-      "loss": 3.5226,
+      "loss": 3.5497,
       "step": 22700
     },
     {
       "epoch": 2.4486061780217416,
-      "grad_norm": 0.5740516185760498,
+      "grad_norm": 0.564789891242981,
       "learning_rate": 0.0004536235319469884,
-      "loss": 3.5704,
+      "loss": 3.545,
       "step": 22750
     },
     {
       "epoch": 2.4539877300613497,
-      "grad_norm": 0.5694280862808228,
+      "grad_norm": 0.6158958077430725,
       "learning_rate": 0.000453300290916927,
-      "loss": 3.5505,
+      "loss": 3.5523,
       "step": 22800
     },
     {
       "epoch": 2.459369282100958,
-      "grad_norm": 0.6105132699012756,
+      "grad_norm": 0.5858321785926819,
       "learning_rate": 0.0004529770498868656,
-      "loss": 3.5632,
+      "loss": 3.5643,
       "step": 22850
     },
     {
       "epoch": 2.464750834140566,
-      "grad_norm": 0.6120337843894958,
+      "grad_norm": 0.570465624332428,
       "learning_rate": 0.00045265380885680414,
-      "loss": 3.5523,
+      "loss": 3.5495,
       "step": 22900
     },
     {
       "epoch": 2.4701323861801745,
-      "grad_norm": 0.5750212669372559,
+      "grad_norm": 0.589330792427063,
       "learning_rate": 0.0004523305678267428,
-      "loss": 3.5434,
+      "loss": 3.5637,
       "step": 22950
     },
     {
       "epoch": 2.4755139382197826,
-      "grad_norm": 0.5812209248542786,
+      "grad_norm": 0.5753523707389832,
       "learning_rate": 0.0004520073267966814,
-      "loss": 3.5626,
+      "loss": 3.5518,
       "step": 23000
     },
     {
       "epoch": 2.4755139382197826,
-      "eval_accuracy": 0.36594868476222564,
-      "eval_loss": 3.5408027172088623,
-      "eval_runtime": 182.9812,
-      "eval_samples_per_second": 98.431,
-      "eval_steps_per_second": 6.154,
+      "eval_accuracy": 0.36554786425907804,
+      "eval_loss": 3.5397651195526123,
+      "eval_runtime": 181.0209,
+      "eval_samples_per_second": 99.497,
+      "eval_steps_per_second": 6.22,
       "step": 23000
     },
     {
       "epoch": 2.4808954902593907,
-      "grad_norm": 0.6218942403793335,
+      "grad_norm": 0.5980810523033142,
       "learning_rate": 0.00045168408576661993,
-      "loss": 3.5572,
+      "loss": 3.5644,
       "step": 23050
     },
     {
       "epoch": 2.4862770422989993,
-      "grad_norm": 0.5775257349014282,
+      "grad_norm": 0.6485756039619446,
       "learning_rate": 0.0004513608447365585,
-      "loss": 3.5484,
+      "loss": 3.5602,
       "step": 23100
     },
     {
       "epoch": 2.4916585943386074,
-      "grad_norm": 0.5762562155723572,
+      "grad_norm": 0.5862653851509094,
       "learning_rate": 0.00045103760370649706,
-      "loss": 3.5744,
+      "loss": 3.5802,
       "step": 23150
     },
     {
       "epoch": 2.4970401463782155,
-      "grad_norm": 0.5888144373893738,
-      "learning_rate": 0.00045071436267643566,
-      "loss": 3.5604,
+      "grad_norm": 0.5570921897888184,
+      "learning_rate": 0.0004507208274970369,
+      "loss": 3.5559,
       "step": 23200
     },
     {
       "epoch": 2.5024216984178236,
-      "grad_norm": 0.6638675928115845,
-      "learning_rate": 0.0004503911216463743,
-      "loss": 3.5623,
+      "grad_norm": 0.6063124537467957,
+      "learning_rate": 0.00045039758646697546,
+      "loss": 3.5462,
       "step": 23250
     },
     {
       "epoch": 2.5078032504574317,
-      "grad_norm": 0.620266318321228,
-      "learning_rate": 0.00045006788061631285,
-      "loss": 3.557,
+      "grad_norm": 0.5342347621917725,
+      "learning_rate": 0.0004500743454369141,
+      "loss": 3.5528,
       "step": 23300
     },
     {
       "epoch": 2.5131848024970402,
-      "grad_norm": 0.5920720100402832,
-      "learning_rate": 0.00044974463958625144,
-      "loss": 3.5369,
+      "grad_norm": 0.5412544012069702,
+      "learning_rate": 0.0004497511044068527,
+      "loss": 3.5389,
       "step": 23350
     },
     {
       "epoch": 2.5185663545366483,
-      "grad_norm": 0.5830405950546265,
-      "learning_rate": 0.00044942139855619004,
-      "loss": 3.5637,
+      "grad_norm": 0.6068680882453918,
+      "learning_rate": 0.00044942786337679125,
+      "loss": 3.5561,
       "step": 23400
     },
     {
       "epoch": 2.5239479065762565,
-      "grad_norm": 0.5792795419692993,
-      "learning_rate": 0.0004490981575261286,
-      "loss": 3.5551,
+      "grad_norm": 0.5819967985153198,
+      "learning_rate": 0.00044910462234672984,
+      "loss": 3.5603,
       "step": 23450
     },
     {
       "epoch": 2.529329458615865,
-      "grad_norm": 0.5703508257865906,
-      "learning_rate": 0.0004487749164960672,
-      "loss": 3.5656,
+      "grad_norm": 0.6125398278236389,
+      "learning_rate": 0.0004487813813166684,
+      "loss": 3.5431,
       "step": 23500
     },
     {
       "epoch": 2.534711010655473,
-      "grad_norm": 0.5803419947624207,
-      "learning_rate": 0.0004484516754660058,
-      "loss": 3.5491,
+      "grad_norm": 0.5376060009002686,
+      "learning_rate": 0.000448458140286607,
+      "loss": 3.5358,
       "step": 23550
     },
     {
       "epoch": 2.540092562695081,
-      "grad_norm": 0.6119604706764221,
-      "learning_rate": 0.00044812843443594436,
-      "loss": 3.556,
+      "grad_norm": 0.5709202289581299,
+      "learning_rate": 0.00044813489925654563,
+      "loss": 3.5621,
       "step": 23600
     },
     {
       "epoch": 2.5454741147346893,
-      "grad_norm": 0.5884591937065125,
-      "learning_rate": 0.00044780519340588296,
-      "loss": 3.5301,
+      "grad_norm": 0.568293035030365,
+      "learning_rate": 0.00044781165822648417,
+      "loss": 3.5418,
       "step": 23650
     },
     {
       "epoch": 2.550855666774298,
-      "grad_norm": 0.5345640778541565,
-      "learning_rate": 0.0004474819523758215,
-      "loss": 3.5249,
+      "grad_norm": 0.5785930156707764,
+      "learning_rate": 0.00044748841719642276,
+      "loss": 3.5416,
       "step": 23700
     },
     {
       "epoch": 2.556237218813906,
-      "grad_norm": 0.5864933729171753,
-      "learning_rate": 0.0004471587113457601,
-      "loss": 3.5468,
+      "grad_norm": 0.5561110973358154,
+      "learning_rate": 0.00044716517616636136,
+      "loss": 3.5365,
       "step": 23750
     },
     {
       "epoch": 2.561618770853514,
-      "grad_norm": 0.5660606622695923,
-      "learning_rate": 0.00044683547031569874,
-      "loss": 3.5576,
+      "grad_norm": 0.5745890736579895,
+      "learning_rate": 0.0004468419351362999,
+      "loss": 3.5404,
       "step": 23800
     },
     {
       "epoch": 2.567000322893122,
-      "grad_norm": 0.5605663061141968,
-      "learning_rate": 0.0004465122292856373,
-      "loss": 3.5404,
+      "grad_norm": 0.5342207551002502,
+      "learning_rate": 0.0004465186941062385,
+      "loss": 3.5512,
       "step": 23850
     },
     {
       "epoch": 2.5723818749327307,
-      "grad_norm": 0.6553633213043213,
-      "learning_rate": 0.0004461889882555759,
-      "loss": 3.5576,
+      "grad_norm": 0.594519317150116,
+      "learning_rate": 0.00044619545307617714,
+      "loss": 3.5628,
       "step": 23900
     },
     {
       "epoch": 2.577763426972339,
-      "grad_norm": 0.6218010783195496,
-      "learning_rate": 0.00044586574722551447,
-      "loss": 3.559,
+      "grad_norm": 0.5280244946479797,
+      "learning_rate": 0.0004458722120461157,
+      "loss": 3.5453,
       "step": 23950
     },
     {
       "epoch": 2.583144979011947,
-      "grad_norm": 0.6036173701286316,
-      "learning_rate": 0.000445542506195453,
-      "loss": 3.5623,
+      "grad_norm": 0.6097589731216431,
+      "learning_rate": 0.0004455489710160543,
+      "loss": 3.5384,
       "step": 24000
     },
     {
       "epoch": 2.583144979011947,
-      "eval_accuracy": 0.3662233592626207,
-      "eval_loss": 3.533743381500244,
-      "eval_runtime": 183.448,
-      "eval_samples_per_second": 98.18,
-      "eval_steps_per_second": 6.138,
+      "eval_accuracy": 0.367126264760359,
+      "eval_loss": 3.5300097465515137,
+      "eval_runtime": 180.3259,
+      "eval_samples_per_second": 99.88,
+      "eval_steps_per_second": 6.244,
       "step": 24000
     },
     {
       "epoch": 2.5885265310515555,
-      "grad_norm": 0.6276186108589172,
-      "learning_rate": 0.0004452192651653916,
-      "loss": 3.5581,
+      "grad_norm": 0.5376815795898438,
+      "learning_rate": 0.0004452257299859928,
+      "loss": 3.5519,
       "step": 24050
     },
     {
       "epoch": 2.5939080830911636,
-      "grad_norm": 0.6104943156242371,
-      "learning_rate": 0.00044489602413533025,
-      "loss": 3.5323,
+      "grad_norm": 0.5851098299026489,
+      "learning_rate": 0.0004449024889559314,
+      "loss": 3.5307,
       "step": 24100
     },
     {
       "epoch": 2.5992896351307717,
-      "grad_norm": 0.5991529822349548,
-      "learning_rate": 0.0004445727831052688,
-      "loss": 3.5428,
+      "grad_norm": 0.5848853588104248,
+      "learning_rate": 0.00044457924792587,
+      "loss": 3.5276,
       "step": 24150
     },
     {
       "epoch": 2.60467118717038,
-      "grad_norm": 0.6116634607315063,
-      "learning_rate": 0.0004442495420752074,
-      "loss": 3.5436,
+      "grad_norm": 0.5740886330604553,
+      "learning_rate": 0.0004442560068958086,
+      "loss": 3.539,
       "step": 24200
     },
     {
       "epoch": 2.610052739209988,
-      "grad_norm": 0.637483537197113,
-      "learning_rate": 0.00044392630104514593,
-      "loss": 3.5658,
+      "grad_norm": 0.5740041732788086,
+      "learning_rate": 0.0004439327658657472,
+      "loss": 3.5394,
       "step": 24250
     },
     {
       "epoch": 2.6154342912495965,
-      "grad_norm": 0.6104105710983276,
-      "learning_rate": 0.0004436030600150845,
-      "loss": 3.5557,
+      "grad_norm": 0.5800835490226746,
+      "learning_rate": 0.0004436095248356858,
+      "loss": 3.5536,
       "step": 24300
     },
     {
       "epoch": 2.6208158432892046,
-      "grad_norm": 0.603550910949707,
-      "learning_rate": 0.0004432798189850231,
-      "loss": 3.5442,
+      "grad_norm": 0.632175087928772,
+      "learning_rate": 0.00044328628380562433,
+      "loss": 3.5448,
       "step": 24350
     },
     {
       "epoch": 2.6261973953288127,
-      "grad_norm": 0.5980435013771057,
-      "learning_rate": 0.0004429565779549617,
-      "loss": 3.5587,
+      "grad_norm": 0.5998339653015137,
+      "learning_rate": 0.0004429630427755629,
+      "loss": 3.5359,
       "step": 24400
     },
     {
       "epoch": 2.6315789473684212,
-      "grad_norm": 0.556158185005188,
-      "learning_rate": 0.0004426333369249003,
-      "loss": 3.5477,
+      "grad_norm": 0.5953707695007324,
+      "learning_rate": 0.0004426398017455016,
+      "loss": 3.566,
       "step": 24450
     },
     {
       "epoch": 2.6369604994080293,
-      "grad_norm": 0.5729323029518127,
-      "learning_rate": 0.0004423100958948389,
-      "loss": 3.5398,
+      "grad_norm": 0.6147051453590393,
+      "learning_rate": 0.0004423165607154401,
+      "loss": 3.5435,
       "step": 24500
     },
     {
       "epoch": 2.6423420514476375,
-      "grad_norm": 0.7184017896652222,
-      "learning_rate": 0.00044198685486477744,
-      "loss": 3.5459,
+      "grad_norm": 0.6090589761734009,
+      "learning_rate": 0.0004419933196853787,
+      "loss": 3.5196,
       "step": 24550
     },
     {
       "epoch": 2.6477236034872456,
-      "grad_norm": 0.6241846680641174,
-      "learning_rate": 0.00044166361383471604,
-      "loss": 3.5393,
+      "grad_norm": 0.6160151362419128,
+      "learning_rate": 0.00044167007865531725,
+      "loss": 3.5354,
       "step": 24600
     },
     {
       "epoch": 2.653105155526854,
-      "grad_norm": 0.625645637512207,
-      "learning_rate": 0.0004413403728046547,
-      "loss": 3.5611,
+      "grad_norm": 0.5995622277259827,
+      "learning_rate": 0.00044134683762525584,
+      "loss": 3.5495,
       "step": 24650
     },
     {
       "epoch": 2.658486707566462,
-      "grad_norm": 0.5856419205665588,
-      "learning_rate": 0.0004410171317745932,
-      "loss": 3.5536,
+      "grad_norm": 0.6008694171905518,
+      "learning_rate": 0.00044102359659519444,
+      "loss": 3.5185,
       "step": 24700
     },
     {
       "epoch": 2.6638682596060703,
-      "grad_norm": 0.6091840863227844,
-      "learning_rate": 0.0004406938907445318,
-      "loss": 3.541,
+      "grad_norm": 0.5636075139045715,
+      "learning_rate": 0.00044070035556513303,
+      "loss": 3.5271,
       "step": 24750
     },
     {
       "epoch": 2.6692498116456784,
-      "grad_norm": 0.6365329027175903,
-      "learning_rate": 0.00044037064971447036,
-      "loss": 3.541,
+      "grad_norm": 0.5857486128807068,
+      "learning_rate": 0.00044037711453507163,
+      "loss": 3.5421,
       "step": 24800
     },
     {
       "epoch": 2.674631363685287,
-      "grad_norm": 0.5892279744148254,
-      "learning_rate": 0.00044004740868440896,
-      "loss": 3.5459,
+      "grad_norm": 0.6047202944755554,
+      "learning_rate": 0.0004400538735050102,
+      "loss": 3.5408,
       "step": 24850
     },
     {
       "epoch": 2.680012915724895,
-      "grad_norm": 0.6100271344184875,
-      "learning_rate": 0.00043972416765434755,
-      "loss": 3.5443,
+      "grad_norm": 0.5527297258377075,
+      "learning_rate": 0.00043973063247494876,
+      "loss": 3.5353,
       "step": 24900
     },
     {
       "epoch": 2.685394467764503,
-      "grad_norm": 0.5854329466819763,
-      "learning_rate": 0.00043940092662428615,
-      "loss": 3.5453,
+      "grad_norm": 0.615338146686554,
+      "learning_rate": 0.00043940739144488736,
+      "loss": 3.5521,
       "step": 24950
     },
     {
       "epoch": 2.6907760198041117,
-      "grad_norm": 0.5759541392326355,
-      "learning_rate": 0.00043907768559422474,
-      "loss": 3.5571,
+      "grad_norm": 0.5913616418838501,
+      "learning_rate": 0.0004390841504148259,
+      "loss": 3.5409,
       "step": 25000
     },
     {
       "epoch": 2.6907760198041117,
-      "eval_accuracy": 0.36772407294911424,
-      "eval_loss": 3.526393413543701,
-      "eval_runtime": 183.1728,
-      "eval_samples_per_second": 98.328,
-      "eval_steps_per_second": 6.147,
+      "eval_accuracy": 0.3674272332596051,
+      "eval_loss": 3.522557497024536,
+      "eval_runtime": 180.2113,
+      "eval_samples_per_second": 99.944,
+      "eval_steps_per_second": 6.248,
       "step": 25000
     },
     {
       "epoch": 2.69615757184372,
-      "grad_norm": 0.5549103021621704,
-      "learning_rate": 0.00043875444456416334,
-      "loss": 3.5413,
+      "grad_norm": 0.5758151412010193,
+      "learning_rate": 0.00043876090938476455,
+      "loss": 3.5389,
       "step": 25050
     },
     {
       "epoch": 2.701539123883328,
-      "grad_norm": 0.625041663646698,
-      "learning_rate": 0.0004384312035341019,
-      "loss": 3.5329,
+      "grad_norm": 0.5552731156349182,
+      "learning_rate": 0.00043843766835470314,
+      "loss": 3.5277,
       "step": 25100
     },
     {
       "epoch": 2.706920675922936,
-      "grad_norm": 0.6317231059074402,
-      "learning_rate": 0.0004381144273246417,
+      "grad_norm": 0.5564490556716919,
+      "learning_rate": 0.00043812089214524295,
       "loss": 3.5234,
       "step": 25150
     },
     {
       "epoch": 2.712302227962544,
-      "grad_norm": 0.5706122517585754,
-      "learning_rate": 0.0004377911862945803,
-      "loss": 3.5438,
+      "grad_norm": 0.5706349611282349,
+      "learning_rate": 0.00043779765111518154,
+      "loss": 3.5484,
       "step": 25200
     },
     {
       "epoch": 2.7176837800021527,
-      "grad_norm": 0.6089178919792175,
-      "learning_rate": 0.00043746794526451887,
-      "loss": 3.5184,
+      "grad_norm": 0.6321981549263,
+      "learning_rate": 0.0004374744100851201,
+      "loss": 3.5224,
       "step": 25250
     },
     {
       "epoch": 2.723065332041761,
-      "grad_norm": 0.6188173890113831,
-      "learning_rate": 0.00043714470423445747,
-      "loss": 3.5516,
+      "grad_norm": 0.6088243126869202,
+      "learning_rate": 0.0004371511690550587,
+      "loss": 3.5388,
       "step": 25300
     },
     {
       "epoch": 2.728446884081369,
-      "grad_norm": 0.5653114914894104,
-      "learning_rate": 0.00043682146320439606,
-      "loss": 3.5304,
+      "grad_norm": 0.5667572617530823,
+      "learning_rate": 0.0004368279280249972,
+      "loss": 3.5316,
       "step": 25350
     },
     {
       "epoch": 2.7338284361209775,
-      "grad_norm": 0.6082587838172913,
-      "learning_rate": 0.00043649822217433466,
-      "loss": 3.5538,
+      "grad_norm": 0.5799518823623657,
+      "learning_rate": 0.00043650468699493587,
+      "loss": 3.5414,
       "step": 25400
     },
     {
       "epoch": 2.7392099881605856,
-      "grad_norm": 0.5532857179641724,
-      "learning_rate": 0.0004361749811442732,
-      "loss": 3.5193,
+      "grad_norm": 0.5914564728736877,
+      "learning_rate": 0.00043618144596487446,
+      "loss": 3.5279,
       "step": 25450
     },
     {
       "epoch": 2.7445915402001937,
-      "grad_norm": 0.6214589476585388,
-      "learning_rate": 0.0004358517401142118,
-      "loss": 3.536,
+      "grad_norm": 0.5532760620117188,
+      "learning_rate": 0.000435858204934813,
+      "loss": 3.5156,
       "step": 25500
     },
     {
       "epoch": 2.749973092239802,
-      "grad_norm": 0.5975282192230225,
-      "learning_rate": 0.00043552849908415033,
-      "loss": 3.5301,
+      "grad_norm": 0.5992439985275269,
+      "learning_rate": 0.0004355349639047516,
+      "loss": 3.5352,
       "step": 25550
     },
     {
       "epoch": 2.7553546442794103,
-      "grad_norm": 0.6577773094177246,
-      "learning_rate": 0.000435205258054089,
-      "loss": 3.5566,
+      "grad_norm": 0.5980358123779297,
+      "learning_rate": 0.0004352117228746902,
+      "loss": 3.5518,
       "step": 25600
     },
     {
       "epoch": 2.7607361963190185,
-      "grad_norm": 0.5862059593200684,
-      "learning_rate": 0.0004348820170240276,
-      "loss": 3.5315,
+      "grad_norm": 0.6080833077430725,
+      "learning_rate": 0.00043488848184462873,
+      "loss": 3.5471,
       "step": 25650
     },
     {
       "epoch": 2.7661177483586266,
-      "grad_norm": 0.6143049001693726,
-      "learning_rate": 0.0004345587759939661,
-      "loss": 3.5541,
+      "grad_norm": 0.6731295585632324,
+      "learning_rate": 0.0004345652408145674,
+      "loss": 3.548,
       "step": 25700
     },
     {
       "epoch": 2.7714993003982347,
-      "grad_norm": 0.5826513767242432,
-      "learning_rate": 0.0004342355349639047,
-      "loss": 3.5328,
+      "grad_norm": 0.576494038105011,
+      "learning_rate": 0.000434241999784506,
+      "loss": 3.5336,
       "step": 25750
     },
     {
       "epoch": 2.776880852437843,
-      "grad_norm": 0.6236239075660706,
-      "learning_rate": 0.0004339122939338433,
-      "loss": 3.5282,
+      "grad_norm": 0.562201738357544,
+      "learning_rate": 0.0004339187587544445,
+      "loss": 3.5494,
       "step": 25800
     },
     {
       "epoch": 2.7822624044774513,
-      "grad_norm": 0.5690028667449951,
-      "learning_rate": 0.00043358905290378184,
-      "loss": 3.5382,
+      "grad_norm": 0.5914180874824524,
+      "learning_rate": 0.0004335955177243831,
+      "loss": 3.527,
       "step": 25850
     },
     {
       "epoch": 2.7876439565170594,
-      "grad_norm": 0.630249559879303,
-      "learning_rate": 0.0004332658118737205,
-      "loss": 3.5471,
+      "grad_norm": 0.5731594562530518,
+      "learning_rate": 0.00043327227669432165,
+      "loss": 3.5329,
       "step": 25900
     },
     {
       "epoch": 2.793025508556668,
-      "grad_norm": 0.561437726020813,
-      "learning_rate": 0.0004329425708436591,
-      "loss": 3.5452,
+      "grad_norm": 0.5664558410644531,
+      "learning_rate": 0.00043294903566426025,
+      "loss": 3.5355,
       "step": 25950
     },
     {
       "epoch": 2.798407060596276,
-      "grad_norm": 0.5968010425567627,
-      "learning_rate": 0.00043261932981359763,
-      "loss": 3.5268,
+      "grad_norm": 0.5670287609100342,
+      "learning_rate": 0.0004326257946341989,
+      "loss": 3.5356,
       "step": 26000
     },
     {
       "epoch": 2.798407060596276,
-      "eval_accuracy": 0.3683658855574344,
-      "eval_loss": 3.5156972408294678,
-      "eval_runtime": 183.4726,
-      "eval_samples_per_second": 98.167,
-      "eval_steps_per_second": 6.137,
+      "eval_accuracy": 0.36834143865767616,
+      "eval_loss": 3.510011672973633,
+      "eval_runtime": 180.2381,
+      "eval_samples_per_second": 99.929,
+      "eval_steps_per_second": 6.247,
       "step": 26000
     },
     {
       "epoch": 2.803788612635884,
-      "grad_norm": 0.5962703227996826,
-      "learning_rate": 0.0004322960887835362,
-      "loss": 3.5435,
+      "grad_norm": 0.5943952202796936,
+      "learning_rate": 0.00043230255360413744,
+      "loss": 3.5392,
       "step": 26050
     },
     {
       "epoch": 2.8091701646754923,
-      "grad_norm": 0.5704419016838074,
-      "learning_rate": 0.00043197284775347476,
-      "loss": 3.5439,
+      "grad_norm": 0.5770288109779358,
+      "learning_rate": 0.00043197931257407603,
+      "loss": 3.5359,
       "step": 26100
     },
     {
       "epoch": 2.8145517167151004,
-      "grad_norm": 0.6311613917350769,
-      "learning_rate": 0.00043164960672341336,
-      "loss": 3.5348,
+      "grad_norm": 0.5877098441123962,
+      "learning_rate": 0.0004316560715440146,
+      "loss": 3.5339,
       "step": 26150
     },
     {
       "epoch": 2.819933268754709,
-      "grad_norm": 0.5643249154090881,
+      "grad_norm": 0.6326656937599182,
       "learning_rate": 0.00043133283051395317,
-      "loss": 3.5529,
+      "loss": 3.5566,
       "step": 26200
     },
     {
       "epoch": 2.825314820794317,
-      "grad_norm": 0.6420338749885559,
+      "grad_norm": 0.5939858555793762,
       "learning_rate": 0.0004310095894838918,
-      "loss": 3.545,
+      "loss": 3.5401,
       "step": 26250
     },
     {
       "epoch": 2.830696372833925,
-      "grad_norm": 0.6328233480453491,
+      "grad_norm": 0.538489818572998,
       "learning_rate": 0.0004306863484538304,
-      "loss": 3.5405,
+      "loss": 3.5488,
       "step": 26300
     },
     {
       "epoch": 2.8360779248735337,
-      "grad_norm": 0.6627986431121826,
+      "grad_norm": 0.5739915370941162,
       "learning_rate": 0.00043036310742376895,
-      "loss": 3.5228,
+      "loss": 3.5204,
       "step": 26350
     },
     {
       "epoch": 2.841459476913142,
-      "grad_norm": 0.5869179964065552,
+      "grad_norm": 0.5786092877388,
       "learning_rate": 0.00043003986639370754,
-      "loss": 3.5245,
+      "loss": 3.5213,
       "step": 26400
     },
     {
       "epoch": 2.84684102895275,
-      "grad_norm": 0.5820196270942688,
+      "grad_norm": 0.5726288557052612,
       "learning_rate": 0.0004297166253636461,
-      "loss": 3.5315,
+      "loss": 3.5369,
       "step": 26450
     },
     {
       "epoch": 2.852222580992358,
-      "grad_norm": 0.6153207421302795,
+      "grad_norm": 0.578011691570282,
       "learning_rate": 0.0004293933843335847,
-      "loss": 3.5206,
+      "loss": 3.5242,
       "step": 26500
     },
     {
       "epoch": 2.857604133031966,
-      "grad_norm": 0.5706645846366882,
+      "grad_norm": 0.6015625,
       "learning_rate": 0.00042907014330352333,
-      "loss": 3.5492,
+      "loss": 3.5361,
       "step": 26550
     },
     {
       "epoch": 2.8629856850715747,
-      "grad_norm": 0.5671915411949158,
+      "grad_norm": 0.5923911929130554,
       "learning_rate": 0.00042874690227346187,
-      "loss": 3.5261,
+      "loss": 3.5423,
       "step": 26600
     },
     {
       "epoch": 2.868367237111183,
-      "grad_norm": 0.6378110647201538,
+      "grad_norm": 0.5731902718544006,
       "learning_rate": 0.00042842366124340046,
-      "loss": 3.5554,
+      "loss": 3.538,
       "step": 26650
     },
     {
       "epoch": 2.873748789150791,
-      "grad_norm": 0.5837578773498535,
+      "grad_norm": 0.6166850924491882,
       "learning_rate": 0.00042810042021333906,
-      "loss": 3.5286,
+      "loss": 3.5398,
       "step": 26700
     },
     {
       "epoch": 2.8791303411903995,
-      "grad_norm": 0.5720592737197876,
+      "grad_norm": 0.6081316471099854,
       "learning_rate": 0.0004277771791832776,
-      "loss": 3.5193,
+      "loss": 3.5272,
       "step": 26750
     },
     {
       "epoch": 2.8845118932300076,
-      "grad_norm": 0.6068984270095825,
+      "grad_norm": 0.5820943713188171,
       "learning_rate": 0.0004274539381532162,
-      "loss": 3.5375,
+      "loss": 3.5307,
       "step": 26800
     },
     {
       "epoch": 2.8898934452696157,
-      "grad_norm": 0.590977132320404,
+      "grad_norm": 0.5724826455116272,
       "learning_rate": 0.00042713069712315484,
-      "loss": 3.515,
+      "loss": 3.5088,
       "step": 26850
     },
     {
       "epoch": 2.895274997309224,
-      "grad_norm": 0.6040093302726746,
+      "grad_norm": 0.6168664693832397,
       "learning_rate": 0.0004268074560930934,
-      "loss": 3.5472,
+      "loss": 3.5398,
       "step": 26900
     },
     {
       "epoch": 2.9006565493488323,
-      "grad_norm": 0.5838053822517395,
+      "grad_norm": 0.5778253078460693,
       "learning_rate": 0.000426484215063032,
-      "loss": 3.5313,
+      "loss": 3.516,
       "step": 26950
     },
     {
       "epoch": 2.9060381013884404,
-      "grad_norm": 0.6135042905807495,
+      "grad_norm": 0.6123675107955933,
       "learning_rate": 0.0004261609740329705,
-      "loss": 3.5395,
+      "loss": 3.518,
       "step": 27000
     },
     {
       "epoch": 2.9060381013884404,
-      "eval_accuracy": 0.3692769400217588,
-      "eval_loss": 3.50553035736084,
-      "eval_runtime": 183.0394,
-      "eval_samples_per_second": 98.4,
-      "eval_steps_per_second": 6.152,
+      "eval_accuracy": 0.3695355338947573,
+      "eval_loss": 3.5021097660064697,
+      "eval_runtime": 180.3865,
+      "eval_samples_per_second": 99.847,
+      "eval_steps_per_second": 6.242,
       "step": 27000
     },
     {
       "epoch": 2.9114196534280485,
-      "grad_norm": 0.6675634384155273,
+      "grad_norm": 0.5784735083580017,
       "learning_rate": 0.0004258377330029091,
-      "loss": 3.5392,
+      "loss": 3.5326,
       "step": 27050
     },
     {
       "epoch": 2.9168012054676566,
-      "grad_norm": 0.5531997680664062,
+      "grad_norm": 0.5168740153312683,
       "learning_rate": 0.00042551449197284776,
-      "loss": 3.5078,
+      "loss": 3.524,
       "step": 27100
     },
     {
       "epoch": 2.922182757507265,
-      "grad_norm": 0.6126731634140015,
+      "grad_norm": 0.6014238595962524,
       "learning_rate": 0.0004251912509427863,
-      "loss": 3.5329,
+      "loss": 3.5285,
       "step": 27150
     },
     {
       "epoch": 2.9275643095468733,
-      "grad_norm": 0.5538141131401062,
+      "grad_norm": 0.5599318742752075,
       "learning_rate": 0.0004248680099127249,
-      "loss": 3.5216,
+      "loss": 3.5311,
       "step": 27200
     },
     {
       "epoch": 2.9329458615864814,
-      "grad_norm": 0.5796554088592529,
+      "grad_norm": 0.5942395925521851,
       "learning_rate": 0.0004245447688826635,
-      "loss": 3.5247,
+      "loss": 3.5191,
       "step": 27250
     },
     {
       "epoch": 2.93832741362609,
-      "grad_norm": 0.5609702467918396,
+      "grad_norm": 0.581760048866272,
       "learning_rate": 0.00042422152785260203,
-      "loss": 3.5229,
+      "loss": 3.5228,
       "step": 27300
     },
     {
       "epoch": 2.943708965665698,
-      "grad_norm": 0.6337541937828064,
+      "grad_norm": 0.5903612375259399,
       "learning_rate": 0.0004238982868225406,
-      "loss": 3.4924,
+      "loss": 3.5014,
       "step": 27350
     },
     {
       "epoch": 2.949090517705306,
-      "grad_norm": 0.6238342523574829,
+      "grad_norm": 0.5760148763656616,
       "learning_rate": 0.0004235750457924793,
-      "loss": 3.5259,
+      "loss": 3.5358,
       "step": 27400
     },
     {
       "epoch": 2.9544720697449143,
-      "grad_norm": 0.707619309425354,
+      "grad_norm": 0.5803918242454529,
       "learning_rate": 0.0004232518047624178,
-      "loss": 3.5301,
+      "loss": 3.5335,
       "step": 27450
     },
     {
       "epoch": 2.9598536217845224,
-      "grad_norm": 0.5899302959442139,
+      "grad_norm": 0.5845760703086853,
       "learning_rate": 0.0004229285637323564,
-      "loss": 3.5246,
+      "loss": 3.5053,
       "step": 27500
     },
     {
       "epoch": 2.965235173824131,
-      "grad_norm": 0.552964985370636,
+      "grad_norm": 0.5800291895866394,
       "learning_rate": 0.00042260532270229495,
-      "loss": 3.5266,
+      "loss": 3.5168,
       "step": 27550
     },
     {
       "epoch": 2.970616725863739,
-      "grad_norm": 0.5927809476852417,
+      "grad_norm": 0.6354805827140808,
       "learning_rate": 0.00042228208167223354,
-      "loss": 3.5179,
+      "loss": 3.5093,
       "step": 27600
     },
     {
       "epoch": 2.975998277903347,
-      "grad_norm": 0.6026394367218018,
+      "grad_norm": 0.6351531744003296,
       "learning_rate": 0.00042195884064217214,
-      "loss": 3.531,
+      "loss": 3.5177,
       "step": 27650
     },
     {
       "epoch": 2.9813798299429557,
-      "grad_norm": 0.6079704165458679,
+      "grad_norm": 0.6364790201187134,
       "learning_rate": 0.00042163559961211073,
-      "loss": 3.5447,
+      "loss": 3.527,
       "step": 27700
     },
     {
       "epoch": 2.986761381982564,
-      "grad_norm": 0.5754170417785645,
+      "grad_norm": 0.5370408892631531,
       "learning_rate": 0.00042131235858204933,
-      "loss": 3.5106,
+      "loss": 3.5141,
       "step": 27750
     },
     {
       "epoch": 2.992142934022172,
-      "grad_norm": 0.5957307815551758,
+      "grad_norm": 0.6115018725395203,
       "learning_rate": 0.0004209891175519879,
-      "loss": 3.5292,
+      "loss": 3.531,
       "step": 27800
     },
     {
       "epoch": 2.9975244860617805,
-      "grad_norm": 0.5584477782249451,
+      "grad_norm": 0.5605997443199158,
       "learning_rate": 0.00042066587652192646,
-      "loss": 3.517,
+      "loss": 3.5233,
       "step": 27850
     },
     {
       "epoch": 3.0029060381013886,
-      "grad_norm": 0.5913463234901428,
+      "grad_norm": 0.5835641026496887,
       "learning_rate": 0.00042034263549186506,
-      "loss": 3.4829,
+      "loss": 3.4592,
       "step": 27900
     },
     {
       "epoch": 3.0082875901409967,
-      "grad_norm": 0.5728430151939392,
+      "grad_norm": 0.5684875249862671,
       "learning_rate": 0.0004200193944618036,
-      "loss": 3.4424,
+      "loss": 3.4344,
       "step": 27950
     },
     {
       "epoch": 3.0136691421806048,
-      "grad_norm": 0.6232100129127502,
+      "grad_norm": 0.6038845181465149,
       "learning_rate": 0.00041969615343174225,
-      "loss": 3.4403,
+      "loss": 3.4181,
       "step": 28000
     },
     {
       "epoch": 3.0136691421806048,
-      "eval_accuracy": 0.3704998282741108,
-      "eval_loss": 3.501554250717163,
-      "eval_runtime": 181.6654,
-      "eval_samples_per_second": 99.144,
-      "eval_steps_per_second": 6.198,
+      "eval_accuracy": 0.3707335406357998,
+      "eval_loss": 3.499069929122925,
+      "eval_runtime": 181.6538,
+      "eval_samples_per_second": 99.15,
+      "eval_steps_per_second": 6.199,
       "step": 28000
     },
     {
       "epoch": 3.0190506942202133,
-      "grad_norm": 0.5570206046104431,
+      "grad_norm": 0.6008001565933228,
       "learning_rate": 0.00041937291240168084,
-      "loss": 3.436,
+      "loss": 3.4198,
       "step": 28050
     },
     {
       "epoch": 3.0244322462598214,
-      "grad_norm": 0.6119909882545471,
+      "grad_norm": 0.603621244430542,
       "learning_rate": 0.0004190496713716194,
-      "loss": 3.4374,
+      "loss": 3.4395,
       "step": 28100
     },
     {
       "epoch": 3.0298137982994295,
-      "grad_norm": 0.6302096843719482,
-      "learning_rate": 0.000418726430341558,
-      "loss": 3.4448,
+      "grad_norm": 0.558818519115448,
+      "learning_rate": 0.00041873289516215924,
+      "loss": 3.4516,
       "step": 28150
     },
     {
       "epoch": 3.0351953503390376,
-      "grad_norm": 0.5777623653411865,
-      "learning_rate": 0.00041840318931149657,
-      "loss": 3.4356,
+      "grad_norm": 0.632050633430481,
+      "learning_rate": 0.0004184096541320978,
+      "loss": 3.4348,
       "step": 28200
     },
     {
       "epoch": 3.040576902378646,
-      "grad_norm": 0.5832511782646179,
-      "learning_rate": 0.00041807994828143517,
-      "loss": 3.4348,
+      "grad_norm": 0.5583595633506775,
+      "learning_rate": 0.0004180864131020364,
+      "loss": 3.4277,
       "step": 28250
     },
     {
       "epoch": 3.0459584544182543,
-      "grad_norm": 0.6186639666557312,
-      "learning_rate": 0.00041775670725137376,
-      "loss": 3.4531,
+      "grad_norm": 0.5803127884864807,
+      "learning_rate": 0.0004177631720719749,
+      "loss": 3.4405,
       "step": 28300
     },
     {
       "epoch": 3.0513400064578624,
-      "grad_norm": 0.5793587565422058,
-      "learning_rate": 0.00041743346622131236,
-      "loss": 3.4351,
+      "grad_norm": 0.6481485962867737,
+      "learning_rate": 0.00041743993104191357,
+      "loss": 3.4311,
       "step": 28350
     },
     {
       "epoch": 3.0567215584974705,
-      "grad_norm": 0.6191322207450867,
-      "learning_rate": 0.0004171102251912509,
-      "loss": 3.4609,
+      "grad_norm": 0.6118427515029907,
+      "learning_rate": 0.00041711669001185216,
+      "loss": 3.4403,
       "step": 28400
     },
     {
       "epoch": 3.062103110537079,
-      "grad_norm": 0.5854801535606384,
-      "learning_rate": 0.0004167869841611895,
-      "loss": 3.445,
+      "grad_norm": 0.6139327883720398,
+      "learning_rate": 0.0004167934489817907,
+      "loss": 3.4492,
       "step": 28450
     },
     {
       "epoch": 3.067484662576687,
-      "grad_norm": 0.5954617857933044,
-      "learning_rate": 0.00041646374313112803,
-      "loss": 3.4516,
+      "grad_norm": 0.6095017790794373,
+      "learning_rate": 0.0004164702079517293,
+      "loss": 3.4327,
       "step": 28500
     },
     {
       "epoch": 3.0728662146162953,
-      "grad_norm": 0.6144933104515076,
-      "learning_rate": 0.0004161405021010667,
-      "loss": 3.4306,
+      "grad_norm": 0.5820819139480591,
+      "learning_rate": 0.0004161469669216679,
+      "loss": 3.4376,
       "step": 28550
     },
     {
       "epoch": 3.0782477666559034,
-      "grad_norm": 0.6332607269287109,
-      "learning_rate": 0.0004158172610710053,
-      "loss": 3.4483,
+      "grad_norm": 0.6147562265396118,
+      "learning_rate": 0.00041582372589160643,
+      "loss": 3.4345,
       "step": 28600
     },
     {
       "epoch": 3.083629318695512,
-      "grad_norm": 0.6351560354232788,
-      "learning_rate": 0.0004154940200409438,
-      "loss": 3.4597,
+      "grad_norm": 0.6053479313850403,
+      "learning_rate": 0.0004155004848615451,
+      "loss": 3.4541,
       "step": 28650
     },
     {
       "epoch": 3.08901087073512,
-      "grad_norm": 0.580599308013916,
-      "learning_rate": 0.0004151707790108824,
-      "loss": 3.4463,
+      "grad_norm": 0.6054416298866272,
+      "learning_rate": 0.0004151772438314837,
+      "loss": 3.4393,
       "step": 28700
     },
     {
       "epoch": 3.094392422774728,
-      "grad_norm": 0.636760950088501,
-      "learning_rate": 0.00041484753798082095,
-      "loss": 3.4619,
+      "grad_norm": 0.600176990032196,
+      "learning_rate": 0.0004148540028014222,
+      "loss": 3.4524,
       "step": 28750
     },
     {
       "epoch": 3.0997739748143363,
-      "grad_norm": 0.6538554430007935,
-      "learning_rate": 0.00041452429695075955,
-      "loss": 3.448,
+      "grad_norm": 0.5862427949905396,
+      "learning_rate": 0.0004145307617713608,
+      "loss": 3.4449,
       "step": 28800
     },
     {
       "epoch": 3.105155526853945,
-      "grad_norm": 0.5733194351196289,
-      "learning_rate": 0.0004142010559206982,
-      "loss": 3.4343,
+      "grad_norm": 0.5885952711105347,
+      "learning_rate": 0.00041420752074129935,
+      "loss": 3.4442,
       "step": 28850
     },
     {
       "epoch": 3.110537078893553,
-      "grad_norm": 0.6369328498840332,
-      "learning_rate": 0.00041387781489063673,
-      "loss": 3.4369,
+      "grad_norm": 0.6215264797210693,
+      "learning_rate": 0.000413884279711238,
+      "loss": 3.4375,
       "step": 28900
     },
     {
       "epoch": 3.115918630933161,
-      "grad_norm": 0.6201010942459106,
-      "learning_rate": 0.00041355457386057533,
-      "loss": 3.453,
+      "grad_norm": 0.5917741060256958,
+      "learning_rate": 0.0004135610386811766,
+      "loss": 3.4585,
       "step": 28950
     },
     {
       "epoch": 3.121300182972769,
-      "grad_norm": 0.6407283544540405,
-      "learning_rate": 0.0004132313328305139,
-      "loss": 3.4636,
+      "grad_norm": 0.6078323721885681,
+      "learning_rate": 0.00041323779765111514,
+      "loss": 3.4367,
       "step": 29000
     },
     {
       "epoch": 3.121300182972769,
-      "eval_accuracy": 0.3711315361638642,
-      "eval_loss": 3.497187614440918,
-      "eval_runtime": 181.453,
-      "eval_samples_per_second": 99.26,
-      "eval_steps_per_second": 6.205,
+      "eval_accuracy": 0.37107069054668806,
+      "eval_loss": 3.4967191219329834,
+      "eval_runtime": 181.4214,
+      "eval_samples_per_second": 99.277,
+      "eval_steps_per_second": 6.207,
       "step": 29000
     },
     {
       "epoch": 3.1266817350123777,
-      "grad_norm": 0.6396225690841675,
-      "learning_rate": 0.00041290809180045246,
-      "loss": 3.4613,
+      "grad_norm": 0.5602723360061646,
+      "learning_rate": 0.00041291455662105373,
+      "loss": 3.4451,
       "step": 29050
     },
     {
       "epoch": 3.132063287051986,
-      "grad_norm": 0.6482574939727783,
-      "learning_rate": 0.0004125848507703911,
-      "loss": 3.4346,
+      "grad_norm": 0.6069493293762207,
+      "learning_rate": 0.0004125913155909923,
+      "loss": 3.4455,
       "step": 29100
     },
     {
       "epoch": 3.137444839091594,
-      "grad_norm": 0.6442139744758606,
-      "learning_rate": 0.0004122616097403297,
-      "loss": 3.4496,
+      "grad_norm": 0.6118255853652954,
+      "learning_rate": 0.00041226807456093087,
+      "loss": 3.4578,
       "step": 29150
     },
     {
       "epoch": 3.1428263911312024,
-      "grad_norm": 0.5868032574653625,
+      "grad_norm": 0.5816478133201599,
       "learning_rate": 0.0004119448335308695,
-      "loss": 3.4653,
+      "loss": 3.4502,
       "step": 29200
     },
     {
       "epoch": 3.1482079431708105,
-      "grad_norm": 0.8085774779319763,
+      "grad_norm": 0.6039485931396484,
       "learning_rate": 0.0004116215925008081,
-      "loss": 3.4591,
+      "loss": 3.4521,
       "step": 29250
     },
     {
       "epoch": 3.1535894952104186,
-      "grad_norm": 0.5951862931251526,
+      "grad_norm": 0.5718762874603271,
       "learning_rate": 0.00041129835147074665,
-      "loss": 3.4455,
+      "loss": 3.4681,
       "step": 29300
     },
     {
       "epoch": 3.1589710472500268,
-      "grad_norm": 0.6378307938575745,
+      "grad_norm": 0.6043664813041687,
       "learning_rate": 0.00041097511044068524,
-      "loss": 3.4538,
+      "loss": 3.4442,
       "step": 29350
     },
     {
       "epoch": 3.1643525992896353,
-      "grad_norm": 0.6090945601463318,
+      "grad_norm": 0.6371094584465027,
       "learning_rate": 0.0004106518694106238,
-      "loss": 3.448,
+      "loss": 3.4496,
       "step": 29400
     },
     {
       "epoch": 3.1697341513292434,
-      "grad_norm": 0.6292023062705994,
+      "grad_norm": 0.5829236507415771,
       "learning_rate": 0.0004103286283805624,
-      "loss": 3.4687,
+      "loss": 3.4686,
       "step": 29450
     },
     {
       "epoch": 3.1751157033688515,
-      "grad_norm": 0.6239890456199646,
+      "grad_norm": 0.5857312083244324,
       "learning_rate": 0.00041000538735050103,
-      "loss": 3.4909,
+      "loss": 3.4659,
       "step": 29500
     },
     {
       "epoch": 3.1804972554084596,
-      "grad_norm": 0.5989189147949219,
+      "grad_norm": 0.6097814440727234,
       "learning_rate": 0.00040968214632043957,
-      "loss": 3.4548,
+      "loss": 3.4451,
       "step": 29550
     },
     {
       "epoch": 3.185878807448068,
-      "grad_norm": 0.6178577542304993,
+      "grad_norm": 0.5862761735916138,
       "learning_rate": 0.00040935890529037816,
-      "loss": 3.4575,
+      "loss": 3.4522,
       "step": 29600
     },
     {
       "epoch": 3.1912603594876763,
-      "grad_norm": 0.6444182395935059,
+      "grad_norm": 0.592506468296051,
       "learning_rate": 0.00040903566426031676,
-      "loss": 3.4525,
+      "loss": 3.4427,
       "step": 29650
     },
     {
       "epoch": 3.1966419115272844,
-      "grad_norm": 0.6180800199508667,
+      "grad_norm": 0.6013423204421997,
       "learning_rate": 0.0004087124232302553,
-      "loss": 3.436,
+      "loss": 3.4579,
       "step": 29700
     },
     {
       "epoch": 3.2020234635668925,
-      "grad_norm": 0.595629096031189,
+      "grad_norm": 0.624481201171875,
       "learning_rate": 0.0004083891822001939,
-      "loss": 3.4835,
+      "loss": 3.4725,
       "step": 29750
     },
     {
       "epoch": 3.207405015606501,
-      "grad_norm": 0.6241080164909363,
+      "grad_norm": 0.5892785787582397,
       "learning_rate": 0.00040806594117013254,
-      "loss": 3.454,
+      "loss": 3.4465,
       "step": 29800
     },
     {
       "epoch": 3.212786567646109,
-      "grad_norm": 0.6285580992698669,
+      "grad_norm": 0.638572096824646,
       "learning_rate": 0.0004077427001400711,
-      "loss": 3.4568,
+      "loss": 3.4549,
       "step": 29850
     },
     {
       "epoch": 3.2181681196857173,
-      "grad_norm": 0.648627758026123,
+      "grad_norm": 0.6022881865501404,
       "learning_rate": 0.0004074194591100097,
-      "loss": 3.441,
+      "loss": 3.442,
       "step": 29900
     },
     {
       "epoch": 3.2235496717253254,
-      "grad_norm": 0.64311283826828,
+      "grad_norm": 0.6144007444381714,
       "learning_rate": 0.0004070962180799482,
-      "loss": 3.4699,
+      "loss": 3.4683,
       "step": 29950
     },
     {
       "epoch": 3.228931223764934,
-      "grad_norm": 0.5918191075325012,
+      "grad_norm": 0.5872783064842224,
       "learning_rate": 0.0004067729770498868,
-      "loss": 3.4348,
+      "loss": 3.4534,
       "step": 30000
     },
     {
       "epoch": 3.228931223764934,
-      "eval_accuracy": 0.37159700513526145,
-      "eval_loss": 3.49194073677063,
-      "eval_runtime": 181.5347,
-      "eval_samples_per_second": 99.215,
-      "eval_steps_per_second": 6.203,
+      "eval_accuracy": 0.37168349283396174,
+      "eval_loss": 3.4922738075256348,
+      "eval_runtime": 181.9551,
+      "eval_samples_per_second": 98.986,
+      "eval_steps_per_second": 6.188,
       "step": 30000
     },
     {
       "epoch": 3.234312775804542,
-      "grad_norm": 0.6755573153495789,
+      "grad_norm": 0.6475398540496826,
       "learning_rate": 0.00040644973601982546,
-      "loss": 3.4519,
+      "loss": 3.4553,
       "step": 30050
     },
     {
       "epoch": 3.23969432784415,
-      "grad_norm": 0.6835849285125732,
+      "grad_norm": 0.6069983839988708,
       "learning_rate": 0.000406126494989764,
-      "loss": 3.4827,
+      "loss": 3.4623,
       "step": 30100
     },
     {
       "epoch": 3.2450758798837587,
-      "grad_norm": 0.6226943135261536,
+      "grad_norm": 0.6207373142242432,
       "learning_rate": 0.0004058032539597026,
-      "loss": 3.4645,
+      "loss": 3.4686,
       "step": 30150
     },
     {
       "epoch": 3.250457431923367,
-      "grad_norm": 0.6609615087509155,
+      "grad_norm": 0.628922700881958,
       "learning_rate": 0.00040548001292964114,
-      "loss": 3.4642,
+      "loss": 3.4527,
       "step": 30200
     },
     {
       "epoch": 3.255838983962975,
-      "grad_norm": 0.5687957406044006,
+      "grad_norm": 0.5696398019790649,
       "learning_rate": 0.00040515677189957973,
-      "loss": 3.4688,
+      "loss": 3.4717,
       "step": 30250
     },
     {
       "epoch": 3.261220536002583,
-      "grad_norm": 0.6109760999679565,
+      "grad_norm": 0.6937188506126404,
       "learning_rate": 0.0004048335308695183,
-      "loss": 3.4437,
+      "loss": 3.445,
       "step": 30300
     },
     {
       "epoch": 3.2666020880421915,
-      "grad_norm": 0.5959461331367493,
+      "grad_norm": 0.6091139912605286,
       "learning_rate": 0.0004045102898394569,
-      "loss": 3.4724,
+      "loss": 3.4661,
       "step": 30350
     },
     {
       "epoch": 3.2719836400817996,
-      "grad_norm": 0.6271927952766418,
+      "grad_norm": 0.6347094774246216,
       "learning_rate": 0.0004041870488093955,
-      "loss": 3.4651,
+      "loss": 3.4472,
       "step": 30400
     },
     {
       "epoch": 3.2773651921214078,
-      "grad_norm": 0.6263259649276733,
+      "grad_norm": 0.5904783010482788,
       "learning_rate": 0.0004038638077793341,
-      "loss": 3.4471,
+      "loss": 3.4709,
       "step": 30450
     },
     {
       "epoch": 3.282746744161016,
-      "grad_norm": 0.6434895396232605,
+      "grad_norm": 0.6150405406951904,
       "learning_rate": 0.00040354056674927265,
-      "loss": 3.4604,
+      "loss": 3.4546,
       "step": 30500
     },
     {
       "epoch": 3.2881282962006244,
-      "grad_norm": 0.661169707775116,
+      "grad_norm": 0.6563023328781128,
       "learning_rate": 0.00040321732571921124,
-      "loss": 3.4547,
+      "loss": 3.455,
       "step": 30550
     },
     {
       "epoch": 3.2935098482402325,
-      "grad_norm": 0.683679461479187,
+      "grad_norm": 0.6171647310256958,
       "learning_rate": 0.0004028940846891498,
-      "loss": 3.4602,
+      "loss": 3.4544,
       "step": 30600
     },
     {
       "epoch": 3.2988914002798406,
-      "grad_norm": 0.6047713756561279,
+      "grad_norm": 0.6179287433624268,
       "learning_rate": 0.00040257084365908843,
-      "loss": 3.4512,
+      "loss": 3.446,
       "step": 30650
     },
     {
       "epoch": 3.304272952319449,
-      "grad_norm": 0.6232369542121887,
+      "grad_norm": 0.6334787011146545,
       "learning_rate": 0.00040224760262902703,
-      "loss": 3.4713,
+      "loss": 3.4667,
       "step": 30700
     },
     {
       "epoch": 3.3096545043590573,
-      "grad_norm": 0.6141099333763123,
+      "grad_norm": 0.6175564527511597,
       "learning_rate": 0.00040192436159896557,
-      "loss": 3.4627,
+      "loss": 3.469,
       "step": 30750
     },
     {
       "epoch": 3.3150360563986654,
-      "grad_norm": 0.6429236531257629,
+      "grad_norm": 0.6480590105056763,
       "learning_rate": 0.00040160112056890416,
-      "loss": 3.4586,
+      "loss": 3.4429,
       "step": 30800
     },
     {
       "epoch": 3.3204176084382735,
-      "grad_norm": 0.5798470377922058,
+      "grad_norm": 0.6085280179977417,
       "learning_rate": 0.00040127787953884276,
-      "loss": 3.4562,
+      "loss": 3.4571,
       "step": 30850
     },
     {
       "epoch": 3.3257991604778816,
-      "grad_norm": 0.6334143280982971,
+      "grad_norm": 0.6256332993507385,
       "learning_rate": 0.00040095463850878135,
-      "loss": 3.4408,
+      "loss": 3.4434,
       "step": 30900
     },
     {
       "epoch": 3.33118071251749,
-      "grad_norm": 0.6591746211051941,
+      "grad_norm": 0.5866464376449585,
       "learning_rate": 0.00040063139747871995,
-      "loss": 3.4526,
+      "loss": 3.4603,
       "step": 30950
     },
     {
       "epoch": 3.3365622645570983,
-      "grad_norm": 0.6119304895401001,
+      "grad_norm": 0.6502524018287659,
       "learning_rate": 0.00040030815644865854,
-      "loss": 3.4676,
+      "loss": 3.4682,
       "step": 31000
     },
     {
       "epoch": 3.3365622645570983,
-      "eval_accuracy": 0.3719315473768422,
-      "eval_loss": 3.486844062805176,
-      "eval_runtime": 182.7259,
-      "eval_samples_per_second": 98.568,
-      "eval_steps_per_second": 6.162,
+      "eval_accuracy": 0.3722422946359916,
+      "eval_loss": 3.4806246757507324,
+      "eval_runtime": 180.1007,
+      "eval_samples_per_second": 100.005,
+      "eval_steps_per_second": 6.252,
       "step": 31000
     },
     {
       "epoch": 3.3419438165967064,
-      "grad_norm": 0.6275706887245178,
+      "grad_norm": 0.6035051345825195,
       "learning_rate": 0.0003999849154185971,
-      "loss": 3.4726,
+      "loss": 3.4659,
       "step": 31050
     },
     {
       "epoch": 3.347325368636315,
-      "grad_norm": 0.5887126326560974,
+      "grad_norm": 0.6270487308502197,
       "learning_rate": 0.0003996616743885357,
-      "loss": 3.4399,
+      "loss": 3.4582,
       "step": 31100
     },
     {
       "epoch": 3.352706920675923,
-      "grad_norm": 0.6963056921958923,
+      "grad_norm": 0.6606770157814026,
       "learning_rate": 0.0003993384333584742,
-      "loss": 3.4559,
+      "loss": 3.4488,
       "step": 31150
     },
     {
       "epoch": 3.358088472715531,
-      "grad_norm": 0.642962634563446,
+      "grad_norm": 0.621417224407196,
       "learning_rate": 0.00039901519232841287,
-      "loss": 3.4906,
+      "loss": 3.4636,
       "step": 31200
     },
     {
       "epoch": 3.3634700247551392,
-      "grad_norm": 0.6062067747116089,
+      "grad_norm": 0.6087856292724609,
       "learning_rate": 0.00039869195129835146,
-      "loss": 3.4634,
+      "loss": 3.4655,
       "step": 31250
     },
     {
       "epoch": 3.368851576794748,
-      "grad_norm": 0.6534258723258972,
+      "grad_norm": 0.6589096188545227,
       "learning_rate": 0.00039836871026829,
-      "loss": 3.4594,
+      "loss": 3.4636,
       "step": 31300
     },
     {
       "epoch": 3.374233128834356,
-      "grad_norm": 0.60047847032547,
+      "grad_norm": 0.6454888582229614,
       "learning_rate": 0.0003980454692382286,
-      "loss": 3.4708,
+      "loss": 3.467,
       "step": 31350
     },
     {
       "epoch": 3.379614680873964,
-      "grad_norm": 0.5911248326301575,
+      "grad_norm": 0.5854974985122681,
       "learning_rate": 0.0003977222282081672,
-      "loss": 3.4498,
+      "loss": 3.4461,
       "step": 31400
     },
     {
       "epoch": 3.384996232913572,
-      "grad_norm": 0.5842762589454651,
+      "grad_norm": 0.6991376876831055,
       "learning_rate": 0.00039739898717810573,
-      "loss": 3.472,
+      "loss": 3.4659,
       "step": 31450
     },
     {
       "epoch": 3.3903777849531807,
-      "grad_norm": 0.6286013126373291,
+      "grad_norm": 0.6003750562667847,
       "learning_rate": 0.0003970757461480444,
-      "loss": 3.4783,
+      "loss": 3.4762,
       "step": 31500
     },
     {
       "epoch": 3.3957593369927888,
-      "grad_norm": 0.6078022718429565,
+      "grad_norm": 0.5914140939712524,
       "learning_rate": 0.000396752505117983,
-      "loss": 3.4681,
+      "loss": 3.4591,
       "step": 31550
     },
     {
       "epoch": 3.401140889032397,
-      "grad_norm": 0.6259421706199646,
+      "grad_norm": 0.6046661734580994,
       "learning_rate": 0.0003964292640879215,
-      "loss": 3.4482,
+      "loss": 3.4536,
       "step": 31600
     },
     {
       "epoch": 3.4065224410720054,
-      "grad_norm": 0.6382803916931152,
+      "grad_norm": 0.5836654305458069,
       "learning_rate": 0.0003961060230578601,
-      "loss": 3.4746,
+      "loss": 3.4565,
       "step": 31650
     },
     {
       "epoch": 3.4119039931116135,
-      "grad_norm": 0.6442477107048035,
+      "grad_norm": 0.604293942451477,
       "learning_rate": 0.00039578278202779865,
-      "loss": 3.4505,
+      "loss": 3.4612,
       "step": 31700
     },
     {
       "epoch": 3.4172855451512216,
-      "grad_norm": 0.578318178653717,
+      "grad_norm": 0.608776330947876,
       "learning_rate": 0.00039545954099773725,
-      "loss": 3.4637,
+      "loss": 3.4494,
       "step": 31750
     },
     {
       "epoch": 3.4226670971908297,
-      "grad_norm": 0.6376295685768127,
+      "grad_norm": 0.8002951145172119,
       "learning_rate": 0.0003951362999676759,
-      "loss": 3.4703,
+      "loss": 3.466,
       "step": 31800
     },
     {
       "epoch": 3.428048649230438,
-      "grad_norm": 0.5918070673942566,
-      "learning_rate": 0.00039481305893761444,
-      "loss": 3.4539,
+      "grad_norm": 0.6348261833190918,
+      "learning_rate": 0.0003948195237582157,
+      "loss": 3.4507,
       "step": 31850
     },
     {
       "epoch": 3.4334302012700464,
-      "grad_norm": 0.6081002354621887,
+      "grad_norm": 0.5637660622596741,
       "learning_rate": 0.0003944962827281543,
-      "loss": 3.4505,
+      "loss": 3.4417,
       "step": 31900
     },
     {
       "epoch": 3.4388117533096545,
-      "grad_norm": 0.6261876225471497,
+      "grad_norm": 0.5938000082969666,
       "learning_rate": 0.00039417304169809284,
-      "loss": 3.4755,
+      "loss": 3.4679,
       "step": 31950
     },
     {
       "epoch": 3.4441933053492626,
-      "grad_norm": 0.6414044499397278,
+      "grad_norm": 0.6145411133766174,
       "learning_rate": 0.00039384980066803143,
-      "loss": 3.4556,
+      "loss": 3.4724,
       "step": 32000
     },
     {
       "epoch": 3.4441933053492626,
-      "eval_accuracy": 0.37261671248740036,
-      "eval_loss": 3.481884002685547,
-      "eval_runtime": 181.6828,
-      "eval_samples_per_second": 99.134,
-      "eval_steps_per_second": 6.198,
+      "eval_accuracy": 0.37322875420445933,
+      "eval_loss": 3.4761388301849365,
+      "eval_runtime": 180.5011,
+      "eval_samples_per_second": 99.783,
+      "eval_steps_per_second": 6.238,
       "step": 32000
     },
     {
       "epoch": 3.449574857388871,
-      "grad_norm": 0.6231973767280579,
+      "grad_norm": 0.6383312940597534,
       "learning_rate": 0.00039352655963796997,
-      "loss": 3.4596,
+      "loss": 3.4685,
       "step": 32050
     },
     {
       "epoch": 3.4549564094284793,
-      "grad_norm": 0.6971520781517029,
+      "grad_norm": 0.6065367460250854,
       "learning_rate": 0.00039320331860790857,
-      "loss": 3.4584,
+      "loss": 3.4559,
       "step": 32100
     },
     {
       "epoch": 3.4603379614680874,
-      "grad_norm": 0.6719905734062195,
+      "grad_norm": 0.6150339841842651,
       "learning_rate": 0.0003928800775778472,
-      "loss": 3.4615,
+      "loss": 3.457,
       "step": 32150
     },
     {
       "epoch": 3.4657195135076955,
-      "grad_norm": 0.7227779626846313,
+      "grad_norm": 0.5929863452911377,
       "learning_rate": 0.00039255683654778576,
-      "loss": 3.4486,
+      "loss": 3.4587,
       "step": 32200
     },
     {
       "epoch": 3.471101065547304,
-      "grad_norm": 0.6186626553535461,
+      "grad_norm": 0.593113362789154,
       "learning_rate": 0.00039223359551772435,
-      "loss": 3.4744,
+      "loss": 3.4586,
       "step": 32250
     },
     {
       "epoch": 3.476482617586912,
-      "grad_norm": 0.6083580851554871,
+      "grad_norm": 0.5945250988006592,
       "learning_rate": 0.00039191035448766294,
-      "loss": 3.4683,
+      "loss": 3.4731,
       "step": 32300
     },
     {
       "epoch": 3.4818641696265202,
-      "grad_norm": 0.6660602688789368,
+      "grad_norm": 0.6576330661773682,
       "learning_rate": 0.0003915871134576015,
-      "loss": 3.4638,
+      "loss": 3.4643,
       "step": 32350
     },
     {
       "epoch": 3.4872457216661283,
-      "grad_norm": 0.6296412944793701,
+      "grad_norm": 0.599430501461029,
       "learning_rate": 0.0003912638724275401,
-      "loss": 3.4557,
+      "loss": 3.4455,
       "step": 32400
     },
     {
       "epoch": 3.492627273705737,
-      "grad_norm": 0.6064853072166443,
+      "grad_norm": 0.6346076130867004,
       "learning_rate": 0.00039094063139747873,
-      "loss": 3.4495,
+      "loss": 3.4385,
       "step": 32450
     },
     {
       "epoch": 3.498008825745345,
-      "grad_norm": 0.6372223496437073,
+      "grad_norm": 0.659796953201294,
       "learning_rate": 0.00039061739036741727,
-      "loss": 3.4593,
+      "loss": 3.4666,
       "step": 32500
     },
     {
       "epoch": 3.503390377784953,
-      "grad_norm": 0.6473979949951172,
+      "grad_norm": 0.596899688243866,
       "learning_rate": 0.00039029414933735586,
-      "loss": 3.4618,
+      "loss": 3.4537,
       "step": 32550
     },
     {
       "epoch": 3.5087719298245617,
-      "grad_norm": 0.5884780883789062,
+      "grad_norm": 0.5908325910568237,
       "learning_rate": 0.0003899709083072944,
-      "loss": 3.4682,
+      "loss": 3.4522,
       "step": 32600
     },
     {
       "epoch": 3.5141534818641698,
-      "grad_norm": 0.6261137127876282,
+      "grad_norm": 0.6653596758842468,
       "learning_rate": 0.000389647667277233,
-      "loss": 3.4469,
+      "loss": 3.4514,
       "step": 32650
     },
     {
       "epoch": 3.519535033903778,
-      "grad_norm": 0.6143213510513306,
+      "grad_norm": 0.6003645658493042,
       "learning_rate": 0.00038932442624717165,
-      "loss": 3.4486,
+      "loss": 3.458,
       "step": 32700
     },
     {
       "epoch": 3.524916585943386,
-      "grad_norm": 0.6216268539428711,
+      "grad_norm": 0.6156851649284363,
       "learning_rate": 0.0003890011852171102,
-      "loss": 3.4687,
+      "loss": 3.4705,
       "step": 32750
     },
     {
       "epoch": 3.530298137982994,
-      "grad_norm": 0.6284915208816528,
+      "grad_norm": 0.6596025228500366,
       "learning_rate": 0.0003886779441870488,
-      "loss": 3.45,
+      "loss": 3.456,
       "step": 32800
     },
     {
       "epoch": 3.5356796900226026,
-      "grad_norm": 0.7261960506439209,
+      "grad_norm": 0.6155264377593994,
       "learning_rate": 0.0003883547031569874,
-      "loss": 3.461,
+      "loss": 3.4624,
       "step": 32850
     },
     {
       "epoch": 3.5410612420622107,
-      "grad_norm": 0.6670135855674744,
+      "grad_norm": 0.8467185497283936,
       "learning_rate": 0.0003880314621269259,
-      "loss": 3.4522,
+      "loss": 3.4456,
       "step": 32900
     },
     {
       "epoch": 3.546442794101819,
-      "grad_norm": 0.6466349959373474,
+      "grad_norm": 0.6377886533737183,
       "learning_rate": 0.0003877082210968645,
-      "loss": 3.4644,
+      "loss": 3.4632,
       "step": 32950
     },
     {
       "epoch": 3.5518243461414274,
-      "grad_norm": 0.5869767069816589,
+      "grad_norm": 0.6660758256912231,
       "learning_rate": 0.00038738498006680316,
-      "loss": 3.4758,
+      "loss": 3.4476,
       "step": 33000
     },
     {
       "epoch": 3.5518243461414274,
-      "eval_accuracy": 0.37266495436958996,
-      "eval_loss": 3.4755313396453857,
-      "eval_runtime": 181.8939,
-      "eval_samples_per_second": 99.019,
-      "eval_steps_per_second": 6.19,
+      "eval_accuracy": 0.3736872693910364,
+      "eval_loss": 3.469855546951294,
+      "eval_runtime": 181.5358,
+      "eval_samples_per_second": 99.215,
+      "eval_steps_per_second": 6.203,
       "step": 33000
     },
     {
       "epoch": 3.5572058981810355,
-      "grad_norm": 0.6126701831817627,
+      "grad_norm": 0.6835556626319885,
       "learning_rate": 0.0003870617390367417,
-      "loss": 3.4703,
+      "loss": 3.4556,
       "step": 33050
     },
     {
       "epoch": 3.5625874502206436,
-      "grad_norm": 0.6517377495765686,
+      "grad_norm": 0.6231185793876648,
       "learning_rate": 0.0003867384980066803,
-      "loss": 3.4547,
+      "loss": 3.4635,
       "step": 33100
     },
     {
       "epoch": 3.5679690022602517,
-      "grad_norm": 0.6763390898704529,
+      "grad_norm": 0.698049008846283,
       "learning_rate": 0.00038641525697661884,
-      "loss": 3.4589,
+      "loss": 3.4601,
       "step": 33150
     },
     {
       "epoch": 3.57335055429986,
-      "grad_norm": 0.6317027807235718,
+      "grad_norm": 0.6271853446960449,
       "learning_rate": 0.00038609201594655743,
-      "loss": 3.4612,
+      "loss": 3.4541,
       "step": 33200
     },
     {
       "epoch": 3.5787321063394684,
-      "grad_norm": 0.6396004557609558,
+      "grad_norm": 0.5717801451683044,
       "learning_rate": 0.000385768774916496,
-      "loss": 3.4317,
+      "loss": 3.4483,
       "step": 33250
     },
     {
       "epoch": 3.5841136583790765,
-      "grad_norm": 0.6065496802330017,
+      "grad_norm": 0.6370962858200073,
       "learning_rate": 0.0003854455338864346,
-      "loss": 3.4426,
+      "loss": 3.4662,
       "step": 33300
     },
     {
       "epoch": 3.5894952104186846,
-      "grad_norm": 0.6124275326728821,
+      "grad_norm": 0.6097142696380615,
       "learning_rate": 0.0003851222928563732,
-      "loss": 3.4479,
+      "loss": 3.4462,
       "step": 33350
     },
     {
       "epoch": 3.594876762458293,
-      "grad_norm": 0.6135236620903015,
+      "grad_norm": 0.6334971785545349,
       "learning_rate": 0.0003847990518263118,
-      "loss": 3.468,
+      "loss": 3.4714,
       "step": 33400
     },
     {
       "epoch": 3.6002583144979012,
-      "grad_norm": 0.6805423498153687,
+      "grad_norm": 0.5804086923599243,
       "learning_rate": 0.00038447581079625035,
-      "loss": 3.4838,
+      "loss": 3.4707,
       "step": 33450
     },
     {
       "epoch": 3.6056398665375093,
-      "grad_norm": 0.5888933539390564,
+      "grad_norm": 0.8195416331291199,
       "learning_rate": 0.00038415256976618895,
-      "loss": 3.4487,
+      "loss": 3.4643,
       "step": 33500
     },
     {
       "epoch": 3.611021418577118,
-      "grad_norm": 0.6695815920829773,
+      "grad_norm": 0.6207466125488281,
       "learning_rate": 0.0003838293287361275,
-      "loss": 3.4717,
+      "loss": 3.4629,
       "step": 33550
     },
     {
       "epoch": 3.616402970616726,
-      "grad_norm": 0.6879743337631226,
+      "grad_norm": 0.6233764290809631,
       "learning_rate": 0.00038350608770606613,
-      "loss": 3.4645,
+      "loss": 3.4553,
       "step": 33600
     },
     {
       "epoch": 3.621784522656334,
-      "grad_norm": 0.549407958984375,
+      "grad_norm": 0.5710169076919556,
       "learning_rate": 0.00038318284667600473,
-      "loss": 3.4642,
+      "loss": 3.4537,
       "step": 33650
     },
     {
       "epoch": 3.627166074695942,
-      "grad_norm": 0.5964043736457825,
+      "grad_norm": 0.7211458086967468,
       "learning_rate": 0.00038285960564594327,
-      "loss": 3.4735,
+      "loss": 3.4667,
       "step": 33700
     },
     {
       "epoch": 3.6325476267355503,
-      "grad_norm": 0.6675652265548706,
+      "grad_norm": 0.6359925866127014,
       "learning_rate": 0.00038253636461588186,
-      "loss": 3.4586,
+      "loss": 3.4642,
       "step": 33750
     },
     {
       "epoch": 3.637929178775159,
-      "grad_norm": 0.6164919137954712,
+      "grad_norm": 0.6265870332717896,
       "learning_rate": 0.00038221312358582046,
-      "loss": 3.4589,
+      "loss": 3.4435,
       "step": 33800
     },
     {
       "epoch": 3.643310730814767,
-      "grad_norm": 0.6311922073364258,
+      "grad_norm": 0.6150044798851013,
       "learning_rate": 0.00038188988255575905,
-      "loss": 3.4587,
+      "loss": 3.4613,
       "step": 33850
     },
     {
       "epoch": 3.648692282854375,
-      "grad_norm": 0.6482579708099365,
+      "grad_norm": 0.6175883412361145,
       "learning_rate": 0.00038156664152569765,
-      "loss": 3.4482,
+      "loss": 3.4483,
       "step": 33900
     },
     {
       "epoch": 3.6540738348939836,
-      "grad_norm": 0.7032673358917236,
+      "grad_norm": 0.6008051037788391,
       "learning_rate": 0.00038124340049563624,
-      "loss": 3.4705,
+      "loss": 3.4536,
       "step": 33950
     },
     {
       "epoch": 3.6594553869335917,
-      "grad_norm": 0.592106819152832,
+      "grad_norm": 0.5510129928588867,
       "learning_rate": 0.0003809201594655748,
-      "loss": 3.455,
+      "loss": 3.4658,
       "step": 34000
     },
     {
       "epoch": 3.6594553869335917,
-      "eval_accuracy": 0.37426226047335065,
-      "eval_loss": 3.4661221504211426,
-      "eval_runtime": 181.561,
-      "eval_samples_per_second": 99.201,
-      "eval_steps_per_second": 6.202,
+      "eval_accuracy": 0.37424509331707595,
+      "eval_loss": 3.464181423187256,
+      "eval_runtime": 180.1795,
+      "eval_samples_per_second": 99.961,
+      "eval_steps_per_second": 6.249,
       "step": 34000
     },
     {
       "epoch": 3.6648369389732,
-      "grad_norm": 0.5778534412384033,
+      "grad_norm": 0.6249406337738037,
       "learning_rate": 0.0003805969184355134,
-      "loss": 3.4494,
+      "loss": 3.4448,
       "step": 34050
     },
     {
       "epoch": 3.670218491012808,
-      "grad_norm": 0.5945811867713928,
+      "grad_norm": 0.5980893969535828,
       "learning_rate": 0.0003802736774054519,
-      "loss": 3.4473,
+      "loss": 3.4447,
       "step": 34100
     },
     {
       "epoch": 3.675600043052416,
-      "grad_norm": 0.6176246404647827,
+      "grad_norm": 0.6434440612792969,
       "learning_rate": 0.00037995043637539057,
-      "loss": 3.455,
+      "loss": 3.4546,
       "step": 34150
     },
     {
       "epoch": 3.6809815950920246,
-      "grad_norm": 0.6745234727859497,
+      "grad_norm": 0.6532515287399292,
       "learning_rate": 0.00037962719534532916,
-      "loss": 3.4559,
+      "loss": 3.4351,
       "step": 34200
     },
     {
       "epoch": 3.6863631471316327,
-      "grad_norm": 0.5692402124404907,
+      "grad_norm": 0.5936521291732788,
       "learning_rate": 0.0003793039543152677,
-      "loss": 3.441,
+      "loss": 3.4624,
       "step": 34250
     },
     {
       "epoch": 3.691744699171241,
-      "grad_norm": 0.6138830184936523,
+      "grad_norm": 0.6500722169876099,
       "learning_rate": 0.0003789807132852063,
-      "loss": 3.4481,
+      "loss": 3.4571,
       "step": 34300
     },
     {
       "epoch": 3.6971262512108494,
-      "grad_norm": 0.6347447037696838,
+      "grad_norm": 0.6223753094673157,
       "learning_rate": 0.0003786574722551449,
-      "loss": 3.4508,
+      "loss": 3.4498,
       "step": 34350
     },
     {
       "epoch": 3.7025078032504575,
-      "grad_norm": 0.6304515600204468,
+      "grad_norm": 0.5894403457641602,
       "learning_rate": 0.00037833423122508343,
-      "loss": 3.4593,
+      "loss": 3.4655,
       "step": 34400
     },
     {
       "epoch": 3.7078893552900656,
-      "grad_norm": 0.6190226078033447,
+      "grad_norm": 0.5831552743911743,
       "learning_rate": 0.0003780109901950221,
-      "loss": 3.4571,
+      "loss": 3.4511,
       "step": 34450
     },
     {
       "epoch": 3.713270907329674,
-      "grad_norm": 0.6307307481765747,
+      "grad_norm": 0.6428442001342773,
       "learning_rate": 0.0003776877491649607,
-      "loss": 3.4505,
+      "loss": 3.4404,
       "step": 34500
     },
     {
       "epoch": 3.7186524593692822,
-      "grad_norm": 0.6326085925102234,
+      "grad_norm": 0.5852512121200562,
       "learning_rate": 0.0003773645081348992,
-      "loss": 3.4599,
+      "loss": 3.4619,
       "step": 34550
     },
     {
       "epoch": 3.7240340114088903,
-      "grad_norm": 0.6394867897033691,
+      "grad_norm": 0.5681474208831787,
       "learning_rate": 0.0003770412671048378,
-      "loss": 3.4568,
+      "loss": 3.4679,
       "step": 34600
     },
     {
       "epoch": 3.7294155634484984,
-      "grad_norm": 0.633033037185669,
+      "grad_norm": 0.6105179190635681,
       "learning_rate": 0.00037671802607477635,
-      "loss": 3.4429,
+      "loss": 3.4377,
       "step": 34650
     },
     {
       "epoch": 3.7347971154881066,
-      "grad_norm": 0.6172804236412048,
+      "grad_norm": 0.6766160130500793,
       "learning_rate": 0.000376394785044715,
-      "loss": 3.4633,
+      "loss": 3.4651,
       "step": 34700
     },
     {
       "epoch": 3.740178667527715,
-      "grad_norm": 0.6102865934371948,
+      "grad_norm": 0.6639159321784973,
       "learning_rate": 0.0003760715440146536,
-      "loss": 3.4695,
+      "loss": 3.4677,
       "step": 34750
     },
     {
       "epoch": 3.745560219567323,
-      "grad_norm": 0.6239819526672363,
+      "grad_norm": 0.5689606666564941,
       "learning_rate": 0.00037574830298459214,
-      "loss": 3.4567,
+      "loss": 3.4487,
       "step": 34800
     },
     {
       "epoch": 3.7509417716069313,
-      "grad_norm": 0.6692067980766296,
+      "grad_norm": 0.6562086343765259,
       "learning_rate": 0.00037542506195453073,
-      "loss": 3.4591,
+      "loss": 3.4656,
       "step": 34850
     },
     {
       "epoch": 3.75632332364654,
-      "grad_norm": 0.589547336101532,
+      "grad_norm": 0.6432766318321228,
       "learning_rate": 0.0003751018209244693,
-      "loss": 3.4723,
+      "loss": 3.4544,
       "step": 34900
     },
     {
       "epoch": 3.761704875686148,
-      "grad_norm": 0.6273283958435059,
+      "grad_norm": 0.6148874759674072,
       "learning_rate": 0.00037477857989440787,
-      "loss": 3.4668,
+      "loss": 3.4577,
       "step": 34950
     },
     {
       "epoch": 3.767086427725756,
-      "grad_norm": 0.6687270402908325,
-      "learning_rate": 0.00037446180368494767,
-      "loss": 3.4728,
+      "grad_norm": 0.7988006472587585,
+      "learning_rate": 0.0003744553388643465,
+      "loss": 3.4444,
       "step": 35000
     },
     {
       "epoch": 3.767086427725756,
-      "eval_accuracy": 0.3749564437735974,
-      "eval_loss": 3.4588773250579834,
-      "eval_runtime": 181.7474,
-      "eval_samples_per_second": 99.099,
-      "eval_steps_per_second": 6.195,
+      "eval_accuracy": 0.3748263862668834,
+      "eval_loss": 3.4560623168945312,
+      "eval_runtime": 181.5342,
+      "eval_samples_per_second": 99.215,
+      "eval_steps_per_second": 6.203,
       "step": 35000
     },
     {
       "epoch": 3.772467979765364,
-      "grad_norm": 0.657584547996521,
-      "learning_rate": 0.00037413856265488627,
-      "loss": 3.4584,
+      "grad_norm": 0.7088155150413513,
+      "learning_rate": 0.0003741320978342851,
+      "loss": 3.4583,
       "step": 35050
     },
     {
       "epoch": 3.7778495318049723,
-      "grad_norm": 0.6070948839187622,
-      "learning_rate": 0.0003738153216248249,
-      "loss": 3.4495,
+      "grad_norm": 0.5917738676071167,
+      "learning_rate": 0.00037380885680422365,
+      "loss": 3.4517,
       "step": 35100
     },
     {
       "epoch": 3.783231083844581,
-      "grad_norm": 0.6411889791488647,
-      "learning_rate": 0.00037349208059476346,
-      "loss": 3.4406,
+      "grad_norm": 0.5819308161735535,
+      "learning_rate": 0.00037348561577416224,
+      "loss": 3.448,
       "step": 35150
     },
     {
       "epoch": 3.788612635884189,
-      "grad_norm": 0.6415666341781616,
-      "learning_rate": 0.00037316883956470205,
-      "loss": 3.4803,
+      "grad_norm": 0.6338306069374084,
+      "learning_rate": 0.0003731623747441008,
+      "loss": 3.4711,
       "step": 35200
     },
     {
       "epoch": 3.793994187923797,
-      "grad_norm": 0.6426709890365601,
-      "learning_rate": 0.00037284559853464064,
-      "loss": 3.4618,
+      "grad_norm": 0.6566649079322815,
+      "learning_rate": 0.0003728391337140394,
+      "loss": 3.4573,
       "step": 35250
     },
     {
       "epoch": 3.7993757399634056,
-      "grad_norm": 0.6597058176994324,
-      "learning_rate": 0.0003725223575045792,
-      "loss": 3.4453,
+      "grad_norm": 0.6187413930892944,
+      "learning_rate": 0.00037251589268397803,
+      "loss": 3.4363,
       "step": 35300
     },
     {
       "epoch": 3.8047572920030137,
-      "grad_norm": 0.6358980536460876,
-      "learning_rate": 0.0003721991164745178,
-      "loss": 3.4715,
+      "grad_norm": 0.6591097712516785,
+      "learning_rate": 0.00037219265165391657,
+      "loss": 3.4629,
       "step": 35350
     },
     {
       "epoch": 3.810138844042622,
-      "grad_norm": 0.6410354375839233,
-      "learning_rate": 0.00037187587544445643,
-      "loss": 3.4647,
+      "grad_norm": 0.6349477171897888,
+      "learning_rate": 0.00037186941062385516,
+      "loss": 3.4627,
       "step": 35400
     },
     {
       "epoch": 3.8155203960822304,
-      "grad_norm": 0.6233660578727722,
-      "learning_rate": 0.00037155263441439497,
-      "loss": 3.4698,
+      "grad_norm": 0.6552199125289917,
+      "learning_rate": 0.0003715461695937937,
+      "loss": 3.4575,
       "step": 35450
     },
     {
       "epoch": 3.8209019481218385,
-      "grad_norm": 0.628305196762085,
-      "learning_rate": 0.00037122939338433356,
-      "loss": 3.4537,
+      "grad_norm": 0.5953060388565063,
+      "learning_rate": 0.0003712229285637323,
+      "loss": 3.4207,
       "step": 35500
     },
     {
       "epoch": 3.8262835001614466,
-      "grad_norm": 0.6736071705818176,
-      "learning_rate": 0.0003709061523542721,
-      "loss": 3.4576,
+      "grad_norm": 0.6357983350753784,
+      "learning_rate": 0.0003708996875336709,
+      "loss": 3.4471,
       "step": 35550
     },
     {
       "epoch": 3.8316650522010547,
-      "grad_norm": 0.6532096266746521,
-      "learning_rate": 0.0003705829113242107,
-      "loss": 3.4494,
+      "grad_norm": 0.6096036434173584,
+      "learning_rate": 0.0003705764465036095,
+      "loss": 3.4506,
       "step": 35600
     },
     {
       "epoch": 3.837046604240663,
-      "grad_norm": 0.5983206629753113,
-      "learning_rate": 0.00037025967029414935,
-      "loss": 3.4538,
+      "grad_norm": 0.6250810623168945,
+      "learning_rate": 0.0003702532054735481,
+      "loss": 3.4284,
       "step": 35650
     },
     {
       "epoch": 3.8424281562802713,
-      "grad_norm": 0.6398340463638306,
-      "learning_rate": 0.0003699364292640879,
-      "loss": 3.4738,
+      "grad_norm": 0.6065813899040222,
+      "learning_rate": 0.0003699299644434867,
+      "loss": 3.4517,
       "step": 35700
     },
     {
       "epoch": 3.8478097083198795,
-      "grad_norm": 0.5982816219329834,
-      "learning_rate": 0.0003696131882340265,
-      "loss": 3.4687,
+      "grad_norm": 0.6180517077445984,
+      "learning_rate": 0.0003696067234134252,
+      "loss": 3.4675,
       "step": 35750
     },
     {
       "epoch": 3.8531912603594876,
-      "grad_norm": 0.6313088536262512,
-      "learning_rate": 0.0003692899472039651,
-      "loss": 3.4527,
+      "grad_norm": 0.6146745085716248,
+      "learning_rate": 0.0003692834823833638,
+      "loss": 3.4389,
       "step": 35800
     },
     {
       "epoch": 3.858572812399096,
-      "grad_norm": 0.6521557569503784,
-      "learning_rate": 0.0003689667061739036,
-      "loss": 3.4547,
+      "grad_norm": 0.6117834448814392,
+      "learning_rate": 0.00036896024135330246,
+      "loss": 3.4298,
       "step": 35850
     },
     {
       "epoch": 3.863954364438704,
-      "grad_norm": 0.6223726272583008,
-      "learning_rate": 0.0003686434651438422,
-      "loss": 3.4714,
+      "grad_norm": 0.6121900081634521,
+      "learning_rate": 0.000368637000323241,
+      "loss": 3.4516,
       "step": 35900
     },
     {
       "epoch": 3.8693359164783123,
-      "grad_norm": 0.6917114853858948,
+      "grad_norm": 0.6214909553527832,
       "learning_rate": 0.00036832022411378086,
-      "loss": 3.4696,
+      "loss": 3.4468,
       "step": 35950
     },
     {
       "epoch": 3.8747174685179204,
-      "grad_norm": 0.606867253780365,
+      "grad_norm": 0.6564396619796753,
       "learning_rate": 0.0003679969830837194,
-      "loss": 3.4347,
+      "loss": 3.4371,
       "step": 36000
     },
     {
       "epoch": 3.8747174685179204,
-      "eval_accuracy": 0.3753462903350757,
-      "eval_loss": 3.4530980587005615,
-      "eval_runtime": 181.4546,
-      "eval_samples_per_second": 99.259,
-      "eval_steps_per_second": 6.205,
+      "eval_accuracy": 0.3759991855379529,
+      "eval_loss": 3.450083017349243,
+      "eval_runtime": 180.0679,
+      "eval_samples_per_second": 100.023,
+      "eval_steps_per_second": 6.253,
       "step": 36000
     },
     {
       "epoch": 3.8800990205575285,
-      "grad_norm": 0.6278716325759888,
+      "grad_norm": 0.6633586883544922,
       "learning_rate": 0.000367673742053658,
-      "loss": 3.4351,
+      "loss": 3.4457,
       "step": 36050
     },
     {
       "epoch": 3.885480572597137,
-      "grad_norm": 0.625916600227356,
+      "grad_norm": 0.6210970282554626,
       "learning_rate": 0.00036735050102359654,
-      "loss": 3.4415,
+      "loss": 3.4522,
       "step": 36100
     },
     {
       "epoch": 3.890862124636745,
-      "grad_norm": 0.6142260432243347,
+      "grad_norm": 0.6026017665863037,
       "learning_rate": 0.00036702725999353513,
-      "loss": 3.4539,
+      "loss": 3.4468,
       "step": 36150
     },
     {
       "epoch": 3.8962436766763533,
-      "grad_norm": 0.6766964793205261,
+      "grad_norm": 0.6041121482849121,
       "learning_rate": 0.0003667040189634737,
-      "loss": 3.4584,
+      "loss": 3.4567,
       "step": 36200
     },
     {
       "epoch": 3.901625228715962,
-      "grad_norm": 0.6694559454917908,
+      "grad_norm": 0.6270918846130371,
       "learning_rate": 0.0003663807779334123,
-      "loss": 3.4564,
+      "loss": 3.4568,
       "step": 36250
     },
     {
       "epoch": 3.90700678075557,
-      "grad_norm": 0.6292983889579773,
+      "grad_norm": 0.6320198774337769,
       "learning_rate": 0.0003660575369033509,
-      "loss": 3.4572,
+      "loss": 3.4723,
       "step": 36300
     },
     {
       "epoch": 3.912388332795178,
-      "grad_norm": 0.6223490834236145,
+      "grad_norm": 0.5653135180473328,
       "learning_rate": 0.0003657342958732895,
-      "loss": 3.4521,
+      "loss": 3.4367,
       "step": 36350
     },
     {
       "epoch": 3.9177698848347866,
-      "grad_norm": 0.6118103861808777,
+      "grad_norm": 0.6417329907417297,
       "learning_rate": 0.00036541105484322805,
-      "loss": 3.4535,
+      "loss": 3.4461,
       "step": 36400
     },
     {
       "epoch": 3.9231514368743947,
-      "grad_norm": 0.645898163318634,
+      "grad_norm": 0.6318935751914978,
       "learning_rate": 0.00036508781381316665,
-      "loss": 3.4518,
+      "loss": 3.4404,
       "step": 36450
     },
     {
       "epoch": 3.928532988914003,
-      "grad_norm": 0.6415484547615051,
+      "grad_norm": 0.5992429256439209,
       "learning_rate": 0.0003647645727831053,
-      "loss": 3.4585,
+      "loss": 3.4482,
       "step": 36500
     },
     {
       "epoch": 3.933914540953611,
-      "grad_norm": 0.6804546117782593,
+      "grad_norm": 0.6318196058273315,
       "learning_rate": 0.00036444133175304384,
-      "loss": 3.4533,
+      "loss": 3.4537,
       "step": 36550
     },
     {
       "epoch": 3.939296092993219,
-      "grad_norm": 0.6136488318443298,
+      "grad_norm": 0.5906317830085754,
       "learning_rate": 0.00036411809072298243,
-      "loss": 3.4474,
+      "loss": 3.445,
       "step": 36600
     },
     {
       "epoch": 3.9446776450328276,
-      "grad_norm": 0.6238126158714294,
+      "grad_norm": 0.6364299058914185,
       "learning_rate": 0.00036379484969292097,
-      "loss": 3.458,
+      "loss": 3.4569,
       "step": 36650
     },
     {
       "epoch": 3.9500591970724357,
-      "grad_norm": 0.6238007545471191,
+      "grad_norm": 0.6347159743309021,
       "learning_rate": 0.00036347160866285956,
-      "loss": 3.4624,
+      "loss": 3.4618,
       "step": 36700
     },
     {
       "epoch": 3.955440749112044,
-      "grad_norm": 0.6192507743835449,
+      "grad_norm": 0.6349920630455017,
       "learning_rate": 0.0003631483676327981,
-      "loss": 3.4544,
+      "loss": 3.4575,
       "step": 36750
     },
     {
       "epoch": 3.9608223011516523,
-      "grad_norm": 0.6187044978141785,
+      "grad_norm": 0.6208349466323853,
       "learning_rate": 0.00036282512660273675,
-      "loss": 3.4372,
+      "loss": 3.4404,
       "step": 36800
     },
     {
       "epoch": 3.9662038531912605,
-      "grad_norm": 0.6006636619567871,
+      "grad_norm": 0.6194992661476135,
       "learning_rate": 0.00036250188557267535,
-      "loss": 3.4588,
+      "loss": 3.4502,
       "step": 36850
     },
     {
       "epoch": 3.9715854052308686,
-      "grad_norm": 0.6458033323287964,
+      "grad_norm": 0.662883996963501,
       "learning_rate": 0.0003621786445426139,
-      "loss": 3.4481,
+      "loss": 3.445,
       "step": 36900
     },
     {
       "epoch": 3.9769669572704767,
-      "grad_norm": 0.6214568614959717,
+      "grad_norm": 0.6490581035614014,
       "learning_rate": 0.0003618554035125525,
-      "loss": 3.4495,
+      "loss": 3.4478,
       "step": 36950
     },
     {
       "epoch": 3.9823485093100848,
-      "grad_norm": 0.6304638981819153,
+      "grad_norm": 0.5769525766372681,
       "learning_rate": 0.0003615321624824911,
-      "loss": 3.46,
+      "loss": 3.443,
       "step": 37000
     },
     {
       "epoch": 3.9823485093100848,
-      "eval_accuracy": 0.37597354345642875,
-      "eval_loss": 3.446321725845337,
-      "eval_runtime": 181.7245,
-      "eval_samples_per_second": 99.112,
-      "eval_steps_per_second": 6.196,
+      "eval_accuracy": 0.3764034829335106,
+      "eval_loss": 3.4439070224761963,
+      "eval_runtime": 180.3404,
+      "eval_samples_per_second": 99.872,
+      "eval_steps_per_second": 6.244,
       "step": 37000
     },
     {
       "epoch": 3.9877300613496933,
-      "grad_norm": 0.5896131992340088,
+      "grad_norm": 0.6289724707603455,
       "learning_rate": 0.0003612089214524296,
-      "loss": 3.4304,
+      "loss": 3.4323,
       "step": 37050
     },
     {
       "epoch": 3.9931116133893014,
-      "grad_norm": 0.6256299614906311,
+      "grad_norm": 0.6251957416534424,
       "learning_rate": 0.00036088568042236827,
-      "loss": 3.4539,
+      "loss": 3.4473,
       "step": 37100
     },
     {
       "epoch": 3.9984931654289095,
-      "grad_norm": 0.6513100862503052,
+      "grad_norm": 0.6285573840141296,
       "learning_rate": 0.00036056243939230686,
-      "loss": 3.4543,
+      "loss": 3.4515,
       "step": 37150
     },
     {
       "epoch": 4.003874717468518,
-      "grad_norm": 0.6809729337692261,
+      "grad_norm": 0.6383672952651978,
       "learning_rate": 0.0003602391983622454,
-      "loss": 3.382,
+      "loss": 3.3654,
       "step": 37200
     },
     {
       "epoch": 4.009256269508126,
-      "grad_norm": 0.6216638088226318,
+      "grad_norm": 0.5977559685707092,
       "learning_rate": 0.000359915957332184,
-      "loss": 3.3691,
+      "loss": 3.3621,
       "step": 37250
     },
     {
       "epoch": 4.014637821547734,
-      "grad_norm": 0.687582790851593,
+      "grad_norm": 0.568278431892395,
       "learning_rate": 0.00035959271630212254,
-      "loss": 3.3739,
+      "loss": 3.3411,
       "step": 37300
     },
     {
       "epoch": 4.020019373587343,
-      "grad_norm": 0.6538896560668945,
+      "grad_norm": 0.649253785610199,
       "learning_rate": 0.00035926947527206113,
-      "loss": 3.3553,
+      "loss": 3.3576,
       "step": 37350
     },
     {
       "epoch": 4.0254009256269505,
-      "grad_norm": 0.6100491881370544,
+      "grad_norm": 0.6648885011672974,
       "learning_rate": 0.0003589462342419998,
-      "loss": 3.3636,
+      "loss": 3.3602,
       "step": 37400
     },
     {
       "epoch": 4.030782477666559,
-      "grad_norm": 0.6782578825950623,
+      "grad_norm": 0.6154427528381348,
       "learning_rate": 0.0003586229932119383,
-      "loss": 3.3712,
+      "loss": 3.3567,
       "step": 37450
     },
     {
       "epoch": 4.036164029706168,
-      "grad_norm": 0.7136054039001465,
+      "grad_norm": 0.6308563947677612,
       "learning_rate": 0.0003582997521818769,
-      "loss": 3.3502,
+      "loss": 3.338,
       "step": 37500
     },
     {
       "epoch": 4.041545581745775,
-      "grad_norm": 1.0063740015029907,
+      "grad_norm": 0.5829379558563232,
       "learning_rate": 0.0003579765111518155,
-      "loss": 3.3518,
+      "loss": 3.3477,
       "step": 37550
     },
     {
       "epoch": 4.046927133785384,
-      "grad_norm": 0.6431665420532227,
+      "grad_norm": 0.6504759192466736,
       "learning_rate": 0.00035765327012175405,
-      "loss": 3.3673,
+      "loss": 3.3557,
       "step": 37600
     },
     {
       "epoch": 4.0523086858249915,
-      "grad_norm": 0.6726707220077515,
+      "grad_norm": 0.6111143231391907,
       "learning_rate": 0.0003573300290916927,
-      "loss": 3.365,
+      "loss": 3.3689,
       "step": 37650
     },
     {
       "epoch": 4.0576902378646,
-      "grad_norm": 0.6649792194366455,
+      "grad_norm": 0.6477968692779541,
       "learning_rate": 0.0003570067880616313,
-      "loss": 3.373,
+      "loss": 3.3882,
       "step": 37700
     },
     {
       "epoch": 4.063071789904209,
-      "grad_norm": 0.6573074460029602,
-      "learning_rate": 0.00035668354703156984,
-      "loss": 3.3614,
+      "grad_norm": 0.6770215034484863,
+      "learning_rate": 0.0003566900118521711,
+      "loss": 3.3522,
       "step": 37750
     },
     {
       "epoch": 4.068453341943816,
-      "grad_norm": 0.6372230052947998,
-      "learning_rate": 0.00035636030600150843,
-      "loss": 3.3596,
+      "grad_norm": 0.6108023524284363,
+      "learning_rate": 0.0003563667708221097,
+      "loss": 3.3573,
       "step": 37800
     },
     {
       "epoch": 4.073834893983425,
-      "grad_norm": 0.6461084485054016,
-      "learning_rate": 0.00035603706497144697,
-      "loss": 3.3685,
+      "grad_norm": 0.658203125,
+      "learning_rate": 0.00035604352979204824,
+      "loss": 3.3553,
       "step": 37850
     },
     {
       "epoch": 4.079216446023033,
-      "grad_norm": 0.6296938061714172,
-      "learning_rate": 0.00035571382394138557,
-      "loss": 3.3641,
+      "grad_norm": 0.6561046838760376,
+      "learning_rate": 0.00035572028876198683,
+      "loss": 3.3438,
       "step": 37900
     },
     {
       "epoch": 4.084597998062641,
-      "grad_norm": 0.658966600894928,
-      "learning_rate": 0.0003553905829113242,
-      "loss": 3.3773,
+      "grad_norm": 0.6152969598770142,
+      "learning_rate": 0.00035539704773192537,
+      "loss": 3.349,
       "step": 37950
     },
     {
       "epoch": 4.08997955010225,
-      "grad_norm": 0.6599716544151306,
-      "learning_rate": 0.00035506734188126275,
-      "loss": 3.3683,
+      "grad_norm": 0.6802772879600525,
+      "learning_rate": 0.00035507380670186397,
+      "loss": 3.3598,
       "step": 38000
     },
     {
       "epoch": 4.08997955010225,
-      "eval_accuracy": 0.3761422813912046,
-      "eval_loss": 3.44930100440979,
-      "eval_runtime": 181.3188,
-      "eval_samples_per_second": 99.333,
-      "eval_steps_per_second": 6.21,
+      "eval_accuracy": 0.37615390725020076,
+      "eval_loss": 3.4493589401245117,
+      "eval_runtime": 180.1809,
+      "eval_samples_per_second": 99.961,
+      "eval_steps_per_second": 6.249,
       "step": 38000
     },
     {
       "epoch": 4.095361102141858,
-      "grad_norm": 0.6307918429374695,
-      "learning_rate": 0.00035474410085120135,
-      "loss": 3.3854,
+      "grad_norm": 0.6440187692642212,
+      "learning_rate": 0.0003547505656718026,
+      "loss": 3.3704,
       "step": 38050
     },
     {
       "epoch": 4.100742654181466,
-      "grad_norm": 0.6225454211235046,
-      "learning_rate": 0.00035442085982113994,
-      "loss": 3.3698,
+      "grad_norm": 0.6484562158584595,
+      "learning_rate": 0.00035442732464174116,
+      "loss": 3.3886,
       "step": 38100
     },
     {
       "epoch": 4.106124206221074,
-      "grad_norm": 0.6287394762039185,
-      "learning_rate": 0.0003540976187910785,
-      "loss": 3.3919,
+      "grad_norm": 0.6503589153289795,
+      "learning_rate": 0.00035410408361167975,
+      "loss": 3.3757,
       "step": 38150
     },
     {
       "epoch": 4.111505758260682,
-      "grad_norm": 0.6210476756095886,
-      "learning_rate": 0.0003537743777610171,
-      "loss": 3.3621,
+      "grad_norm": 0.6673949956893921,
+      "learning_rate": 0.0003537808425816183,
+      "loss": 3.3554,
       "step": 38200
     },
     {
       "epoch": 4.1168873103002905,
-      "grad_norm": 0.6711219549179077,
-      "learning_rate": 0.00035345113673095573,
-      "loss": 3.3686,
+      "grad_norm": 0.646493136882782,
+      "learning_rate": 0.0003534576015515569,
+      "loss": 3.3822,
       "step": 38250
     },
     {
       "epoch": 4.122268862339899,
-      "grad_norm": 0.6280274391174316,
-      "learning_rate": 0.00035312789570089427,
-      "loss": 3.3742,
+      "grad_norm": 0.6250292658805847,
+      "learning_rate": 0.00035313436052149553,
+      "loss": 3.3642,
       "step": 38300
     },
     {
       "epoch": 4.127650414379507,
-      "grad_norm": 0.6348408460617065,
-      "learning_rate": 0.00035280465467083286,
-      "loss": 3.3871,
+      "grad_norm": 0.6226435303688049,
+      "learning_rate": 0.0003528111194914341,
+      "loss": 3.3765,
       "step": 38350
     },
     {
       "epoch": 4.133031966419115,
-      "grad_norm": 0.6895505785942078,
-      "learning_rate": 0.0003524814136407714,
-      "loss": 3.3586,
+      "grad_norm": 0.6346940398216248,
+      "learning_rate": 0.00035248787846137267,
+      "loss": 3.36,
       "step": 38400
     },
     {
       "epoch": 4.138413518458724,
-      "grad_norm": 0.7086821794509888,
-      "learning_rate": 0.00035215817261071,
-      "loss": 3.3695,
+      "grad_norm": 0.605279803276062,
+      "learning_rate": 0.00035216463743131126,
+      "loss": 3.3698,
       "step": 38450
     },
     {
       "epoch": 4.1437950704983315,
-      "grad_norm": 0.6616617441177368,
-      "learning_rate": 0.00035183493158064865,
-      "loss": 3.382,
+      "grad_norm": 0.6094624996185303,
+      "learning_rate": 0.0003518413964012498,
+      "loss": 3.3648,
       "step": 38500
     },
     {
       "epoch": 4.14917662253794,
-      "grad_norm": 0.6920830607414246,
-      "learning_rate": 0.0003515116905505872,
-      "loss": 3.4122,
+      "grad_norm": 0.6399276256561279,
+      "learning_rate": 0.0003515181553711884,
+      "loss": 3.4123,
       "step": 38550
     },
     {
       "epoch": 4.154558174577549,
-      "grad_norm": 0.7117165923118591,
-      "learning_rate": 0.0003511884495205258,
-      "loss": 3.3802,
+      "grad_norm": 0.702944278717041,
+      "learning_rate": 0.00035119491434112705,
+      "loss": 3.3685,
       "step": 38600
     },
     {
       "epoch": 4.159939726617156,
-      "grad_norm": 0.6522443890571594,
-      "learning_rate": 0.0003508652084904644,
-      "loss": 3.3718,
+      "grad_norm": 0.613124668598175,
+      "learning_rate": 0.0003508716733110656,
+      "loss": 3.3872,
       "step": 38650
     },
     {
       "epoch": 4.165321278656765,
-      "grad_norm": 0.7154022455215454,
-      "learning_rate": 0.0003505419674604029,
-      "loss": 3.3929,
+      "grad_norm": 0.5904596447944641,
+      "learning_rate": 0.0003505484322810042,
+      "loss": 3.3746,
       "step": 38700
     },
     {
       "epoch": 4.1707028306963725,
-      "grad_norm": 0.6403151750564575,
-      "learning_rate": 0.0003502187264303415,
-      "loss": 3.3938,
+      "grad_norm": 0.6269122958183289,
+      "learning_rate": 0.0003502251912509427,
+      "loss": 3.3911,
       "step": 38750
     },
     {
       "epoch": 4.176084382735981,
-      "grad_norm": 0.5948874354362488,
-      "learning_rate": 0.00034989548540028016,
-      "loss": 3.3894,
+      "grad_norm": 0.6426027417182922,
+      "learning_rate": 0.0003499019502208813,
+      "loss": 3.377,
       "step": 38800
     },
     {
       "epoch": 4.18146593477559,
-      "grad_norm": 0.6313942074775696,
-      "learning_rate": 0.0003495722443702187,
-      "loss": 3.384,
+      "grad_norm": 0.656601071357727,
+      "learning_rate": 0.0003495787091908199,
+      "loss": 3.3798,
       "step": 38850
     },
     {
       "epoch": 4.186847486815197,
-      "grad_norm": 0.6668844223022461,
-      "learning_rate": 0.0003492490033401573,
-      "loss": 3.3911,
+      "grad_norm": 0.6394505500793457,
+      "learning_rate": 0.0003492554681607585,
+      "loss": 3.382,
       "step": 38900
     },
     {
       "epoch": 4.192229038854806,
-      "grad_norm": 0.6406861543655396,
-      "learning_rate": 0.00034892576231009584,
-      "loss": 3.3706,
+      "grad_norm": 0.6042181849479675,
+      "learning_rate": 0.0003489322271306971,
+      "loss": 3.3857,
       "step": 38950
     },
     {
       "epoch": 4.197610590894414,
-      "grad_norm": 0.6086711883544922,
-      "learning_rate": 0.00034860252128003443,
-      "loss": 3.3682,
+      "grad_norm": 0.5869829654693604,
+      "learning_rate": 0.0003486089861006357,
+      "loss": 3.3833,
       "step": 39000
     },
     {
       "epoch": 4.197610590894414,
-      "eval_accuracy": 0.3768287503364165,
-      "eval_loss": 3.4479315280914307,
-      "eval_runtime": 181.8639,
-      "eval_samples_per_second": 99.036,
-      "eval_steps_per_second": 6.191,
+      "eval_accuracy": 0.37726357919345005,
+      "eval_loss": 3.4431540966033936,
+      "eval_runtime": 180.2754,
+      "eval_samples_per_second": 99.908,
+      "eval_steps_per_second": 6.246,
       "step": 39000
     },
     {
       "epoch": 4.202992142934022,
-      "grad_norm": 0.6679412722587585,
-      "learning_rate": 0.000348279280249973,
-      "loss": 3.3828,
+      "grad_norm": 0.6573967933654785,
+      "learning_rate": 0.00034828574507057424,
+      "loss": 3.3738,
       "step": 39050
     },
     {
       "epoch": 4.208373694973631,
-      "grad_norm": 0.6079999804496765,
-      "learning_rate": 0.0003479560392199116,
-      "loss": 3.3802,
+      "grad_norm": 0.6622105240821838,
+      "learning_rate": 0.00034796250404051283,
+      "loss": 3.381,
       "step": 39100
     },
     {
       "epoch": 4.213755247013238,
-      "grad_norm": 0.629065215587616,
+      "grad_norm": 0.6184806823730469,
       "learning_rate": 0.00034763926301045137,
-      "loss": 3.3894,
+      "loss": 3.3919,
       "step": 39150
     },
     {
       "epoch": 4.219136799052847,
-      "grad_norm": 0.6328132152557373,
+      "grad_norm": 0.652938723564148,
       "learning_rate": 0.00034731602198039,
-      "loss": 3.3766,
+      "loss": 3.374,
       "step": 39200
     },
     {
       "epoch": 4.224518351092455,
-      "grad_norm": 0.6607147455215454,
+      "grad_norm": 0.6615681648254395,
       "learning_rate": 0.0003469927809503286,
-      "loss": 3.3742,
+      "loss": 3.3722,
       "step": 39250
     },
     {
       "epoch": 4.229899903132063,
-      "grad_norm": 0.6375463008880615,
+      "grad_norm": 0.6392892599105835,
       "learning_rate": 0.00034666953992026716,
-      "loss": 3.3871,
+      "loss": 3.3706,
       "step": 39300
     },
     {
       "epoch": 4.2352814551716715,
-      "grad_norm": 0.6818897724151611,
+      "grad_norm": 0.678726851940155,
       "learning_rate": 0.00034634629889020575,
-      "loss": 3.3868,
+      "loss": 3.4001,
       "step": 39350
     },
     {
       "epoch": 4.24066300721128,
-      "grad_norm": 0.6507583856582642,
+      "grad_norm": 0.643818736076355,
       "learning_rate": 0.00034602305786014435,
-      "loss": 3.3852,
+      "loss": 3.4025,
       "step": 39400
     },
     {
       "epoch": 4.246044559250888,
-      "grad_norm": 0.649346113204956,
+      "grad_norm": 0.6321120858192444,
       "learning_rate": 0.00034569981683008294,
-      "loss": 3.3816,
+      "loss": 3.388,
       "step": 39450
     },
     {
       "epoch": 4.251426111290496,
-      "grad_norm": 0.6513179540634155,
+      "grad_norm": 0.682064414024353,
       "learning_rate": 0.00034537657580002154,
-      "loss": 3.3709,
+      "loss": 3.3712,
       "step": 39500
     },
     {
       "epoch": 4.256807663330104,
-      "grad_norm": 0.7123865485191345,
+      "grad_norm": 0.657920777797699,
       "learning_rate": 0.00034505333476996013,
-      "loss": 3.3986,
+      "loss": 3.3968,
       "step": 39550
     },
     {
       "epoch": 4.2621892153697125,
-      "grad_norm": 0.7108580470085144,
+      "grad_norm": 0.6285057663917542,
       "learning_rate": 0.00034473009373989867,
-      "loss": 3.3923,
+      "loss": 3.3724,
       "step": 39600
     },
     {
       "epoch": 4.267570767409321,
-      "grad_norm": 0.6283742785453796,
+      "grad_norm": 0.594862163066864,
       "learning_rate": 0.00034440685270983727,
-      "loss": 3.401,
+      "loss": 3.3892,
       "step": 39650
     },
     {
       "epoch": 4.272952319448929,
-      "grad_norm": 0.6473816633224487,
+      "grad_norm": 0.6109697818756104,
       "learning_rate": 0.0003440836116797758,
-      "loss": 3.391,
+      "loss": 3.3754,
       "step": 39700
     },
     {
       "epoch": 4.278333871488537,
-      "grad_norm": 0.6747614145278931,
+      "grad_norm": 0.6800058484077454,
       "learning_rate": 0.00034376037064971445,
-      "loss": 3.401,
+      "loss": 3.3921,
       "step": 39750
     },
     {
       "epoch": 4.283715423528146,
-      "grad_norm": 0.6350759863853455,
+      "grad_norm": 0.6627162098884583,
       "learning_rate": 0.00034343712961965305,
-      "loss": 3.4075,
+      "loss": 3.4045,
       "step": 39800
     },
     {
       "epoch": 4.2890969755677535,
-      "grad_norm": 0.7255776524543762,
+      "grad_norm": 0.6985821723937988,
       "learning_rate": 0.0003431138885895916,
-      "loss": 3.3835,
+      "loss": 3.3997,
       "step": 39850
     },
     {
       "epoch": 4.294478527607362,
-      "grad_norm": 0.7065731287002563,
-      "learning_rate": 0.00034279711238013145,
-      "loss": 3.396,
+      "grad_norm": 0.6768974661827087,
+      "learning_rate": 0.0003427906475595302,
+      "loss": 3.383,
       "step": 39900
     },
     {
       "epoch": 4.299860079646971,
-      "grad_norm": 0.67261803150177,
-      "learning_rate": 0.00034247387135007,
-      "loss": 3.3983,
+      "grad_norm": 0.6420722603797913,
+      "learning_rate": 0.0003424674065294688,
+      "loss": 3.3841,
       "step": 39950
     },
     {
       "epoch": 4.305241631686578,
-      "grad_norm": 0.6579726934432983,
-      "learning_rate": 0.0003421506303200086,
-      "loss": 3.3876,
+      "grad_norm": 0.6062000393867493,
+      "learning_rate": 0.0003421441654994073,
+      "loss": 3.3864,
       "step": 40000
     },
     {
       "epoch": 4.305241631686578,
-      "eval_accuracy": 0.3775528131808122,
-      "eval_loss": 3.441394805908203,
-      "eval_runtime": 181.6119,
-      "eval_samples_per_second": 99.173,
-      "eval_steps_per_second": 6.2,
+      "eval_accuracy": 0.3777341548505746,
+      "eval_loss": 3.4375386238098145,
+      "eval_runtime": 180.0714,
+      "eval_samples_per_second": 100.021,
+      "eval_steps_per_second": 6.253,
       "step": 40000
     },
     {
       "epoch": 4.310623183726187,
-      "grad_norm": 0.6545993089675903,
-      "learning_rate": 0.0003418273892899471,
-      "loss": 3.382,
+      "grad_norm": 0.6825680732727051,
+      "learning_rate": 0.00034182092446934597,
+      "loss": 3.3811,
       "step": 40050
     },
     {
       "epoch": 4.3160047357657945,
-      "grad_norm": 0.5974786877632141,
-      "learning_rate": 0.0003415041482598858,
-      "loss": 3.3722,
+      "grad_norm": 0.6287062764167786,
+      "learning_rate": 0.00034149768343928456,
+      "loss": 3.3686,
       "step": 40100
     },
     {
       "epoch": 4.321386287805403,
-      "grad_norm": 0.6922400593757629,
-      "learning_rate": 0.00034118090722982437,
-      "loss": 3.3802,
+      "grad_norm": 0.7107475399971008,
+      "learning_rate": 0.0003411744424092231,
+      "loss": 3.3775,
       "step": 40150
     },
     {
       "epoch": 4.326767839845012,
-      "grad_norm": 0.6838219165802002,
-      "learning_rate": 0.0003408576661997629,
-      "loss": 3.39,
+      "grad_norm": 0.6253789663314819,
+      "learning_rate": 0.0003408512013791617,
+      "loss": 3.3867,
       "step": 40200
     },
     {
       "epoch": 4.332149391884619,
-      "grad_norm": 0.6342960000038147,
-      "learning_rate": 0.0003405344251697015,
-      "loss": 3.3819,
+      "grad_norm": 0.625191867351532,
+      "learning_rate": 0.00034052796034910024,
+      "loss": 3.3857,
       "step": 40250
     },
     {
       "epoch": 4.337530943924228,
-      "grad_norm": 0.7120860815048218,
+      "grad_norm": 0.6348721981048584,
       "learning_rate": 0.0003402111841396401,
-      "loss": 3.374,
+      "loss": 3.3708,
       "step": 40300
     },
     {
       "epoch": 4.342912495963836,
-      "grad_norm": 0.648949921131134,
+      "grad_norm": 0.6521097421646118,
       "learning_rate": 0.00033988794310957864,
-      "loss": 3.3776,
+      "loss": 3.387,
       "step": 40350
     },
     {
       "epoch": 4.348294048003444,
-      "grad_norm": 0.7276732325553894,
+      "grad_norm": 0.687777042388916,
       "learning_rate": 0.0003395647020795173,
-      "loss": 3.4097,
+      "loss": 3.3829,
       "step": 40400
     },
     {
       "epoch": 4.3536756000430525,
-      "grad_norm": 0.6344953179359436,
+      "grad_norm": 0.6225450038909912,
       "learning_rate": 0.0003392414610494559,
-      "loss": 3.3754,
+      "loss": 3.3816,
       "step": 40450
     },
     {
       "epoch": 4.359057152082661,
-      "grad_norm": 0.6971585154533386,
+      "grad_norm": 0.6652933359146118,
       "learning_rate": 0.0003389182200193944,
-      "loss": 3.4039,
+      "loss": 3.3944,
       "step": 40500
     },
     {
       "epoch": 4.364438704122269,
-      "grad_norm": 0.6294052600860596,
+      "grad_norm": 0.6676579713821411,
       "learning_rate": 0.000338594978989333,
-      "loss": 3.3791,
+      "loss": 3.3864,
       "step": 40550
     },
     {
       "epoch": 4.369820256161877,
-      "grad_norm": 0.6447839140892029,
+      "grad_norm": 0.7050386071205139,
       "learning_rate": 0.00033827173795927156,
-      "loss": 3.3871,
+      "loss": 3.4043,
       "step": 40600
     },
     {
       "epoch": 4.375201808201485,
-      "grad_norm": 0.6472920179367065,
+      "grad_norm": 0.687006413936615,
       "learning_rate": 0.00033794849692921015,
-      "loss": 3.3886,
+      "loss": 3.3879,
       "step": 40650
     },
     {
       "epoch": 4.3805833602410935,
-      "grad_norm": 0.6552985310554504,
+      "grad_norm": 0.6358218789100647,
       "learning_rate": 0.0003376252558991488,
-      "loss": 3.3904,
+      "loss": 3.3971,
       "step": 40700
     },
     {
       "epoch": 4.385964912280702,
-      "grad_norm": 0.6818690896034241,
+      "grad_norm": 0.6476041674613953,
       "learning_rate": 0.00033730201486908734,
-      "loss": 3.3947,
+      "loss": 3.3801,
       "step": 40750
     },
     {
       "epoch": 4.39134646432031,
-      "grad_norm": 0.646091639995575,
+      "grad_norm": 0.6176375150680542,
       "learning_rate": 0.00033697877383902594,
-      "loss": 3.4024,
+      "loss": 3.3866,
       "step": 40800
     },
     {
       "epoch": 4.396728016359918,
-      "grad_norm": 0.6999764442443848,
+      "grad_norm": 0.6508418917655945,
       "learning_rate": 0.00033665553280896453,
-      "loss": 3.391,
+      "loss": 3.4075,
       "step": 40850
     },
     {
       "epoch": 4.402109568399527,
-      "grad_norm": 0.6619383096694946,
+      "grad_norm": 0.6436052918434143,
       "learning_rate": 0.00033633229177890307,
-      "loss": 3.4125,
+      "loss": 3.4054,
       "step": 40900
     },
     {
       "epoch": 4.4074911204391345,
-      "grad_norm": 0.8046988844871521,
+      "grad_norm": 0.691495954990387,
       "learning_rate": 0.00033600905074884167,
-      "loss": 3.373,
+      "loss": 3.367,
       "step": 40950
     },
     {
       "epoch": 4.412872672478743,
-      "grad_norm": 0.692973256111145,
+      "grad_norm": 0.6765034198760986,
       "learning_rate": 0.0003356858097187803,
-      "loss": 3.3813,
+      "loss": 3.3935,
       "step": 41000
     },
     {
       "epoch": 4.412872672478743,
-      "eval_accuracy": 0.37769666960427856,
-      "eval_loss": 3.4349679946899414,
-      "eval_runtime": 181.1247,
-      "eval_samples_per_second": 99.44,
-      "eval_steps_per_second": 6.217,
+      "eval_accuracy": 0.3784238833824209,
+      "eval_loss": 3.4338836669921875,
+      "eval_runtime": 181.6523,
+      "eval_samples_per_second": 99.151,
+      "eval_steps_per_second": 6.199,
       "step": 41000
     },
     {
       "epoch": 4.418254224518351,
-      "grad_norm": 0.6343230605125427,
+      "grad_norm": 0.6076163649559021,
       "learning_rate": 0.00033536256868871886,
-      "loss": 3.3868,
+      "loss": 3.3764,
       "step": 41050
     },
     {
       "epoch": 4.423635776557959,
-      "grad_norm": 0.6828956604003906,
+      "grad_norm": 0.6354140043258667,
       "learning_rate": 0.00033503932765865745,
-      "loss": 3.3779,
+      "loss": 3.3881,
       "step": 41100
     },
     {
       "epoch": 4.429017328597568,
-      "grad_norm": 0.6604346632957458,
+      "grad_norm": 0.6258163452148438,
       "learning_rate": 0.000334716086628596,
-      "loss": 3.3862,
+      "loss": 3.3808,
       "step": 41150
     },
     {
       "epoch": 4.4343988806371755,
-      "grad_norm": 0.7162014842033386,
+      "grad_norm": 0.6244299411773682,
       "learning_rate": 0.0003343928455985346,
-      "loss": 3.4053,
+      "loss": 3.388,
       "step": 41200
     },
     {
       "epoch": 4.439780432676784,
-      "grad_norm": 0.7162100672721863,
+      "grad_norm": 0.6745234727859497,
       "learning_rate": 0.00033406960456847324,
-      "loss": 3.397,
+      "loss": 3.3919,
       "step": 41250
     },
     {
       "epoch": 4.445161984716393,
-      "grad_norm": 0.6890962719917297,
+      "grad_norm": 0.6292052865028381,
       "learning_rate": 0.0003337463635384118,
-      "loss": 3.4001,
+      "loss": 3.3986,
       "step": 41300
     },
     {
       "epoch": 4.450543536756,
-      "grad_norm": 0.714141309261322,
+      "grad_norm": 0.6365064978599548,
       "learning_rate": 0.00033342312250835037,
-      "loss": 3.3843,
+      "loss": 3.3859,
       "step": 41350
     },
     {
       "epoch": 4.455925088795609,
-      "grad_norm": 0.6847618222236633,
+      "grad_norm": 0.619064450263977,
       "learning_rate": 0.00033309988147828896,
-      "loss": 3.4102,
+      "loss": 3.3833,
       "step": 41400
     },
     {
       "epoch": 4.461306640835216,
-      "grad_norm": 0.6703453063964844,
+      "grad_norm": 0.6593164801597595,
       "learning_rate": 0.0003327766404482275,
-      "loss": 3.4016,
+      "loss": 3.415,
       "step": 41450
     },
     {
       "epoch": 4.466688192874825,
-      "grad_norm": 0.6432178616523743,
+      "grad_norm": 0.6667826175689697,
       "learning_rate": 0.0003324533994181661,
-      "loss": 3.3857,
+      "loss": 3.3872,
       "step": 41500
     },
     {
       "epoch": 4.4720697449144335,
-      "grad_norm": 0.6851868033409119,
+      "grad_norm": 0.7157216668128967,
       "learning_rate": 0.00033213015838810475,
-      "loss": 3.4059,
+      "loss": 3.3954,
       "step": 41550
     },
     {
       "epoch": 4.477451296954041,
-      "grad_norm": 0.6780423521995544,
+      "grad_norm": 0.6244058609008789,
       "learning_rate": 0.0003318069173580433,
-      "loss": 3.3957,
+      "loss": 3.3731,
       "step": 41600
     },
     {
       "epoch": 4.48283284899365,
-      "grad_norm": 0.6518639922142029,
+      "grad_norm": 0.6343091726303101,
       "learning_rate": 0.0003314836763279819,
-      "loss": 3.3808,
+      "loss": 3.3754,
       "step": 41650
     },
     {
       "epoch": 4.488214401033258,
-      "grad_norm": 0.630031406879425,
+      "grad_norm": 0.6154260039329529,
       "learning_rate": 0.0003311604352979204,
-      "loss": 3.3863,
+      "loss": 3.3646,
       "step": 41700
     },
     {
       "epoch": 4.493595953072866,
-      "grad_norm": 0.6582158207893372,
+      "grad_norm": 0.6237065196037292,
       "learning_rate": 0.000330837194267859,
-      "loss": 3.3722,
+      "loss": 3.3872,
       "step": 41750
     },
     {
       "epoch": 4.4989775051124745,
-      "grad_norm": 0.7116129994392395,
+      "grad_norm": 0.6346547603607178,
       "learning_rate": 0.0003305139532377976,
-      "loss": 3.389,
+      "loss": 3.4098,
       "step": 41800
     },
     {
       "epoch": 4.504359057152083,
-      "grad_norm": 0.6760125756263733,
+      "grad_norm": 0.6222251057624817,
       "learning_rate": 0.0003301907122077362,
-      "loss": 3.396,
+      "loss": 3.3723,
       "step": 41850
     },
     {
       "epoch": 4.509740609191691,
-      "grad_norm": 0.6478492617607117,
+      "grad_norm": 0.665420413017273,
       "learning_rate": 0.0003298674711776748,
-      "loss": 3.401,
+      "loss": 3.4172,
       "step": 41900
     },
     {
       "epoch": 4.515122161231299,
-      "grad_norm": 0.6583666205406189,
+      "grad_norm": 0.6179385185241699,
       "learning_rate": 0.0003295442301476134,
-      "loss": 3.3866,
+      "loss": 3.3913,
       "step": 41950
     },
     {
       "epoch": 4.520503713270907,
-      "grad_norm": 0.6689998507499695,
+      "grad_norm": 0.6129652261734009,
       "learning_rate": 0.00032922098911755194,
-      "loss": 3.3924,
+      "loss": 3.3866,
       "step": 42000
     },
     {
       "epoch": 4.520503713270907,
-      "eval_accuracy": 0.37855687451710585,
-      "eval_loss": 3.431107521057129,
-      "eval_runtime": 181.9242,
-      "eval_samples_per_second": 99.003,
-      "eval_steps_per_second": 6.189,
+      "eval_accuracy": 0.3788473036862339,
+      "eval_loss": 3.429064989089966,
+      "eval_runtime": 181.7062,
+      "eval_samples_per_second": 99.122,
+      "eval_steps_per_second": 6.197,
       "step": 42000
     },
     {
       "epoch": 4.5258852653105155,
-      "grad_norm": 0.655572235584259,
+      "grad_norm": 0.5837913155555725,
       "learning_rate": 0.00032889774808749053,
-      "loss": 3.407,
+      "loss": 3.3868,
       "step": 42050
     },
     {
       "epoch": 4.531266817350124,
-      "grad_norm": 0.6405171155929565,
+      "grad_norm": 0.6249428391456604,
       "learning_rate": 0.0003285745070574292,
-      "loss": 3.4029,
+      "loss": 3.3927,
       "step": 42100
     },
     {
       "epoch": 4.536648369389732,
-      "grad_norm": 0.6641229391098022,
+      "grad_norm": 0.638615608215332,
       "learning_rate": 0.0003282512660273677,
-      "loss": 3.3999,
+      "loss": 3.3798,
       "step": 42150
     },
     {
       "epoch": 4.54202992142934,
-      "grad_norm": 0.6896253824234009,
+      "grad_norm": 0.6654457449913025,
       "learning_rate": 0.0003279280249973063,
-      "loss": 3.4085,
+      "loss": 3.4005,
       "step": 42200
     },
     {
       "epoch": 4.547411473468949,
-      "grad_norm": 0.6799988746643066,
+      "grad_norm": 0.6309723854064941,
       "learning_rate": 0.00032760478396724486,
-      "loss": 3.4027,
+      "loss": 3.3766,
       "step": 42250
     },
     {
       "epoch": 4.5527930255085565,
-      "grad_norm": 0.6822940111160278,
+      "grad_norm": 0.6769714951515198,
       "learning_rate": 0.00032728154293718345,
-      "loss": 3.3753,
+      "loss": 3.3672,
       "step": 42300
     },
     {
       "epoch": 4.558174577548165,
-      "grad_norm": 0.656019389629364,
+      "grad_norm": 0.6504793167114258,
       "learning_rate": 0.00032695830190712205,
-      "loss": 3.3807,
+      "loss": 3.3841,
       "step": 42350
     },
     {
       "epoch": 4.563556129587774,
-      "grad_norm": 0.650692880153656,
+      "grad_norm": 0.6260042190551758,
       "learning_rate": 0.00032663506087706064,
-      "loss": 3.4062,
+      "loss": 3.4012,
       "step": 42400
     },
     {
       "epoch": 4.568937681627381,
-      "grad_norm": 0.6641072630882263,
+      "grad_norm": 0.6566450595855713,
       "learning_rate": 0.00032631181984699924,
-      "loss": 3.4034,
+      "loss": 3.3801,
       "step": 42450
     },
     {
       "epoch": 4.57431923366699,
-      "grad_norm": 0.6998100280761719,
+      "grad_norm": 0.6926515102386475,
       "learning_rate": 0.00032598857881693783,
-      "loss": 3.394,
+      "loss": 3.4049,
       "step": 42500
     },
     {
       "epoch": 4.579700785706597,
-      "grad_norm": 0.7412688732147217,
+      "grad_norm": 0.7046941518783569,
       "learning_rate": 0.00032566533778687637,
-      "loss": 3.3865,
+      "loss": 3.3702,
       "step": 42550
     },
     {
       "epoch": 4.585082337746206,
-      "grad_norm": 0.6878949999809265,
-      "learning_rate": 0.00032534209675681497,
-      "loss": 3.3875,
+      "grad_norm": 0.7117016315460205,
+      "learning_rate": 0.0003253485615774162,
+      "loss": 3.3922,
       "step": 42600
     },
     {
       "epoch": 4.5904638897858145,
-      "grad_norm": 0.718602180480957,
-      "learning_rate": 0.0003250188557267535,
-      "loss": 3.4083,
+      "grad_norm": 0.7032127380371094,
+      "learning_rate": 0.00032502532054735477,
+      "loss": 3.4032,
       "step": 42650
     },
     {
       "epoch": 4.595845441825422,
-      "grad_norm": 0.6992116570472717,
-      "learning_rate": 0.00032469561469669215,
-      "loss": 3.3877,
+      "grad_norm": 0.6811186671257019,
+      "learning_rate": 0.00032470207951729337,
+      "loss": 3.4026,
       "step": 42700
     },
     {
       "epoch": 4.601226993865031,
-      "grad_norm": 0.6840443015098572,
-      "learning_rate": 0.00032437237366663075,
-      "loss": 3.3814,
+      "grad_norm": 0.6895545721054077,
+      "learning_rate": 0.0003243788384872319,
+      "loss": 3.3894,
       "step": 42750
     },
     {
       "epoch": 4.606608545904638,
-      "grad_norm": 0.6726332902908325,
-      "learning_rate": 0.0003240491326365693,
-      "loss": 3.4107,
+      "grad_norm": 0.6768222451210022,
+      "learning_rate": 0.00032405559745717056,
+      "loss": 3.4082,
       "step": 42800
     },
     {
       "epoch": 4.611990097944247,
-      "grad_norm": 0.7318175435066223,
-      "learning_rate": 0.0003237258916065079,
-      "loss": 3.3882,
+      "grad_norm": 0.6644182205200195,
+      "learning_rate": 0.00032373235642710915,
+      "loss": 3.4108,
       "step": 42850
     },
     {
       "epoch": 4.6173716499838555,
-      "grad_norm": 0.6592293977737427,
-      "learning_rate": 0.0003234026505764465,
-      "loss": 3.4084,
+      "grad_norm": 0.6393153071403503,
+      "learning_rate": 0.0003234091153970477,
+      "loss": 3.3813,
       "step": 42900
     },
     {
       "epoch": 4.622753202023463,
-      "grad_norm": 0.70222407579422,
-      "learning_rate": 0.000323079409546385,
-      "loss": 3.3907,
+      "grad_norm": 0.651145339012146,
+      "learning_rate": 0.0003230858743669863,
+      "loss": 3.3914,
       "step": 42950
     },
     {
       "epoch": 4.628134754063072,
-      "grad_norm": 0.6887460350990295,
+      "grad_norm": 0.6680070161819458,
       "learning_rate": 0.0003227626333369248,
-      "loss": 3.3912,
+      "loss": 3.3818,
       "step": 43000
     },
     {
       "epoch": 4.628134754063072,
-      "eval_accuracy": 0.3786048990935198,
-      "eval_loss": 3.426353693008423,
-      "eval_runtime": 183.3187,
-      "eval_samples_per_second": 98.25,
-      "eval_steps_per_second": 6.142,
+      "eval_accuracy": 0.3787929772423267,
+      "eval_loss": 3.4235572814941406,
+      "eval_runtime": 181.274,
+      "eval_samples_per_second": 99.358,
+      "eval_steps_per_second": 6.212,
       "step": 43000
     },
     {
       "epoch": 4.63351630610268,
-      "grad_norm": 0.6995670795440674,
+      "grad_norm": 0.6707707643508911,
       "learning_rate": 0.0003224393923068635,
-      "loss": 3.3847,
+      "loss": 3.3785,
       "step": 43050
     },
     {
       "epoch": 4.638897858142288,
-      "grad_norm": 0.7121273875236511,
+      "grad_norm": 0.6937111020088196,
       "learning_rate": 0.00032211615127680207,
-      "loss": 3.3972,
+      "loss": 3.3941,
       "step": 43100
     },
     {
       "epoch": 4.6442794101818965,
-      "grad_norm": 0.6917558908462524,
+      "grad_norm": 0.6383497714996338,
       "learning_rate": 0.0003217929102467406,
-      "loss": 3.4062,
+      "loss": 3.3954,
       "step": 43150
     },
     {
       "epoch": 4.649660962221505,
-      "grad_norm": 0.7152304649353027,
+      "grad_norm": 0.6747254133224487,
       "learning_rate": 0.0003214696692166792,
-      "loss": 3.3848,
+      "loss": 3.3839,
       "step": 43200
     },
     {
       "epoch": 4.655042514261113,
-      "grad_norm": 0.6601977944374084,
+      "grad_norm": 0.6464184522628784,
       "learning_rate": 0.0003211464281866178,
-      "loss": 3.3972,
+      "loss": 3.3906,
       "step": 43250
     },
     {
       "epoch": 4.660424066300721,
-      "grad_norm": 0.6530590057373047,
+      "grad_norm": 0.6488027572631836,
       "learning_rate": 0.00032082318715655634,
-      "loss": 3.388,
+      "loss": 3.3716,
       "step": 43300
     },
     {
       "epoch": 4.665805618340329,
-      "grad_norm": 0.6723693013191223,
+      "grad_norm": 0.6586099863052368,
       "learning_rate": 0.000320499946126495,
-      "loss": 3.3952,
+      "loss": 3.3756,
       "step": 43350
     },
     {
       "epoch": 4.6711871703799375,
-      "grad_norm": 0.6972584128379822,
+      "grad_norm": 0.6510335803031921,
       "learning_rate": 0.0003201767050964336,
-      "loss": 3.4,
+      "loss": 3.3914,
       "step": 43400
     },
     {
       "epoch": 4.676568722419546,
-      "grad_norm": 0.692314088344574,
+      "grad_norm": 0.6524762511253357,
       "learning_rate": 0.0003198534640663721,
-      "loss": 3.3883,
+      "loss": 3.387,
       "step": 43450
     },
     {
       "epoch": 4.681950274459154,
-      "grad_norm": 0.6952795386314392,
+      "grad_norm": 0.6539650559425354,
       "learning_rate": 0.0003195302230363107,
-      "loss": 3.3966,
+      "loss": 3.3995,
       "step": 43500
     },
     {
       "epoch": 4.687331826498762,
-      "grad_norm": 0.6832876205444336,
+      "grad_norm": 0.6376681327819824,
       "learning_rate": 0.00031920698200624926,
-      "loss": 3.3708,
+      "loss": 3.374,
       "step": 43550
     },
     {
       "epoch": 4.692713378538371,
-      "grad_norm": 0.684529185295105,
+      "grad_norm": 0.6648434400558472,
       "learning_rate": 0.00031888374097618785,
-      "loss": 3.3997,
+      "loss": 3.3923,
       "step": 43600
     },
     {
       "epoch": 4.6980949305779784,
-      "grad_norm": 0.6592495441436768,
+      "grad_norm": 0.6919891834259033,
       "learning_rate": 0.0003185604999461265,
-      "loss": 3.3893,
+      "loss": 3.3818,
       "step": 43650
     },
     {
       "epoch": 4.703476482617587,
-      "grad_norm": 0.7246899604797363,
+      "grad_norm": 0.6662169694900513,
       "learning_rate": 0.00031823725891606504,
-      "loss": 3.3858,
+      "loss": 3.3936,
       "step": 43700
     },
     {
       "epoch": 4.7088580346571955,
-      "grad_norm": 0.7060039043426514,
+      "grad_norm": 0.6808415651321411,
       "learning_rate": 0.00031791401788600364,
-      "loss": 3.373,
+      "loss": 3.3752,
       "step": 43750
     },
     {
       "epoch": 4.714239586696803,
-      "grad_norm": 0.727345883846283,
+      "grad_norm": 0.6524971127510071,
       "learning_rate": 0.00031759077685594223,
       "loss": 3.3876,
       "step": 43800
     },
     {
       "epoch": 4.719621138736412,
-      "grad_norm": 0.6933861970901489,
+      "grad_norm": 0.6357795000076294,
       "learning_rate": 0.00031726753582588077,
-      "loss": 3.4002,
+      "loss": 3.3865,
       "step": 43850
     },
     {
       "epoch": 4.725002690776019,
-      "grad_norm": 0.7261818647384644,
+      "grad_norm": 0.6607389450073242,
       "learning_rate": 0.0003169442947958194,
-      "loss": 3.397,
+      "loss": 3.3936,
       "step": 43900
     },
     {
       "epoch": 4.730384242815628,
-      "grad_norm": 0.6434661149978638,
+      "grad_norm": 0.7226917147636414,
       "learning_rate": 0.000316621053765758,
-      "loss": 3.4044,
+      "loss": 3.3948,
       "step": 43950
     },
     {
       "epoch": 4.7357657948552365,
-      "grad_norm": 0.6868217587471008,
+      "grad_norm": 0.645797610282898,
       "learning_rate": 0.00031629781273569656,
-      "loss": 3.4047,
+      "loss": 3.4125,
       "step": 44000
     },
     {
       "epoch": 4.7357657948552365,
-      "eval_accuracy": 0.3793180966491341,
-      "eval_loss": 3.421250581741333,
-      "eval_runtime": 187.5428,
-      "eval_samples_per_second": 96.037,
-      "eval_steps_per_second": 6.004,
+      "eval_accuracy": 0.37952421117731816,
+      "eval_loss": 3.4157555103302,
+      "eval_runtime": 181.5161,
+      "eval_samples_per_second": 99.225,
+      "eval_steps_per_second": 6.203,
       "step": 44000
     },
     {
       "epoch": 4.741147346894844,
-      "grad_norm": 0.6731848120689392,
+      "grad_norm": 0.6595436930656433,
       "learning_rate": 0.00031597457170563515,
-      "loss": 3.3793,
+      "loss": 3.3895,
       "step": 44050
     },
     {
       "epoch": 4.746528898934453,
-      "grad_norm": 0.6794449687004089,
+      "grad_norm": 0.6886684894561768,
       "learning_rate": 0.0003156513306755737,
-      "loss": 3.39,
+      "loss": 3.3911,
       "step": 44100
     },
     {
       "epoch": 4.751910450974061,
-      "grad_norm": 0.7639029622077942,
+      "grad_norm": 0.7380845546722412,
       "learning_rate": 0.0003153280896455123,
-      "loss": 3.402,
+      "loss": 3.3957,
       "step": 44150
     },
     {
       "epoch": 4.757292003013669,
-      "grad_norm": 0.686454176902771,
+      "grad_norm": 0.6567789912223816,
       "learning_rate": 0.00031500484861545094,
-      "loss": 3.3893,
+      "loss": 3.386,
       "step": 44200
     },
     {
       "epoch": 4.7626735550532775,
-      "grad_norm": 0.7179557085037231,
+      "grad_norm": 0.6533080339431763,
       "learning_rate": 0.0003146816075853895,
-      "loss": 3.3874,
+      "loss": 3.3913,
       "step": 44250
     },
     {
       "epoch": 4.768055107092886,
-      "grad_norm": 0.6821876168251038,
+      "grad_norm": 0.6387834548950195,
       "learning_rate": 0.00031435836655532807,
-      "loss": 3.3928,
+      "loss": 3.389,
       "step": 44300
     },
     {
       "epoch": 4.773436659132494,
-      "grad_norm": 0.6775598526000977,
+      "grad_norm": 0.6194447875022888,
       "learning_rate": 0.00031403512552526667,
-      "loss": 3.381,
+      "loss": 3.4026,
       "step": 44350
     },
     {
       "epoch": 4.778818211172102,
-      "grad_norm": 0.666349470615387,
+      "grad_norm": 0.6570907235145569,
       "learning_rate": 0.0003137118844952052,
-      "loss": 3.4018,
+      "loss": 3.4048,
       "step": 44400
     },
     {
       "epoch": 4.78419976321171,
-      "grad_norm": 0.6841625571250916,
+      "grad_norm": 0.6597467660903931,
       "learning_rate": 0.0003133886434651438,
-      "loss": 3.3952,
+      "loss": 3.3791,
       "step": 44450
     },
     {
       "epoch": 4.7895813152513185,
-      "grad_norm": 0.6649987697601318,
+      "grad_norm": 0.6301759481430054,
       "learning_rate": 0.00031306540243508245,
-      "loss": 3.4116,
+      "loss": 3.3883,
       "step": 44500
     },
     {
       "epoch": 4.794962867290927,
-      "grad_norm": 0.6441112756729126,
+      "grad_norm": 0.6611241102218628,
       "learning_rate": 0.000312742161405021,
-      "loss": 3.3951,
+      "loss": 3.3877,
       "step": 44550
     },
     {
       "epoch": 4.800344419330535,
-      "grad_norm": 0.6589279174804688,
+      "grad_norm": 0.6324141025543213,
       "learning_rate": 0.0003124189203749596,
-      "loss": 3.3902,
+      "loss": 3.3713,
       "step": 44600
     },
     {
       "epoch": 4.805725971370143,
-      "grad_norm": 0.677620530128479,
+      "grad_norm": 0.6432929635047913,
       "learning_rate": 0.0003120956793448981,
-      "loss": 3.3967,
+      "loss": 3.387,
       "step": 44650
     },
     {
       "epoch": 4.811107523409751,
-      "grad_norm": 0.7282724380493164,
+      "grad_norm": 0.6320468187332153,
       "learning_rate": 0.0003117724383148367,
-      "loss": 3.386,
+      "loss": 3.3829,
       "step": 44700
     },
     {
       "epoch": 4.8164890754493594,
-      "grad_norm": 0.6708424687385559,
+      "grad_norm": 0.6340096592903137,
       "learning_rate": 0.00031144919728477526,
-      "loss": 3.3793,
+      "loss": 3.3765,
       "step": 44750
     },
     {
       "epoch": 4.821870627488968,
-      "grad_norm": 0.678166925907135,
+      "grad_norm": 0.6637021899223328,
       "learning_rate": 0.0003111259562547139,
-      "loss": 3.4066,
+      "loss": 3.3984,
       "step": 44800
     },
     {
       "epoch": 4.827252179528576,
-      "grad_norm": 0.6643491387367249,
+      "grad_norm": 0.6757526397705078,
       "learning_rate": 0.0003108027152246525,
-      "loss": 3.3997,
+      "loss": 3.4011,
       "step": 44850
     },
     {
       "epoch": 4.832633731568184,
-      "grad_norm": 0.7268407940864563,
+      "grad_norm": 0.6414923667907715,
       "learning_rate": 0.0003104794741945911,
-      "loss": 3.3941,
+      "loss": 3.3972,
       "step": 44900
     },
     {
       "epoch": 4.838015283607793,
-      "grad_norm": 0.666113555431366,
+      "grad_norm": 0.658253014087677,
       "learning_rate": 0.00031015623316452964,
-      "loss": 3.3933,
+      "loss": 3.3915,
       "step": 44950
     },
     {
       "epoch": 4.8433968356474,
-      "grad_norm": 0.6685688495635986,
-      "learning_rate": 0.00030983299213446823,
-      "loss": 3.3776,
+      "grad_norm": 0.7013706564903259,
+      "learning_rate": 0.00030983945695506945,
+      "loss": 3.3749,
       "step": 45000
     },
     {
       "epoch": 4.8433968356474,
-      "eval_accuracy": 0.37968338765796633,
-      "eval_loss": 3.4163522720336914,
-      "eval_runtime": 202.3061,
-      "eval_samples_per_second": 89.028,
-      "eval_steps_per_second": 5.566,
+      "eval_accuracy": 0.3797631388776222,
+      "eval_loss": 3.4132089614868164,
+      "eval_runtime": 181.4084,
+      "eval_samples_per_second": 99.284,
+      "eval_steps_per_second": 6.207,
       "step": 45000
     },
     {
       "epoch": 4.848778387687009,
-      "grad_norm": 0.657361626625061,
-      "learning_rate": 0.0003095097511044069,
-      "loss": 3.4067,
+      "grad_norm": 0.6422181725502014,
+      "learning_rate": 0.00030951621592500804,
+      "loss": 3.3985,
       "step": 45050
     },
     {
       "epoch": 4.8541599397266175,
-      "grad_norm": 0.6610834002494812,
-      "learning_rate": 0.0003091865100743454,
-      "loss": 3.4013,
+      "grad_norm": 0.6657373905181885,
+      "learning_rate": 0.00030919297489494663,
+      "loss": 3.3932,
       "step": 45100
     },
     {
       "epoch": 4.859541491766225,
-      "grad_norm": 0.6313742995262146,
+      "grad_norm": 0.6616642475128174,
       "learning_rate": 0.00030886973386488523,
-      "loss": 3.4168,
+      "loss": 3.3955,
       "step": 45150
     },
     {
       "epoch": 4.864923043805834,
-      "grad_norm": 0.6884166598320007,
+      "grad_norm": 0.6808633208274841,
       "learning_rate": 0.0003085464928348238,
-      "loss": 3.3887,
+      "loss": 3.3996,
       "step": 45200
     },
     {
       "epoch": 4.870304595845441,
-      "grad_norm": 0.6939536333084106,
+      "grad_norm": 0.6713807582855225,
       "learning_rate": 0.0003082232518047624,
-      "loss": 3.3801,
+      "loss": 3.3786,
       "step": 45250
     },
     {
       "epoch": 4.87568614788505,
-      "grad_norm": 0.6692671775817871,
+      "grad_norm": 0.6750205755233765,
       "learning_rate": 0.00030790001077470096,
-      "loss": 3.3979,
+      "loss": 3.3914,
       "step": 45300
     },
     {
       "epoch": 4.8810676999246585,
-      "grad_norm": 0.6553664803504944,
+      "grad_norm": 0.6538323163986206,
       "learning_rate": 0.00030757676974463955,
-      "loss": 3.3896,
+      "loss": 3.3956,
       "step": 45350
     },
     {
       "epoch": 4.886449251964266,
-      "grad_norm": 0.6963851451873779,
+      "grad_norm": 0.6709042191505432,
       "learning_rate": 0.0003072535287145781,
-      "loss": 3.4001,
+      "loss": 3.3953,
       "step": 45400
     },
     {
       "epoch": 4.891830804003875,
-      "grad_norm": 0.6625418066978455,
+      "grad_norm": 0.6458733677864075,
       "learning_rate": 0.00030693028768451674,
-      "loss": 3.3867,
+      "loss": 3.3989,
       "step": 45450
     },
     {
       "epoch": 4.897212356043483,
-      "grad_norm": 0.6888773441314697,
+      "grad_norm": 0.6456732749938965,
       "learning_rate": 0.00030660704665445534,
-      "loss": 3.3772,
+      "loss": 3.3665,
       "step": 45500
     },
     {
       "epoch": 4.902593908083091,
-      "grad_norm": 0.657309889793396,
+      "grad_norm": 0.6412439346313477,
       "learning_rate": 0.0003062838056243939,
-      "loss": 3.3902,
+      "loss": 3.3766,
       "step": 45550
     },
     {
       "epoch": 4.9079754601226995,
-      "grad_norm": 0.6960524320602417,
+      "grad_norm": 0.6516876220703125,
       "learning_rate": 0.00030596056459433247,
-      "loss": 3.3857,
+      "loss": 3.4037,
       "step": 45600
     },
     {
       "epoch": 4.913357012162308,
-      "grad_norm": 0.6952453255653381,
+      "grad_norm": 0.6648831963539124,
       "learning_rate": 0.00030563732356427107,
-      "loss": 3.4173,
+      "loss": 3.3884,
       "step": 45650
     },
     {
       "epoch": 4.918738564201916,
-      "grad_norm": 0.6704519987106323,
+      "grad_norm": 0.6400995254516602,
       "learning_rate": 0.00030531408253420966,
-      "loss": 3.3778,
+      "loss": 3.365,
       "step": 45700
     },
     {
       "epoch": 4.924120116241524,
-      "grad_norm": 0.7266472578048706,
+      "grad_norm": 0.6892823576927185,
       "learning_rate": 0.00030499084150414826,
-      "loss": 3.3823,
+      "loss": 3.3762,
       "step": 45750
     },
     {
       "epoch": 4.929501668281132,
-      "grad_norm": 0.6816993355751038,
+      "grad_norm": 0.6849403381347656,
       "learning_rate": 0.00030466760047408685,
-      "loss": 3.3877,
+      "loss": 3.391,
       "step": 45800
     },
     {
       "epoch": 4.9348832203207404,
-      "grad_norm": 0.6936447620391846,
+      "grad_norm": 0.6666726469993591,
       "learning_rate": 0.0003043443594440254,
-      "loss": 3.3832,
+      "loss": 3.3807,
       "step": 45850
     },
     {
       "epoch": 4.940264772360349,
-      "grad_norm": 0.6859702467918396,
+      "grad_norm": 0.663345456123352,
       "learning_rate": 0.000304021118413964,
-      "loss": 3.4126,
+      "loss": 3.3941,
       "step": 45900
     },
     {
       "epoch": 4.945646324399957,
-      "grad_norm": 0.6758688688278198,
+      "grad_norm": 0.6493843197822571,
       "learning_rate": 0.0003036978773839025,
-      "loss": 3.3839,
+      "loss": 3.3903,
       "step": 45950
     },
     {
       "epoch": 4.951027876439565,
-      "grad_norm": 0.6840824484825134,
+      "grad_norm": 0.7202813029289246,
       "learning_rate": 0.0003033746363538412,
-      "loss": 3.3902,
+      "loss": 3.3877,
       "step": 46000
     },
     {
       "epoch": 4.951027876439565,
-      "eval_accuracy": 0.3802585960460562,
-      "eval_loss": 3.4134035110473633,
-      "eval_runtime": 191.2911,
-      "eval_samples_per_second": 94.155,
-      "eval_steps_per_second": 5.886,
+      "eval_accuracy": 0.380573254809167,
+      "eval_loss": 3.4068665504455566,
+      "eval_runtime": 181.503,
+      "eval_samples_per_second": 99.233,
+      "eval_steps_per_second": 6.204,
       "step": 46000
     },
     {
       "epoch": 4.956409428479174,
-      "grad_norm": 0.640100359916687,
+      "grad_norm": 0.6572828888893127,
       "learning_rate": 0.00030305139532377977,
-      "loss": 3.4115,
+      "loss": 3.3793,
       "step": 46050
     },
     {
       "epoch": 4.961790980518781,
-      "grad_norm": 0.6694591641426086,
+      "grad_norm": 0.7318044900894165,
       "learning_rate": 0.0003027281542937183,
-      "loss": 3.4089,
+      "loss": 3.3941,
       "step": 46100
     },
     {
       "epoch": 4.96717253255839,
-      "grad_norm": 0.6837192177772522,
+      "grad_norm": 0.73976731300354,
       "learning_rate": 0.0003024049132636569,
-      "loss": 3.4014,
+      "loss": 3.4062,
       "step": 46150
     },
     {
       "epoch": 4.9725540845979985,
-      "grad_norm": 0.6926203966140747,
+      "grad_norm": 0.6453207731246948,
       "learning_rate": 0.0003020816722335955,
-      "loss": 3.3931,
+      "loss": 3.3783,
       "step": 46200
     },
     {
       "epoch": 4.977935636637606,
-      "grad_norm": 0.672174870967865,
+      "grad_norm": 0.664265513420105,
       "learning_rate": 0.00030175843120353404,
-      "loss": 3.3686,
+      "loss": 3.3797,
       "step": 46250
     },
     {
       "epoch": 4.983317188677215,
-      "grad_norm": 0.6593598127365112,
+      "grad_norm": 0.6867574453353882,
       "learning_rate": 0.0003014351901734727,
-      "loss": 3.4056,
+      "loss": 3.3972,
       "step": 46300
     },
     {
       "epoch": 4.988698740716822,
-      "grad_norm": 0.6591873168945312,
+      "grad_norm": 0.6498213410377502,
       "learning_rate": 0.0003011119491434113,
-      "loss": 3.4209,
+      "loss": 3.399,
       "step": 46350
     },
     {
       "epoch": 4.994080292756431,
-      "grad_norm": 0.6413924694061279,
+      "grad_norm": 0.7483736276626587,
       "learning_rate": 0.0003007887081133498,
-      "loss": 3.3968,
+      "loss": 3.3893,
       "step": 46400
     },
     {
       "epoch": 4.9994618447960395,
-      "grad_norm": 0.7270100116729736,
+      "grad_norm": 0.6604465246200562,
       "learning_rate": 0.0003004654670832884,
-      "loss": 3.4009,
+      "loss": 3.4003,
       "step": 46450
     },
     {
       "epoch": 5.004843396835647,
-      "grad_norm": 0.6969584226608276,
+      "grad_norm": 0.6823358535766602,
       "learning_rate": 0.00030014222605322696,
-      "loss": 3.3214,
+      "loss": 3.3102,
       "step": 46500
     },
     {
       "epoch": 5.010224948875256,
-      "grad_norm": 0.7092729806900024,
+      "grad_norm": 0.6361904740333557,
       "learning_rate": 0.00029981898502316555,
-      "loss": 3.3004,
+      "loss": 3.2982,
       "step": 46550
     },
     {
       "epoch": 5.015606500914864,
-      "grad_norm": 0.7010934352874756,
+      "grad_norm": 0.6895138621330261,
       "learning_rate": 0.00029949574399310415,
-      "loss": 3.3056,
+      "loss": 3.2978,
       "step": 46600
     },
     {
       "epoch": 5.020988052954472,
-      "grad_norm": 0.6872923970222473,
+      "grad_norm": 0.647340714931488,
       "learning_rate": 0.00029917250296304274,
-      "loss": 3.3137,
+      "loss": 3.3033,
       "step": 46650
     },
     {
       "epoch": 5.0263696049940805,
-      "grad_norm": 0.6743913292884827,
+      "grad_norm": 0.7126519083976746,
       "learning_rate": 0.00029884926193298134,
-      "loss": 3.3004,
+      "loss": 3.3111,
       "step": 46700
     },
     {
       "epoch": 5.031751157033688,
-      "grad_norm": 0.7134643197059631,
+      "grad_norm": 0.7115910649299622,
       "learning_rate": 0.0002985260209029199,
-      "loss": 3.2892,
+      "loss": 3.2931,
       "step": 46750
     },
     {
       "epoch": 5.037132709073297,
-      "grad_norm": 0.6986505389213562,
+      "grad_norm": 0.6613615155220032,
       "learning_rate": 0.00029820277987285853,
-      "loss": 3.2902,
+      "loss": 3.2897,
       "step": 46800
     },
     {
       "epoch": 5.042514261112905,
-      "grad_norm": 0.6677493453025818,
+      "grad_norm": 0.6790776252746582,
       "learning_rate": 0.00029787953884279707,
-      "loss": 3.3036,
+      "loss": 3.2852,
       "step": 46850
     },
     {
       "epoch": 5.047895813152513,
-      "grad_norm": 0.6945018172264099,
+      "grad_norm": 0.7005345225334167,
       "learning_rate": 0.00029755629781273566,
-      "loss": 3.3008,
+      "loss": 3.3114,
       "step": 46900
     },
     {
       "epoch": 5.0532773651921215,
-      "grad_norm": 0.7299492955207825,
+      "grad_norm": 0.6714920401573181,
       "learning_rate": 0.00029723305678267426,
-      "loss": 3.3021,
+      "loss": 3.3099,
       "step": 46950
     },
     {
       "epoch": 5.05865891723173,
-      "grad_norm": 0.6931959390640259,
+      "grad_norm": 0.6989243030548096,
       "learning_rate": 0.00029690981575261285,
-      "loss": 3.2993,
+      "loss": 3.2956,
       "step": 47000
     },
     {
       "epoch": 5.05865891723173,
-      "eval_accuracy": 0.38065050701240305,
-      "eval_loss": 3.4127960205078125,
-      "eval_runtime": 201.6655,
-      "eval_samples_per_second": 89.311,
-      "eval_steps_per_second": 5.584,
+      "eval_accuracy": 0.38082902370708227,
+      "eval_loss": 3.410607099533081,
+      "eval_runtime": 181.2751,
+      "eval_samples_per_second": 99.357,
+      "eval_steps_per_second": 6.212,
       "step": 47000
     },
     {
       "epoch": 5.064040469271338,
-      "grad_norm": 0.7374845147132874,
+      "grad_norm": 0.6836389303207397,
       "learning_rate": 0.0002965865747225514,
-      "loss": 3.3001,
+      "loss": 3.2981,
       "step": 47050
     },
     {
       "epoch": 5.069422021310946,
-      "grad_norm": 0.6965045928955078,
+      "grad_norm": 0.692420244216919,
       "learning_rate": 0.00029626333369249,
-      "loss": 3.3073,
+      "loss": 3.2967,
       "step": 47100
     },
     {
       "epoch": 5.074803573350554,
-      "grad_norm": 0.6908303499221802,
+      "grad_norm": 0.6568200588226318,
       "learning_rate": 0.0002959400926624286,
-      "loss": 3.3072,
+      "loss": 3.2946,
       "step": 47150
     },
     {
       "epoch": 5.080185125390162,
-      "grad_norm": 0.6920403242111206,
+      "grad_norm": 0.6334496736526489,
       "learning_rate": 0.0002956168516323672,
-      "loss": 3.318,
+      "loss": 3.2959,
       "step": 47200
     },
     {
       "epoch": 5.085566677429771,
-      "grad_norm": 0.7072156071662903,
+      "grad_norm": 0.6666896939277649,
       "learning_rate": 0.00029529361060230577,
-      "loss": 3.3103,
+      "loss": 3.3038,
       "step": 47250
     },
     {
       "epoch": 5.090948229469379,
-      "grad_norm": 0.691167414188385,
+      "grad_norm": 0.6709624528884888,
       "learning_rate": 0.0002949703695722443,
-      "loss": 3.3072,
+      "loss": 3.3149,
       "step": 47300
     },
     {
       "epoch": 5.096329781508987,
-      "grad_norm": 0.6893109679222107,
+      "grad_norm": 0.6761017441749573,
       "learning_rate": 0.00029464712854218296,
-      "loss": 3.3113,
+      "loss": 3.2953,
       "step": 47350
     },
     {
       "epoch": 5.101711333548596,
-      "grad_norm": 0.6975725889205933,
+      "grad_norm": 0.6951745748519897,
       "learning_rate": 0.0002943238875121215,
-      "loss": 3.3042,
+      "loss": 3.3114,
       "step": 47400
     },
     {
       "epoch": 5.107092885588203,
-      "grad_norm": 0.7136406302452087,
+      "grad_norm": 0.6841645240783691,
       "learning_rate": 0.0002940006464820601,
-      "loss": 3.3148,
+      "loss": 3.3147,
       "step": 47450
     },
     {
       "epoch": 5.112474437627812,
-      "grad_norm": 0.7137263417243958,
+      "grad_norm": 0.7073809504508972,
       "learning_rate": 0.0002936774054519987,
-      "loss": 3.3261,
+      "loss": 3.323,
       "step": 47500
     },
     {
       "epoch": 5.1178559896674205,
-      "grad_norm": 0.6680945754051208,
+      "grad_norm": 0.673910915851593,
       "learning_rate": 0.0002933541644219373,
-      "loss": 3.337,
+      "loss": 3.3265,
       "step": 47550
     },
     {
       "epoch": 5.123237541707028,
-      "grad_norm": 0.8087601661682129,
+      "grad_norm": 0.7163915038108826,
       "learning_rate": 0.0002930309233918758,
-      "loss": 3.3087,
+      "loss": 3.309,
       "step": 47600
     },
     {
       "epoch": 5.128619093746637,
-      "grad_norm": 0.6731941103935242,
+      "grad_norm": 0.6799605488777161,
       "learning_rate": 0.0002927076823618144,
-      "loss": 3.3181,
+      "loss": 3.3028,
       "step": 47650
     },
     {
       "epoch": 5.134000645786244,
-      "grad_norm": 0.7058475017547607,
+      "grad_norm": 0.7438784837722778,
       "learning_rate": 0.000292384441331753,
-      "loss": 3.3101,
+      "loss": 3.3019,
       "step": 47700
     },
     {
       "epoch": 5.139382197825853,
-      "grad_norm": 0.7096105813980103,
+      "grad_norm": 0.6754348874092102,
       "learning_rate": 0.0002920612003016916,
-      "loss": 3.331,
+      "loss": 3.3178,
       "step": 47750
     },
     {
       "epoch": 5.1447637498654615,
-      "grad_norm": 0.7284489274024963,
+      "grad_norm": 0.6973186731338501,
       "learning_rate": 0.0002917379592716302,
-      "loss": 3.3443,
+      "loss": 3.336,
       "step": 47800
     },
     {
       "epoch": 5.150145301905069,
-      "grad_norm": 0.6959323883056641,
+      "grad_norm": 0.654205322265625,
       "learning_rate": 0.00029141471824156874,
-      "loss": 3.3013,
+      "loss": 3.2946,
       "step": 47850
     },
     {
       "epoch": 5.155526853944678,
-      "grad_norm": 0.7227029204368591,
+      "grad_norm": 0.7202890515327454,
       "learning_rate": 0.00029109147721150734,
-      "loss": 3.3017,
+      "loss": 3.295,
       "step": 47900
     },
     {
       "epoch": 5.160908405984286,
-      "grad_norm": 0.6779053807258606,
+      "grad_norm": 0.6722302436828613,
       "learning_rate": 0.00029076823618144593,
-      "loss": 3.3166,
+      "loss": 3.3236,
       "step": 47950
     },
     {
       "epoch": 5.166289958023894,
-      "grad_norm": 0.7135473489761353,
+      "grad_norm": 0.7714488506317139,
       "learning_rate": 0.00029044499515138453,
-      "loss": 3.3255,
+      "loss": 3.3253,
       "step": 48000
     },
     {
       "epoch": 5.166289958023894,
-      "eval_accuracy": 0.3803929996682827,
-      "eval_loss": 3.412876844406128,
-      "eval_runtime": 192.717,
-      "eval_samples_per_second": 93.458,
-      "eval_steps_per_second": 5.843,
+      "eval_accuracy": 0.38084999371443046,
+      "eval_loss": 3.409940719604492,
+      "eval_runtime": 181.4501,
+      "eval_samples_per_second": 99.261,
+      "eval_steps_per_second": 6.206,
       "step": 48000
     },
     {
       "epoch": 5.1716715100635025,
-      "grad_norm": 0.6599064469337463,
+      "grad_norm": 0.6736670136451721,
       "learning_rate": 0.0002901217541213231,
-      "loss": 3.3175,
+      "loss": 3.3197,
       "step": 48050
     },
     {
       "epoch": 5.17705306210311,
-      "grad_norm": 0.7011419534683228,
+      "grad_norm": 0.6848095655441284,
       "learning_rate": 0.0002897985130912617,
-      "loss": 3.3167,
+      "loss": 3.3089,
       "step": 48100
     },
     {
       "epoch": 5.182434614142719,
-      "grad_norm": 0.7181657552719116,
-      "learning_rate": 0.00028947527206120026,
-      "loss": 3.3397,
+      "grad_norm": 0.7036287188529968,
+      "learning_rate": 0.0002894817368818015,
+      "loss": 3.3168,
       "step": 48150
     },
     {
       "epoch": 5.187816166182327,
-      "grad_norm": 0.6871926784515381,
+      "grad_norm": 0.7322375774383545,
       "learning_rate": 0.00028915849585174006,
-      "loss": 3.3162,
+      "loss": 3.3153,
       "step": 48200
     },
     {
       "epoch": 5.193197718221935,
-      "grad_norm": 0.7430890202522278,
+      "grad_norm": 0.7263621091842651,
       "learning_rate": 0.00028883525482167866,
-      "loss": 3.3273,
+      "loss": 3.3129,
       "step": 48250
     },
     {
       "epoch": 5.198579270261543,
-      "grad_norm": 0.7614420056343079,
+      "grad_norm": 0.67429119348526,
       "learning_rate": 0.00028851201379161725,
-      "loss": 3.3204,
+      "loss": 3.322,
       "step": 48300
     },
     {
       "epoch": 5.203960822301152,
-      "grad_norm": 0.6756367683410645,
+      "grad_norm": 0.668222188949585,
       "learning_rate": 0.00028818877276155585,
-      "loss": 3.327,
+      "loss": 3.3124,
       "step": 48350
     },
     {
       "epoch": 5.20934237434076,
-      "grad_norm": 0.7239832282066345,
+      "grad_norm": 0.7060642242431641,
       "learning_rate": 0.0002878655317314944,
-      "loss": 3.3298,
+      "loss": 3.3264,
       "step": 48400
     },
     {
       "epoch": 5.214723926380368,
-      "grad_norm": 0.7229234576225281,
+      "grad_norm": 0.6885815262794495,
       "learning_rate": 0.00028754229070143304,
-      "loss": 3.3227,
+      "loss": 3.3362,
       "step": 48450
     },
     {
       "epoch": 5.220105478419977,
-      "grad_norm": 0.715870201587677,
+      "grad_norm": 0.7397440671920776,
       "learning_rate": 0.0002872190496713716,
-      "loss": 3.3315,
+      "loss": 3.3313,
       "step": 48500
     },
     {
       "epoch": 5.225487030459584,
-      "grad_norm": 0.714283287525177,
+      "grad_norm": 0.7179089784622192,
       "learning_rate": 0.0002868958086413102,
-      "loss": 3.3058,
+      "loss": 3.313,
       "step": 48550
     },
     {
       "epoch": 5.230868582499193,
-      "grad_norm": 0.7448010444641113,
+      "grad_norm": 0.7655637860298157,
       "learning_rate": 0.00028657256761124877,
-      "loss": 3.3418,
+      "loss": 3.3249,
       "step": 48600
     },
     {
       "epoch": 5.236250134538801,
-      "grad_norm": 0.6719698309898376,
+      "grad_norm": 0.7360113263130188,
       "learning_rate": 0.00028624932658118736,
-      "loss": 3.338,
+      "loss": 3.3241,
       "step": 48650
     },
     {
       "epoch": 5.241631686578409,
-      "grad_norm": 0.7021603584289551,
+      "grad_norm": 0.694924533367157,
       "learning_rate": 0.0002859260855511259,
-      "loss": 3.3204,
+      "loss": 3.3419,
       "step": 48700
     },
     {
       "epoch": 5.247013238618018,
-      "grad_norm": 0.7057817578315735,
+      "grad_norm": 0.7072723507881165,
       "learning_rate": 0.0002856028445210645,
-      "loss": 3.3298,
+      "loss": 3.3223,
       "step": 48750
     },
     {
       "epoch": 5.252394790657625,
-      "grad_norm": 0.7323166131973267,
+      "grad_norm": 0.71399986743927,
       "learning_rate": 0.0002852796034910031,
-      "loss": 3.3229,
+      "loss": 3.323,
       "step": 48800
     },
     {
       "epoch": 5.257776342697234,
-      "grad_norm": 0.7022436261177063,
+      "grad_norm": 0.6603875756263733,
       "learning_rate": 0.0002849563624609417,
-      "loss": 3.3345,
+      "loss": 3.3414,
       "step": 48850
     },
     {
       "epoch": 5.2631578947368425,
-      "grad_norm": 0.7284606099128723,
+      "grad_norm": 0.7345569729804993,
       "learning_rate": 0.0002846331214308803,
-      "loss": 3.3281,
+      "loss": 3.3143,
       "step": 48900
     },
     {
       "epoch": 5.26853944677645,
-      "grad_norm": 0.7403408288955688,
+      "grad_norm": 0.6999163627624512,
       "learning_rate": 0.0002843098804008188,
-      "loss": 3.3185,
+      "loss": 3.3147,
       "step": 48950
     },
     {
       "epoch": 5.273920998816059,
-      "grad_norm": 0.7374369502067566,
+      "grad_norm": 0.7195019721984863,
       "learning_rate": 0.00028398663937075747,
-      "loss": 3.3417,
+      "loss": 3.3344,
       "step": 49000
     },
     {
       "epoch": 5.273920998816059,
-      "eval_accuracy": 0.3811685639815025,
-      "eval_loss": 3.4100325107574463,
-      "eval_runtime": 186.625,
-      "eval_samples_per_second": 96.509,
-      "eval_steps_per_second": 6.033,
+      "eval_accuracy": 0.38142563671407154,
+      "eval_loss": 3.406376600265503,
+      "eval_runtime": 181.6284,
+      "eval_samples_per_second": 99.164,
+      "eval_steps_per_second": 6.199,
       "step": 49000
     },
     {
       "epoch": 5.279302550855666,
-      "grad_norm": 0.7417536377906799,
+      "grad_norm": 0.7078256011009216,
       "learning_rate": 0.000283663398340696,
-      "loss": 3.3255,
+      "loss": 3.3061,
       "step": 49050
     },
     {
       "epoch": 5.284684102895275,
-      "grad_norm": 0.7349126935005188,
+      "grad_norm": 0.728187084197998,
       "learning_rate": 0.0002833401573106346,
-      "loss": 3.3224,
+      "loss": 3.3378,
       "step": 49100
     },
     {
       "epoch": 5.2900656549348835,
-      "grad_norm": 0.6606712341308594,
+      "grad_norm": 0.6851932406425476,
       "learning_rate": 0.0002830169162805732,
-      "loss": 3.3305,
+      "loss": 3.3281,
       "step": 49150
     },
     {
       "epoch": 5.295447206974491,
-      "grad_norm": 0.6845487952232361,
+      "grad_norm": 0.6696680188179016,
       "learning_rate": 0.0002826936752505118,
-      "loss": 3.3289,
+      "loss": 3.3258,
       "step": 49200
     },
     {
       "epoch": 5.3008287590141,
-      "grad_norm": 0.7850626111030579,
+      "grad_norm": 0.727760374546051,
       "learning_rate": 0.00028237043422045034,
-      "loss": 3.338,
+      "loss": 3.326,
       "step": 49250
     },
     {
       "epoch": 5.306210311053708,
-      "grad_norm": 0.7478814721107483,
+      "grad_norm": 0.6646648049354553,
       "learning_rate": 0.00028204719319038893,
-      "loss": 3.3253,
+      "loss": 3.3273,
       "step": 49300
     },
     {
       "epoch": 5.311591863093316,
-      "grad_norm": 0.6879907250404358,
+      "grad_norm": 0.6908875703811646,
       "learning_rate": 0.0002817239521603275,
-      "loss": 3.3235,
+      "loss": 3.3141,
       "step": 49350
     },
     {
       "epoch": 5.316973415132924,
-      "grad_norm": 0.7131553292274475,
+      "grad_norm": 0.6827760934829712,
       "learning_rate": 0.0002814007111302661,
-      "loss": 3.3212,
+      "loss": 3.3215,
       "step": 49400
     },
     {
       "epoch": 5.322354967172533,
-      "grad_norm": 0.6914525032043457,
+      "grad_norm": 0.6606786847114563,
       "learning_rate": 0.0002810774701002047,
-      "loss": 3.3049,
+      "loss": 3.3199,
       "step": 49450
     },
     {
       "epoch": 5.327736519212141,
-      "grad_norm": 0.7019218802452087,
+      "grad_norm": 0.972646951675415,
       "learning_rate": 0.00028075422907014325,
-      "loss": 3.3435,
+      "loss": 3.3366,
       "step": 49500
     },
     {
       "epoch": 5.333118071251749,
-      "grad_norm": 0.7138478755950928,
+      "grad_norm": 0.6670774817466736,
       "learning_rate": 0.00028043098804008185,
-      "loss": 3.3277,
+      "loss": 3.3208,
       "step": 49550
     },
     {
       "epoch": 5.338499623291357,
-      "grad_norm": 0.7029947638511658,
+      "grad_norm": 0.7039186954498291,
       "learning_rate": 0.00028010774701002044,
-      "loss": 3.3302,
+      "loss": 3.3095,
       "step": 49600
     },
     {
       "epoch": 5.343881175330965,
-      "grad_norm": 0.7279183864593506,
+      "grad_norm": 0.6665313839912415,
       "learning_rate": 0.00027978450597995904,
-      "loss": 3.3124,
+      "loss": 3.3146,
       "step": 49650
     },
     {
       "epoch": 5.349262727370574,
-      "grad_norm": 0.7827022671699524,
+      "grad_norm": 0.722172737121582,
       "learning_rate": 0.0002794612649498976,
-      "loss": 3.324,
+      "loss": 3.3315,
       "step": 49700
     },
     {
       "epoch": 5.354644279410182,
-      "grad_norm": 0.7062620520591736,
+      "grad_norm": 0.6693257093429565,
       "learning_rate": 0.00027913802391983623,
-      "loss": 3.3445,
+      "loss": 3.3371,
       "step": 49750
     },
     {
       "epoch": 5.36002583144979,
-      "grad_norm": 0.6948559284210205,
+      "grad_norm": 0.6833609342575073,
       "learning_rate": 0.00027881478288977477,
-      "loss": 3.347,
+      "loss": 3.3287,
       "step": 49800
     },
     {
       "epoch": 5.365407383489399,
-      "grad_norm": 0.7399136424064636,
+      "grad_norm": 0.699906587600708,
       "learning_rate": 0.00027849154185971336,
-      "loss": 3.3452,
+      "loss": 3.3302,
       "step": 49850
     },
     {
       "epoch": 5.370788935529006,
-      "grad_norm": 0.7397146821022034,
+      "grad_norm": 0.6370905041694641,
       "learning_rate": 0.00027816830082965196,
-      "loss": 3.3324,
+      "loss": 3.3316,
       "step": 49900
     },
     {
       "epoch": 5.376170487568615,
-      "grad_norm": 0.7319924831390381,
+      "grad_norm": 0.6729517579078674,
       "learning_rate": 0.00027784505979959055,
-      "loss": 3.3462,
+      "loss": 3.3238,
       "step": 49950
     },
     {
       "epoch": 5.3815520396082235,
-      "grad_norm": 0.7393072843551636,
+      "grad_norm": 0.6911770105361938,
       "learning_rate": 0.00027752181876952915,
-      "loss": 3.3513,
+      "loss": 3.3206,
       "step": 50000
     },
     {
       "epoch": 5.3815520396082235,
-      "eval_accuracy": 0.38153733188274486,
-      "eval_loss": 3.4052817821502686,
-      "eval_runtime": 202.9452,
-      "eval_samples_per_second": 88.748,
-      "eval_steps_per_second": 5.548,
+      "eval_accuracy": 0.3819499955506642,
+      "eval_loss": 3.401759147644043,
+      "eval_runtime": 181.4692,
+      "eval_samples_per_second": 99.251,
+      "eval_steps_per_second": 6.205,
       "step": 50000
     },
     {
       "epoch": 5.386933591647831,
-      "grad_norm": 0.6867526173591614,
+      "grad_norm": 0.7145494818687439,
       "learning_rate": 0.0002771985777394677,
-      "loss": 3.3115,
+      "loss": 3.3263,
       "step": 50050
     },
     {
       "epoch": 5.39231514368744,
-      "grad_norm": 0.6913804411888123,
+      "grad_norm": 0.7484979033470154,
       "learning_rate": 0.0002768753367094063,
-      "loss": 3.3469,
+      "loss": 3.3174,
       "step": 50100
     },
     {
       "epoch": 5.397696695727047,
-      "grad_norm": 0.7089942097663879,
+      "grad_norm": 0.700779378414154,
       "learning_rate": 0.0002765520956793449,
-      "loss": 3.3267,
+      "loss": 3.318,
       "step": 50150
     },
     {
       "epoch": 5.403078247766656,
-      "grad_norm": 0.6663913726806641,
+      "grad_norm": 0.7063888311386108,
       "learning_rate": 0.00027622885464928347,
-      "loss": 3.3485,
+      "loss": 3.3295,
       "step": 50200
     },
     {
       "epoch": 5.4084597998062645,
-      "grad_norm": 0.6780911684036255,
+      "grad_norm": 0.6838209629058838,
       "learning_rate": 0.000275905613619222,
-      "loss": 3.323,
+      "loss": 3.3353,
       "step": 50250
     },
     {
       "epoch": 5.413841351845872,
-      "grad_norm": 0.6843054294586182,
+      "grad_norm": 0.7398064136505127,
       "learning_rate": 0.00027558237258916066,
-      "loss": 3.3323,
+      "loss": 3.334,
       "step": 50300
     },
     {
       "epoch": 5.419222903885481,
-      "grad_norm": 0.6963738799095154,
+      "grad_norm": 0.72173672914505,
       "learning_rate": 0.0002752591315590992,
-      "loss": 3.3252,
+      "loss": 3.3292,
       "step": 50350
     },
     {
       "epoch": 5.424604455925088,
-      "grad_norm": 0.7334024906158447,
+      "grad_norm": 0.7020224332809448,
       "learning_rate": 0.0002749358905290378,
-      "loss": 3.3618,
+      "loss": 3.3459,
       "step": 50400
     },
     {
       "epoch": 5.429986007964697,
-      "grad_norm": 0.7072751522064209,
+      "grad_norm": 0.7031869292259216,
       "learning_rate": 0.0002746126494989764,
-      "loss": 3.3496,
+      "loss": 3.3451,
       "step": 50450
     },
     {
       "epoch": 5.435367560004305,
-      "grad_norm": 0.6565462350845337,
+      "grad_norm": 0.6865600943565369,
       "learning_rate": 0.000274289408468915,
-      "loss": 3.3377,
+      "loss": 3.336,
       "step": 50500
     },
     {
       "epoch": 5.440749112043913,
-      "grad_norm": 0.7185359597206116,
+      "grad_norm": 0.6960546374320984,
       "learning_rate": 0.0002739661674388535,
-      "loss": 3.3336,
+      "loss": 3.3309,
       "step": 50550
     },
     {
       "epoch": 5.446130664083522,
-      "grad_norm": 0.6985934972763062,
+      "grad_norm": 0.7380642294883728,
       "learning_rate": 0.0002736429264087921,
-      "loss": 3.3497,
+      "loss": 3.3327,
       "step": 50600
     },
     {
       "epoch": 5.45151221612313,
-      "grad_norm": 0.778202474117279,
+      "grad_norm": 0.7257747054100037,
       "learning_rate": 0.0002733196853787307,
-      "loss": 3.332,
+      "loss": 3.3324,
       "step": 50650
     },
     {
       "epoch": 5.456893768162738,
-      "grad_norm": 0.7283360958099365,
+      "grad_norm": 0.6956778168678284,
       "learning_rate": 0.0002729964443486693,
-      "loss": 3.3343,
+      "loss": 3.3365,
       "step": 50700
     },
     {
       "epoch": 5.462275320202346,
-      "grad_norm": 0.7053759098052979,
+      "grad_norm": 0.6830939054489136,
       "learning_rate": 0.0002726732033186079,
-      "loss": 3.3314,
+      "loss": 3.3402,
       "step": 50750
     },
     {
       "epoch": 5.467656872241955,
-      "grad_norm": 0.7685927748680115,
+      "grad_norm": 0.6951338052749634,
       "learning_rate": 0.00027234996228854644,
-      "loss": 3.3358,
+      "loss": 3.3293,
       "step": 50800
     },
     {
       "epoch": 5.473038424281563,
-      "grad_norm": 0.6882610321044922,
+      "grad_norm": 0.719935953617096,
       "learning_rate": 0.0002720267212584851,
-      "loss": 3.3294,
+      "loss": 3.3189,
       "step": 50850
     },
     {
       "epoch": 5.478419976321171,
-      "grad_norm": 0.7401322722434998,
+      "grad_norm": 0.6854377388954163,
       "learning_rate": 0.00027170348022842363,
-      "loss": 3.333,
+      "loss": 3.3369,
       "step": 50900
     },
     {
       "epoch": 5.483801528360779,
-      "grad_norm": 0.7363923192024231,
+      "grad_norm": 0.6951292157173157,
       "learning_rate": 0.00027138023919836223,
-      "loss": 3.3233,
+      "loss": 3.3284,
       "step": 50950
     },
     {
       "epoch": 5.489183080400387,
-      "grad_norm": 0.6889253258705139,
-      "learning_rate": 0.00027106346298890204,
-      "loss": 3.334,
+      "grad_norm": 0.6584838628768921,
+      "learning_rate": 0.0002710569981683008,
+      "loss": 3.3296,
       "step": 51000
     },
     {
       "epoch": 5.489183080400387,
-      "eval_accuracy": 0.38220174429173037,
-      "eval_loss": 3.4004783630371094,
-      "eval_runtime": 190.6388,
-      "eval_samples_per_second": 94.477,
-      "eval_steps_per_second": 5.906,
+      "eval_accuracy": 0.38222923347234744,
+      "eval_loss": 3.396408796310425,
+      "eval_runtime": 181.7082,
+      "eval_samples_per_second": 99.12,
+      "eval_steps_per_second": 6.197,
       "step": 51000
     },
     {
       "epoch": 5.494564632439996,
-      "grad_norm": 0.7611465454101562,
-      "learning_rate": 0.00027074022195884063,
-      "loss": 3.3499,
+      "grad_norm": 0.7184561491012573,
+      "learning_rate": 0.0002707337571382394,
+      "loss": 3.3382,
       "step": 51050
     },
     {
       "epoch": 5.499946184479604,
-      "grad_norm": 0.6965069770812988,
-      "learning_rate": 0.0002704169809287792,
-      "loss": 3.3126,
+      "grad_norm": 0.676673412322998,
+      "learning_rate": 0.00027041051610817796,
+      "loss": 3.3359,
       "step": 51100
     },
     {
       "epoch": 5.505327736519212,
-      "grad_norm": 0.6993700861930847,
-      "learning_rate": 0.00027009373989871776,
-      "loss": 3.3497,
+      "grad_norm": 0.7439379692077637,
+      "learning_rate": 0.00027008727507811655,
+      "loss": 3.3459,
       "step": 51150
     },
     {
       "epoch": 5.510709288558821,
-      "grad_norm": 0.7188165187835693,
-      "learning_rate": 0.00026977049886865636,
-      "loss": 3.3468,
+      "grad_norm": 0.6832498908042908,
+      "learning_rate": 0.00026976403404805515,
+      "loss": 3.3312,
       "step": 51200
     },
     {
       "epoch": 5.516090840598428,
-      "grad_norm": 0.7151656746864319,
-      "learning_rate": 0.00026944725783859495,
-      "loss": 3.3301,
+      "grad_norm": 0.7079065442085266,
+      "learning_rate": 0.00026944079301799374,
+      "loss": 3.3353,
       "step": 51250
     },
     {
       "epoch": 5.521472392638037,
-      "grad_norm": 0.7025049924850464,
-      "learning_rate": 0.00026912401680853355,
-      "loss": 3.3395,
+      "grad_norm": 0.7613362669944763,
+      "learning_rate": 0.00026911755198793234,
+      "loss": 3.3278,
       "step": 51300
     },
     {
       "epoch": 5.5268539446776455,
-      "grad_norm": 0.7040926814079285,
-      "learning_rate": 0.0002688007757784721,
-      "loss": 3.3292,
+      "grad_norm": 0.7304935455322266,
+      "learning_rate": 0.0002687943109578709,
+      "loss": 3.3251,
       "step": 51350
     },
     {
       "epoch": 5.532235496717253,
-      "grad_norm": 0.7505577802658081,
-      "learning_rate": 0.00026847753474841074,
-      "loss": 3.3488,
+      "grad_norm": 0.6913647651672363,
+      "learning_rate": 0.00026847106992780947,
+      "loss": 3.3664,
       "step": 51400
     },
     {
       "epoch": 5.537617048756862,
-      "grad_norm": 0.7513281106948853,
+      "grad_norm": 0.7008777260780334,
       "learning_rate": 0.0002681542937183493,
-      "loss": 3.3521,
+      "loss": 3.325,
       "step": 51450
     },
     {
       "epoch": 5.542998600796469,
-      "grad_norm": 0.790105938911438,
+      "grad_norm": 0.7103151679039001,
       "learning_rate": 0.0002678310526882879,
-      "loss": 3.3423,
+      "loss": 3.3433,
       "step": 51500
     },
     {
       "epoch": 5.548380152836078,
-      "grad_norm": 0.7354403138160706,
+      "grad_norm": 0.703574001789093,
       "learning_rate": 0.00026750781165822647,
       "loss": 3.3304,
       "step": 51550
     },
     {
       "epoch": 5.553761704875686,
-      "grad_norm": 0.7015941739082336,
+      "grad_norm": 0.6837782263755798,
       "learning_rate": 0.00026718457062816506,
-      "loss": 3.3444,
+      "loss": 3.3369,
       "step": 51600
     },
     {
       "epoch": 5.559143256915294,
-      "grad_norm": 0.7292041182518005,
+      "grad_norm": 0.6965835690498352,
       "learning_rate": 0.00026686132959810366,
-      "loss": 3.3471,
+      "loss": 3.3378,
       "step": 51650
     },
     {
       "epoch": 5.564524808954903,
-      "grad_norm": 0.7539200186729431,
+      "grad_norm": 0.7734165191650391,
       "learning_rate": 0.0002665380885680422,
-      "loss": 3.3464,
+      "loss": 3.3376,
       "step": 51700
     },
     {
       "epoch": 5.569906360994511,
-      "grad_norm": 0.730979859828949,
+      "grad_norm": 0.6557449698448181,
       "learning_rate": 0.0002662148475379808,
-      "loss": 3.3338,
+      "loss": 3.3087,
       "step": 51750
     },
     {
       "epoch": 5.575287913034119,
-      "grad_norm": 0.7592946887016296,
+      "grad_norm": 0.712734580039978,
       "learning_rate": 0.0002658916065079194,
-      "loss": 3.3366,
+      "loss": 3.3312,
       "step": 51800
     },
     {
       "epoch": 5.580669465073727,
-      "grad_norm": 0.7109083533287048,
+      "grad_norm": 0.7088370323181152,
       "learning_rate": 0.000265568365477858,
-      "loss": 3.3529,
+      "loss": 3.3166,
       "step": 51850
     },
     {
       "epoch": 5.586051017113336,
-      "grad_norm": 0.7419684529304504,
+      "grad_norm": 0.689104437828064,
       "learning_rate": 0.0002652451244477965,
-      "loss": 3.333,
+      "loss": 3.3337,
       "step": 51900
     },
     {
       "epoch": 5.591432569152944,
-      "grad_norm": 0.7088695764541626,
+      "grad_norm": 0.7380256056785583,
       "learning_rate": 0.00026492188341773517,
-      "loss": 3.3527,
+      "loss": 3.3388,
       "step": 51950
     },
     {
       "epoch": 5.596814121192552,
-      "grad_norm": 0.721508800983429,
+      "grad_norm": 0.7266364693641663,
       "learning_rate": 0.0002645986423876737,
-      "loss": 3.334,
+      "loss": 3.3342,
       "step": 52000
     },
     {
       "epoch": 5.596814121192552,
-      "eval_accuracy": 0.3824253519348526,
-      "eval_loss": 3.3942997455596924,
-      "eval_runtime": 201.2401,
-      "eval_samples_per_second": 89.5,
-      "eval_steps_per_second": 5.595,
+      "eval_accuracy": 0.3828343214045863,
+      "eval_loss": 3.3914682865142822,
+      "eval_runtime": 181.6827,
+      "eval_samples_per_second": 99.134,
+      "eval_steps_per_second": 6.198,
       "step": 52000
     },
     {
       "epoch": 5.60219567323216,
-      "grad_norm": 0.7105787396430969,
+      "grad_norm": 0.7361786961555481,
       "learning_rate": 0.0002642754013576123,
-      "loss": 3.3619,
+      "loss": 3.3551,
       "step": 52050
     },
     {
       "epoch": 5.607577225271768,
-      "grad_norm": 0.7615808844566345,
+      "grad_norm": 0.6772825121879578,
       "learning_rate": 0.0002639521603275509,
-      "loss": 3.333,
+      "loss": 3.3393,
       "step": 52100
     },
     {
       "epoch": 5.612958777311377,
-      "grad_norm": 0.7386506199836731,
+      "grad_norm": 0.7617408037185669,
       "learning_rate": 0.0002636289192974895,
-      "loss": 3.3353,
+      "loss": 3.3206,
       "step": 52150
     },
     {
       "epoch": 5.618340329350985,
-      "grad_norm": 0.7279359102249146,
+      "grad_norm": 0.713563084602356,
       "learning_rate": 0.00026330567826742804,
-      "loss": 3.3381,
+      "loss": 3.3424,
       "step": 52200
     },
     {
       "epoch": 5.623721881390593,
-      "grad_norm": 0.7196009755134583,
+      "grad_norm": 0.699005663394928,
       "learning_rate": 0.00026298243723736663,
-      "loss": 3.3369,
+      "loss": 3.3161,
       "step": 52250
     },
     {
       "epoch": 5.629103433430201,
-      "grad_norm": 0.7014727592468262,
+      "grad_norm": 0.6893244981765747,
       "learning_rate": 0.0002626591962073052,
-      "loss": 3.3391,
+      "loss": 3.3522,
       "step": 52300
     },
     {
       "epoch": 5.634484985469809,
-      "grad_norm": 0.7047598958015442,
+      "grad_norm": 0.7023751139640808,
       "learning_rate": 0.0002623359551772438,
-      "loss": 3.3345,
+      "loss": 3.3459,
       "step": 52350
     },
     {
       "epoch": 5.639866537509418,
-      "grad_norm": 0.764114499092102,
+      "grad_norm": 0.7125876545906067,
       "learning_rate": 0.0002620127141471824,
-      "loss": 3.3531,
+      "loss": 3.3495,
       "step": 52400
     },
     {
       "epoch": 5.645248089549026,
-      "grad_norm": 0.7131911516189575,
+      "grad_norm": 0.7014057040214539,
       "learning_rate": 0.00026168947311712095,
-      "loss": 3.3343,
+      "loss": 3.3543,
       "step": 52450
     },
     {
       "epoch": 5.650629641588634,
-      "grad_norm": 0.7622030973434448,
+      "grad_norm": 0.7502124905586243,
       "learning_rate": 0.00026136623208705955,
-      "loss": 3.3465,
+      "loss": 3.3401,
       "step": 52500
     },
     {
       "epoch": 5.656011193628243,
-      "grad_norm": 0.7329249382019043,
+      "grad_norm": 0.7329356670379639,
       "learning_rate": 0.00026104299105699814,
-      "loss": 3.3301,
+      "loss": 3.3222,
       "step": 52550
     },
     {
       "epoch": 5.66139274566785,
-      "grad_norm": 0.7030691504478455,
+      "grad_norm": 0.6983018517494202,
       "learning_rate": 0.00026071975002693674,
-      "loss": 3.3294,
+      "loss": 3.3261,
       "step": 52600
     },
     {
       "epoch": 5.666774297707459,
-      "grad_norm": 0.7377075552940369,
+      "grad_norm": 0.7393155097961426,
       "learning_rate": 0.00026039650899687533,
-      "loss": 3.3589,
+      "loss": 3.34,
       "step": 52650
     },
     {
       "epoch": 5.672155849747067,
-      "grad_norm": 0.6939163208007812,
+      "grad_norm": 0.7433198690414429,
       "learning_rate": 0.00026007326796681393,
-      "loss": 3.3312,
+      "loss": 3.3342,
       "step": 52700
     },
     {
       "epoch": 5.677537401786675,
-      "grad_norm": 0.7312877178192139,
+      "grad_norm": 0.7320528626441956,
       "learning_rate": 0.00025975002693675247,
-      "loss": 3.343,
+      "loss": 3.3448,
       "step": 52750
     },
     {
       "epoch": 5.682918953826284,
-      "grad_norm": 0.7162089943885803,
+      "grad_norm": 0.7194253206253052,
       "learning_rate": 0.00025942678590669106,
-      "loss": 3.3215,
+      "loss": 3.3331,
       "step": 52800
     },
     {
       "epoch": 5.688300505865891,
-      "grad_norm": 0.7658792734146118,
+      "grad_norm": 0.7102786898612976,
       "learning_rate": 0.00025910354487662966,
-      "loss": 3.3147,
+      "loss": 3.3139,
       "step": 52850
     },
     {
       "epoch": 5.6936820579055,
-      "grad_norm": 0.7245572209358215,
+      "grad_norm": 0.7151712775230408,
       "learning_rate": 0.00025878030384656825,
-      "loss": 3.3414,
+      "loss": 3.329,
       "step": 52900
     },
     {
       "epoch": 5.699063609945108,
-      "grad_norm": 0.716014564037323,
+      "grad_norm": 0.7396782040596008,
       "learning_rate": 0.00025845706281650685,
-      "loss": 3.3339,
+      "loss": 3.3481,
       "step": 52950
     },
     {
       "epoch": 5.704445161984716,
-      "grad_norm": 0.7805140018463135,
+      "grad_norm": 0.7402505874633789,
       "learning_rate": 0.0002581338217864454,
-      "loss": 3.3328,
+      "loss": 3.3316,
       "step": 53000
     },
     {
       "epoch": 5.704445161984716,
-      "eval_accuracy": 0.3828685470642479,
-      "eval_loss": 3.3905136585235596,
-      "eval_runtime": 201.7044,
-      "eval_samples_per_second": 89.294,
-      "eval_steps_per_second": 5.582,
+      "eval_accuracy": 0.38359706467704396,
+      "eval_loss": 3.3873817920684814,
+      "eval_runtime": 181.344,
+      "eval_samples_per_second": 99.32,
+      "eval_steps_per_second": 6.209,
       "step": 53000
     },
     {
       "epoch": 5.709826714024325,
-      "grad_norm": 0.748485803604126,
+      "grad_norm": 0.6990689635276794,
       "learning_rate": 0.000257810580756384,
-      "loss": 3.3468,
+      "loss": 3.3314,
       "step": 53050
     },
     {
       "epoch": 5.715208266063933,
-      "grad_norm": 0.7458595037460327,
+      "grad_norm": 0.6929753422737122,
       "learning_rate": 0.0002574873397263226,
-      "loss": 3.3035,
+      "loss": 3.3294,
       "step": 53100
     },
     {
       "epoch": 5.720589818103541,
-      "grad_norm": 0.6711325645446777,
+      "grad_norm": 0.7152795791625977,
       "learning_rate": 0.00025716409869626117,
-      "loss": 3.3394,
+      "loss": 3.3197,
       "step": 53150
     },
     {
       "epoch": 5.725971370143149,
-      "grad_norm": 0.7426289319992065,
+      "grad_norm": 0.7226844429969788,
       "learning_rate": 0.0002568408576661997,
-      "loss": 3.3419,
+      "loss": 3.3388,
       "step": 53200
     },
     {
       "epoch": 5.731352922182758,
-      "grad_norm": 0.716453492641449,
+      "grad_norm": 0.6918237805366516,
       "learning_rate": 0.00025651761663613836,
-      "loss": 3.3418,
+      "loss": 3.3262,
       "step": 53250
     },
     {
       "epoch": 5.736734474222366,
-      "grad_norm": 0.7246510982513428,
+      "grad_norm": 0.7147321105003357,
       "learning_rate": 0.0002561943756060769,
-      "loss": 3.3546,
+      "loss": 3.3496,
       "step": 53300
     },
     {
       "epoch": 5.742116026261974,
-      "grad_norm": 0.705548882484436,
+      "grad_norm": 0.6990745067596436,
       "learning_rate": 0.0002558711345760155,
-      "loss": 3.3553,
+      "loss": 3.3323,
       "step": 53350
     },
     {
       "epoch": 5.747497578301582,
-      "grad_norm": 0.7819035053253174,
+      "grad_norm": 0.7370365858078003,
       "learning_rate": 0.0002555478935459541,
-      "loss": 3.325,
+      "loss": 3.3357,
       "step": 53400
     },
     {
       "epoch": 5.75287913034119,
-      "grad_norm": 0.7534633278846741,
+      "grad_norm": 0.732414722442627,
       "learning_rate": 0.00025522465251589263,
-      "loss": 3.3426,
+      "loss": 3.3358,
       "step": 53450
     },
     {
       "epoch": 5.758260682380799,
-      "grad_norm": 0.7411068081855774,
+      "grad_norm": 0.6718565225601196,
       "learning_rate": 0.0002549014114858312,
-      "loss": 3.3346,
+      "loss": 3.3266,
       "step": 53500
     },
     {
       "epoch": 5.763642234420407,
-      "grad_norm": 0.7141110897064209,
+      "grad_norm": 0.6832071542739868,
       "learning_rate": 0.0002545781704557698,
-      "loss": 3.3429,
+      "loss": 3.3443,
       "step": 53550
     },
     {
       "epoch": 5.769023786460015,
-      "grad_norm": 0.7201070189476013,
+      "grad_norm": 0.6670867204666138,
       "learning_rate": 0.0002542549294257084,
-      "loss": 3.3355,
+      "loss": 3.3396,
       "step": 53600
     },
     {
       "epoch": 5.774405338499624,
-      "grad_norm": 0.774978518486023,
+      "grad_norm": 0.6814727783203125,
       "learning_rate": 0.000253931688395647,
-      "loss": 3.3345,
+      "loss": 3.3414,
       "step": 53650
     },
     {
       "epoch": 5.779786890539231,
-      "grad_norm": 0.7548540830612183,
+      "grad_norm": 0.7520219087600708,
       "learning_rate": 0.0002536084473655856,
-      "loss": 3.3389,
+      "loss": 3.3162,
       "step": 53700
     },
     {
       "epoch": 5.78516844257884,
-      "grad_norm": 0.6902544498443604,
+      "grad_norm": 0.7816487550735474,
       "learning_rate": 0.00025328520633552415,
-      "loss": 3.3277,
+      "loss": 3.3271,
       "step": 53750
     },
     {
       "epoch": 5.790549994618448,
-      "grad_norm": 0.7907364964485168,
+      "grad_norm": 0.6912047863006592,
       "learning_rate": 0.00025296196530546274,
-      "loss": 3.3482,
+      "loss": 3.3446,
       "step": 53800
     },
     {
       "epoch": 5.795931546658056,
-      "grad_norm": 0.7004760503768921,
+      "grad_norm": 0.7378522157669067,
       "learning_rate": 0.00025263872427540133,
-      "loss": 3.3535,
+      "loss": 3.3456,
       "step": 53850
     },
     {
       "epoch": 5.801313098697665,
-      "grad_norm": 0.7246608138084412,
+      "grad_norm": 0.7622185349464417,
       "learning_rate": 0.00025231548324533993,
-      "loss": 3.3375,
+      "loss": 3.3399,
       "step": 53900
     },
     {
       "epoch": 5.806694650737272,
-      "grad_norm": 0.713356077671051,
+      "grad_norm": 0.7270689606666565,
       "learning_rate": 0.0002519922422152785,
-      "loss": 3.3441,
+      "loss": 3.3567,
       "step": 53950
     },
     {
       "epoch": 5.812076202776881,
-      "grad_norm": 0.7337074279785156,
+      "grad_norm": 0.7550874352455139,
       "learning_rate": 0.00025166900118521706,
-      "loss": 3.3229,
+      "loss": 3.312,
       "step": 54000
     },
     {
       "epoch": 5.812076202776881,
-      "eval_accuracy": 0.3832519831053452,
-      "eval_loss": 3.3878395557403564,
-      "eval_runtime": 188.3897,
-      "eval_samples_per_second": 95.605,
-      "eval_steps_per_second": 5.977,
+      "eval_accuracy": 0.38383990388130934,
+      "eval_loss": 3.383046865463257,
+      "eval_runtime": 181.3151,
+      "eval_samples_per_second": 99.335,
+      "eval_steps_per_second": 6.21,
       "step": 54000
     },
     {
       "epoch": 5.817457754816489,
-      "grad_norm": 0.7938897609710693,
-      "learning_rate": 0.0002513522249757569,
-      "loss": 3.329,
+      "grad_norm": 0.729927659034729,
+      "learning_rate": 0.00025134576015515566,
+      "loss": 3.3308,
       "step": 54050
     },
     {
       "epoch": 5.822839306856097,
-      "grad_norm": 0.707722544670105,
-      "learning_rate": 0.00025102898394569547,
-      "loss": 3.3292,
+      "grad_norm": 0.693851113319397,
+      "learning_rate": 0.00025102251912509425,
+      "loss": 3.326,
       "step": 54100
     },
     {
       "epoch": 5.828220858895706,
-      "grad_norm": 0.7328303456306458,
-      "learning_rate": 0.00025070574291563406,
-      "loss": 3.3384,
+      "grad_norm": 0.7468603253364563,
+      "learning_rate": 0.00025069927809503285,
+      "loss": 3.3314,
       "step": 54150
     },
     {
       "epoch": 5.833602410935313,
-      "grad_norm": 0.7096349000930786,
-      "learning_rate": 0.00025038250188557265,
-      "loss": 3.3243,
+      "grad_norm": 0.7270225882530212,
+      "learning_rate": 0.0002503760370649714,
+      "loss": 3.3303,
       "step": 54200
     },
     {
       "epoch": 5.838983962974922,
-      "grad_norm": 0.7162660360336304,
-      "learning_rate": 0.00025005926085551125,
-      "loss": 3.3405,
+      "grad_norm": 0.7075627446174622,
+      "learning_rate": 0.00025005279603491004,
+      "loss": 3.3314,
       "step": 54250
     },
     {
       "epoch": 5.84436551501453,
-      "grad_norm": 0.7353925108909607,
-      "learning_rate": 0.0002497360198254498,
-      "loss": 3.3328,
+      "grad_norm": 0.6881726980209351,
+      "learning_rate": 0.0002497295550048486,
+      "loss": 3.3224,
       "step": 54300
     },
     {
       "epoch": 5.849747067054138,
-      "grad_norm": 0.6959530115127563,
-      "learning_rate": 0.00024941277879538844,
-      "loss": 3.3515,
+      "grad_norm": 0.7222962975502014,
+      "learning_rate": 0.00024940631397478717,
+      "loss": 3.3412,
       "step": 54350
     },
     {
       "epoch": 5.855128619093747,
-      "grad_norm": 0.7663270235061646,
-      "learning_rate": 0.000249089537765327,
-      "loss": 3.3076,
+      "grad_norm": 0.7224239706993103,
+      "learning_rate": 0.00024908307294472577,
+      "loss": 3.3057,
       "step": 54400
     },
     {
       "epoch": 5.860510171133355,
-      "grad_norm": 0.6977930665016174,
-      "learning_rate": 0.0002487662967352656,
-      "loss": 3.3244,
+      "grad_norm": 0.6597974896430969,
+      "learning_rate": 0.00024875983191466436,
+      "loss": 3.3277,
       "step": 54450
     },
     {
       "epoch": 5.865891723172963,
-      "grad_norm": 0.7649244666099548,
-      "learning_rate": 0.00024844305570520417,
-      "loss": 3.3401,
+      "grad_norm": 0.7524548172950745,
+      "learning_rate": 0.0002484365908846029,
+      "loss": 3.3397,
       "step": 54500
     },
     {
       "epoch": 5.871273275212571,
-      "grad_norm": 0.7535280585289001,
-      "learning_rate": 0.00024811981467514276,
-      "loss": 3.3485,
+      "grad_norm": 0.705040454864502,
+      "learning_rate": 0.0002481133498545415,
+      "loss": 3.3342,
       "step": 54550
     },
     {
       "epoch": 5.87665482725218,
-      "grad_norm": 0.6970841884613037,
-      "learning_rate": 0.00024779657364508136,
-      "loss": 3.3303,
+      "grad_norm": 0.7237173318862915,
+      "learning_rate": 0.0002477901088244801,
+      "loss": 3.3304,
       "step": 54600
     },
     {
       "epoch": 5.882036379291788,
-      "grad_norm": 0.7283545732498169,
-      "learning_rate": 0.0002474733326150199,
-      "loss": 3.3424,
+      "grad_norm": 0.7052015066146851,
+      "learning_rate": 0.0002474668677944187,
+      "loss": 3.3336,
       "step": 54650
     },
     {
       "epoch": 5.887417931331396,
-      "grad_norm": 0.721764326095581,
-      "learning_rate": 0.0002471500915849585,
-      "loss": 3.341,
+      "grad_norm": 0.7461618781089783,
+      "learning_rate": 0.0002471436267643573,
+      "loss": 3.3475,
       "step": 54700
     },
     {
       "epoch": 5.892799483371004,
-      "grad_norm": 0.687520444393158,
-      "learning_rate": 0.0002468268505548971,
-      "loss": 3.3415,
+      "grad_norm": 0.6796166300773621,
+      "learning_rate": 0.0002468203857342958,
+      "loss": 3.3442,
       "step": 54750
     },
     {
       "epoch": 5.898181035410612,
-      "grad_norm": 0.7138300538063049,
-      "learning_rate": 0.0002465036095248357,
-      "loss": 3.3447,
+      "grad_norm": 0.7980848550796509,
+      "learning_rate": 0.00024649714470423447,
+      "loss": 3.3362,
       "step": 54800
     },
     {
       "epoch": 5.903562587450221,
-      "grad_norm": 0.6954256296157837,
-      "learning_rate": 0.0002461803684947742,
-      "loss": 3.3466,
+      "grad_norm": 0.8394770622253418,
+      "learning_rate": 0.000246173903674173,
+      "loss": 3.3288,
       "step": 54850
     },
     {
       "epoch": 5.9089441394898286,
-      "grad_norm": 0.7353760004043579,
-      "learning_rate": 0.0002458571274647128,
-      "loss": 3.3411,
+      "grad_norm": 0.7369672060012817,
+      "learning_rate": 0.0002458506626441116,
+      "loss": 3.3302,
       "step": 54900
     },
     {
       "epoch": 5.914325691529437,
-      "grad_norm": 0.8013879656791687,
-      "learning_rate": 0.0002455338864346514,
-      "loss": 3.3581,
+      "grad_norm": 0.720160186290741,
+      "learning_rate": 0.0002455274216140502,
+      "loss": 3.3404,
       "step": 54950
     },
     {
       "epoch": 5.919707243569046,
-      "grad_norm": 0.7387758493423462,
-      "learning_rate": 0.00024521064540459,
-      "loss": 3.3521,
+      "grad_norm": 0.7501629590988159,
+      "learning_rate": 0.0002452041805839888,
+      "loss": 3.3368,
       "step": 55000
     },
     {
       "epoch": 5.919707243569046,
-      "eval_accuracy": 0.3836273788327442,
-      "eval_loss": 3.3835108280181885,
-      "eval_runtime": 197.7765,
-      "eval_samples_per_second": 91.067,
-      "eval_steps_per_second": 5.693,
+      "eval_accuracy": 0.3840953468205612,
+      "eval_loss": 3.378291368484497,
+      "eval_runtime": 181.6053,
+      "eval_samples_per_second": 99.177,
+      "eval_steps_per_second": 6.2,
       "step": 55000
     },
     {
       "epoch": 5.925088795608653,
-      "grad_norm": 0.7418875098228455,
-      "learning_rate": 0.0002448874043745286,
-      "loss": 3.3297,
+      "grad_norm": 0.694675087928772,
+      "learning_rate": 0.00024488093955392734,
+      "loss": 3.3315,
       "step": 55050
     },
     {
       "epoch": 5.930470347648262,
-      "grad_norm": 0.6959251165390015,
-      "learning_rate": 0.00024456416334446714,
-      "loss": 3.3306,
+      "grad_norm": 0.6841257214546204,
+      "learning_rate": 0.00024455769852386593,
+      "loss": 3.3369,
       "step": 55100
     },
     {
       "epoch": 5.93585189968787,
-      "grad_norm": 0.6997560858726501,
-      "learning_rate": 0.00024424092231440574,
-      "loss": 3.3368,
+      "grad_norm": 0.7018930315971375,
+      "learning_rate": 0.0002442344574938045,
+      "loss": 3.3253,
       "step": 55150
     },
     {
       "epoch": 5.941233451727478,
-      "grad_norm": 0.6774752736091614,
-      "learning_rate": 0.00024391768128434436,
-      "loss": 3.3398,
+      "grad_norm": 0.6705164909362793,
+      "learning_rate": 0.0002439112164637431,
+      "loss": 3.3367,
       "step": 55200
     },
     {
       "epoch": 5.946615003767087,
-      "grad_norm": 0.728191614151001,
-      "learning_rate": 0.00024359444025428293,
-      "loss": 3.3474,
+      "grad_norm": 0.7306385636329651,
+      "learning_rate": 0.00024358797543368171,
+      "loss": 3.3565,
       "step": 55250
     },
     {
       "epoch": 5.951996555806694,
-      "grad_norm": 0.7751122713088989,
-      "learning_rate": 0.0002432711992242215,
-      "loss": 3.3345,
+      "grad_norm": 0.7234976887702942,
+      "learning_rate": 0.00024326473440362028,
+      "loss": 3.3376,
       "step": 55300
     },
     {
       "epoch": 5.957378107846303,
-      "grad_norm": 0.6916903257369995,
-      "learning_rate": 0.0002429479581941601,
-      "loss": 3.3525,
+      "grad_norm": 0.6947909593582153,
+      "learning_rate": 0.00024294149337355885,
+      "loss": 3.3506,
       "step": 55350
     },
     {
       "epoch": 5.962759659885911,
-      "grad_norm": 0.704298734664917,
-      "learning_rate": 0.00024262471716409868,
-      "loss": 3.3457,
+      "grad_norm": 0.7112725973129272,
+      "learning_rate": 0.00024261825234349744,
+      "loss": 3.3216,
       "step": 55400
     },
     {
       "epoch": 5.968141211925519,
-      "grad_norm": 0.7079257369041443,
-      "learning_rate": 0.00024230147613403728,
-      "loss": 3.3441,
+      "grad_norm": 0.7279510498046875,
+      "learning_rate": 0.00024229501131343604,
+      "loss": 3.3297,
       "step": 55450
     },
     {
       "epoch": 5.973522763965128,
-      "grad_norm": 0.8067532777786255,
-      "learning_rate": 0.00024197823510397584,
-      "loss": 3.3431,
+      "grad_norm": 0.7029486298561096,
+      "learning_rate": 0.00024198469992457706,
+      "loss": 3.3368,
       "step": 55500
     },
     {
       "epoch": 5.978904316004736,
-      "grad_norm": 0.7205045819282532,
-      "learning_rate": 0.0002416549940739144,
-      "loss": 3.3412,
+      "grad_norm": 0.701117992401123,
+      "learning_rate": 0.00024166145889451568,
+      "loss": 3.3295,
       "step": 55550
     },
     {
       "epoch": 5.984285868044344,
-      "grad_norm": 0.729873538017273,
-      "learning_rate": 0.00024133175304385303,
-      "loss": 3.3155,
+      "grad_norm": 0.6714588403701782,
+      "learning_rate": 0.00024133821786445425,
+      "loss": 3.3246,
       "step": 55600
     },
     {
       "epoch": 5.989667420083952,
-      "grad_norm": 0.6925444602966309,
-      "learning_rate": 0.0002410085120137916,
-      "loss": 3.3504,
+      "grad_norm": 0.6800738573074341,
+      "learning_rate": 0.0002410149768343928,
+      "loss": 3.3205,
       "step": 55650
     },
     {
       "epoch": 5.995048972123561,
-      "grad_norm": 0.7201747298240662,
-      "learning_rate": 0.00024068527098373017,
-      "loss": 3.3241,
+      "grad_norm": 0.7238479852676392,
+      "learning_rate": 0.0002406917358043314,
+      "loss": 3.3421,
       "step": 55700
     },
     {
       "epoch": 6.000430524163169,
-      "grad_norm": 0.7336894273757935,
-      "learning_rate": 0.0002403620299536688,
-      "loss": 3.3297,
+      "grad_norm": 0.7229350209236145,
+      "learning_rate": 0.00024036849477427,
+      "loss": 3.3279,
       "step": 55750
     },
     {
       "epoch": 6.005812076202777,
-      "grad_norm": 0.6997312307357788,
-      "learning_rate": 0.00024003878892360736,
-      "loss": 3.232,
+      "grad_norm": 0.7066360116004944,
+      "learning_rate": 0.00024004525374420857,
+      "loss": 3.2384,
       "step": 55800
     },
     {
       "epoch": 6.011193628242385,
-      "grad_norm": 0.7346082329750061,
-      "learning_rate": 0.00023971554789354593,
-      "loss": 3.2338,
+      "grad_norm": 0.7275376915931702,
+      "learning_rate": 0.00023972201271414716,
+      "loss": 3.2388,
       "step": 55850
     },
     {
       "epoch": 6.016575180281993,
-      "grad_norm": 0.7229931950569153,
-      "learning_rate": 0.00023939230686348452,
-      "loss": 3.2316,
+      "grad_norm": 0.7457011342048645,
+      "learning_rate": 0.00023939877168408573,
+      "loss": 3.2336,
       "step": 55900
     },
     {
       "epoch": 6.021956732321602,
-      "grad_norm": 0.7411378622055054,
-      "learning_rate": 0.0002390690658334231,
-      "loss": 3.2539,
+      "grad_norm": 0.7064995765686035,
+      "learning_rate": 0.00023907553065402433,
+      "loss": 3.2549,
       "step": 55950
     },
     {
       "epoch": 6.0273382843612096,
-      "grad_norm": 0.6895947456359863,
-      "learning_rate": 0.00023874582480336168,
-      "loss": 3.2608,
+      "grad_norm": 0.7151305675506592,
+      "learning_rate": 0.00023875228962396292,
+      "loss": 3.2563,
       "step": 56000
     },
     {
       "epoch": 6.0273382843612096,
-      "eval_accuracy": 0.3839755026853018,
-      "eval_loss": 3.3861966133117676,
-      "eval_runtime": 211.6559,
-      "eval_samples_per_second": 85.096,
-      "eval_steps_per_second": 5.32,
+      "eval_accuracy": 0.3845146383146372,
+      "eval_loss": 3.3794331550598145,
+      "eval_runtime": 181.4783,
+      "eval_samples_per_second": 99.246,
+      "eval_steps_per_second": 6.205,
       "step": 56000
     },
     {
       "epoch": 6.032719836400818,
-      "grad_norm": 0.7139065265655518,
-      "learning_rate": 0.00023842258377330028,
-      "loss": 3.2516,
+      "grad_norm": 0.6810989379882812,
+      "learning_rate": 0.0002384290485939015,
+      "loss": 3.2435,
       "step": 56050
     },
     {
       "epoch": 6.038101388440427,
-      "grad_norm": 0.7533993124961853,
+      "grad_norm": 0.7020310163497925,
       "learning_rate": 0.00023810580756384006,
-      "loss": 3.2595,
+      "loss": 3.2497,
       "step": 56100
     },
     {
       "epoch": 6.043482940480034,
-      "grad_norm": 0.7737258076667786,
+      "grad_norm": 0.7535426616668701,
       "learning_rate": 0.00023778256653377868,
-      "loss": 3.2565,
+      "loss": 3.2491,
       "step": 56150
     },
     {
       "epoch": 6.048864492519643,
-      "grad_norm": 0.709679126739502,
+      "grad_norm": 0.7178252339363098,
       "learning_rate": 0.00023745932550371725,
-      "loss": 3.2624,
+      "loss": 3.2746,
       "step": 56200
     },
     {
       "epoch": 6.0542460445592505,
-      "grad_norm": 0.7647435069084167,
+      "grad_norm": 0.7159394025802612,
       "learning_rate": 0.00023713608447365584,
-      "loss": 3.2578,
+      "loss": 3.2571,
       "step": 56250
     },
     {
       "epoch": 6.059627596598859,
-      "grad_norm": 0.7607208490371704,
+      "grad_norm": 0.7339926362037659,
       "learning_rate": 0.00023681284344359444,
-      "loss": 3.2503,
+      "loss": 3.2572,
       "step": 56300
     },
     {
       "epoch": 6.065009148638468,
-      "grad_norm": 0.76082843542099,
+      "grad_norm": 0.6966595649719238,
       "learning_rate": 0.000236489602413533,
-      "loss": 3.2469,
+      "loss": 3.2587,
       "step": 56350
     },
     {
       "epoch": 6.070390700678075,
-      "grad_norm": 0.7211427688598633,
+      "grad_norm": 0.7386974692344666,
       "learning_rate": 0.0002361663613834716,
-      "loss": 3.2735,
+      "loss": 3.2598,
       "step": 56400
     },
     {
       "epoch": 6.075772252717684,
-      "grad_norm": 0.7562023401260376,
+      "grad_norm": 0.7356476783752441,
       "learning_rate": 0.00023584312035341017,
-      "loss": 3.2647,
+      "loss": 3.2582,
       "step": 56450
     },
     {
       "epoch": 6.081153804757292,
-      "grad_norm": 0.7383215427398682,
+      "grad_norm": 0.6955512166023254,
       "learning_rate": 0.00023551987932334876,
-      "loss": 3.2576,
+      "loss": 3.2741,
       "step": 56500
     },
     {
       "epoch": 6.0865353567969,
-      "grad_norm": 0.7635093331336975,
+      "grad_norm": 0.7615606784820557,
       "learning_rate": 0.00023519663829328735,
-      "loss": 3.2469,
+      "loss": 3.2503,
       "step": 56550
     },
     {
       "epoch": 6.091916908836509,
-      "grad_norm": 0.740917980670929,
+      "grad_norm": 0.6932699680328369,
       "learning_rate": 0.00023487339726322592,
-      "loss": 3.2686,
+      "loss": 3.2401,
       "step": 56600
     },
     {
       "epoch": 6.097298460876116,
-      "grad_norm": 0.8048155307769775,
+      "grad_norm": 0.7025453448295593,
       "learning_rate": 0.0002345501562331645,
-      "loss": 3.2547,
+      "loss": 3.2706,
       "step": 56650
     },
     {
       "epoch": 6.102680012915725,
-      "grad_norm": 0.7598278522491455,
+      "grad_norm": 0.7902150750160217,
       "learning_rate": 0.0002342269152031031,
-      "loss": 3.2496,
+      "loss": 3.2586,
       "step": 56700
     },
     {
       "epoch": 6.108061564955333,
-      "grad_norm": 0.7499753832817078,
+      "grad_norm": 0.7198737859725952,
       "learning_rate": 0.00023390367417304168,
-      "loss": 3.2653,
+      "loss": 3.2627,
       "step": 56750
     },
     {
       "epoch": 6.113443116994941,
-      "grad_norm": 0.7181298136711121,
+      "grad_norm": 0.7331535816192627,
       "learning_rate": 0.00023358043314298025,
-      "loss": 3.2557,
+      "loss": 3.2529,
       "step": 56800
     },
     {
       "epoch": 6.11882466903455,
-      "grad_norm": 0.7646988034248352,
+      "grad_norm": 0.7271235585212708,
       "learning_rate": 0.00023325719211291887,
-      "loss": 3.2724,
+      "loss": 3.265,
       "step": 56850
     },
     {
       "epoch": 6.124206221074158,
-      "grad_norm": 0.7216889262199402,
+      "grad_norm": 0.7499111890792847,
       "learning_rate": 0.00023293395108285744,
-      "loss": 3.2523,
+      "loss": 3.2462,
       "step": 56900
     },
     {
       "epoch": 6.129587773113766,
-      "grad_norm": 0.7427191734313965,
+      "grad_norm": 0.684013307094574,
       "learning_rate": 0.000232610710052796,
-      "loss": 3.2762,
+      "loss": 3.2612,
       "step": 56950
     },
     {
       "epoch": 6.134969325153374,
-      "grad_norm": 0.742785632610321,
+      "grad_norm": 0.7347684502601624,
       "learning_rate": 0.0002322874690227346,
-      "loss": 3.2733,
+      "loss": 3.2657,
       "step": 57000
     },
     {
       "epoch": 6.134969325153374,
-      "eval_accuracy": 0.38427451543256724,
-      "eval_loss": 3.3876760005950928,
-      "eval_runtime": 199.7347,
-      "eval_samples_per_second": 90.175,
-      "eval_steps_per_second": 5.637,
+      "eval_accuracy": 0.3845138777444225,
+      "eval_loss": 3.380876064300537,
+      "eval_runtime": 181.1528,
+      "eval_samples_per_second": 99.424,
+      "eval_steps_per_second": 6.216,
       "step": 57000
     },
     {
       "epoch": 6.140350877192983,
-      "grad_norm": 0.7947841882705688,
+      "grad_norm": 0.7603256106376648,
       "learning_rate": 0.0002319642279926732,
-      "loss": 3.2932,
+      "loss": 3.274,
       "step": 57050
     },
     {
       "epoch": 6.1457324292325906,
-      "grad_norm": 0.7083519101142883,
+      "grad_norm": 0.7237735986709595,
       "learning_rate": 0.00023164098696261176,
-      "loss": 3.2631,
+      "loss": 3.2656,
       "step": 57100
     },
     {
       "epoch": 6.151113981272199,
-      "grad_norm": 0.7376063466072083,
+      "grad_norm": 0.7621842622756958,
       "learning_rate": 0.00023131774593255036,
-      "loss": 3.2754,
+      "loss": 3.2773,
       "step": 57150
     },
     {
       "epoch": 6.156495533311807,
-      "grad_norm": 0.7607095241546631,
+      "grad_norm": 0.7423885464668274,
       "learning_rate": 0.00023099450490248892,
-      "loss": 3.282,
+      "loss": 3.2761,
       "step": 57200
     },
     {
       "epoch": 6.161877085351415,
-      "grad_norm": 0.7469072341918945,
+      "grad_norm": 0.7608686089515686,
       "learning_rate": 0.00023067126387242754,
-      "loss": 3.2629,
+      "loss": 3.263,
       "step": 57250
     },
     {
       "epoch": 6.167258637391024,
-      "grad_norm": 0.7765781283378601,
+      "grad_norm": 0.786966860294342,
       "learning_rate": 0.0002303480228423661,
-      "loss": 3.2591,
+      "loss": 3.2493,
       "step": 57300
     },
     {
       "epoch": 6.1726401894306315,
-      "grad_norm": 0.7496424913406372,
+      "grad_norm": 0.797631025314331,
       "learning_rate": 0.00023002478181230468,
-      "loss": 3.2608,
+      "loss": 3.251,
       "step": 57350
     },
     {
       "epoch": 6.17802174147024,
-      "grad_norm": 0.7514758706092834,
+      "grad_norm": 0.7511717677116394,
       "learning_rate": 0.00022970154078224327,
-      "loss": 3.2634,
+      "loss": 3.2594,
       "step": 57400
     },
     {
       "epoch": 6.183403293509849,
-      "grad_norm": 0.7652694582939148,
+      "grad_norm": 0.7554537057876587,
       "learning_rate": 0.00022937829975218187,
-      "loss": 3.2699,
+      "loss": 3.2597,
       "step": 57450
     },
     {
       "epoch": 6.188784845549456,
-      "grad_norm": 0.734973132610321,
+      "grad_norm": 0.7471697926521301,
       "learning_rate": 0.00022905505872212044,
-      "loss": 3.2727,
+      "loss": 3.2733,
       "step": 57500
     },
     {
       "epoch": 6.194166397589065,
-      "grad_norm": 0.7655386328697205,
+      "grad_norm": 0.743618905544281,
       "learning_rate": 0.00022873181769205903,
-      "loss": 3.2568,
+      "loss": 3.2479,
       "step": 57550
     },
     {
       "epoch": 6.1995479496286725,
-      "grad_norm": 0.744789719581604,
+      "grad_norm": 0.7307191491127014,
       "learning_rate": 0.0002284085766619976,
-      "loss": 3.2845,
+      "loss": 3.2556,
       "step": 57600
     },
     {
       "epoch": 6.204929501668281,
-      "grad_norm": 0.7866900563240051,
+      "grad_norm": 0.7556213736534119,
       "learning_rate": 0.0002280853356319362,
-      "loss": 3.2839,
+      "loss": 3.2716,
       "step": 57650
     },
     {
       "epoch": 6.21031105370789,
-      "grad_norm": 0.753262460231781,
+      "grad_norm": 0.7372322678565979,
       "learning_rate": 0.0002277620946018748,
-      "loss": 3.2868,
+      "loss": 3.2734,
       "step": 57700
     },
     {
       "epoch": 6.215692605747497,
-      "grad_norm": 0.7368378639221191,
+      "grad_norm": 0.7232465744018555,
       "learning_rate": 0.00022743885357181336,
-      "loss": 3.2648,
+      "loss": 3.2577,
       "step": 57750
     },
     {
       "epoch": 6.221074157787106,
-      "grad_norm": 0.7708497643470764,
+      "grad_norm": 0.7028861045837402,
       "learning_rate": 0.00022711561254175192,
-      "loss": 3.2733,
+      "loss": 3.2691,
       "step": 57800
     },
     {
       "epoch": 6.226455709826714,
-      "grad_norm": 0.7500898241996765,
+      "grad_norm": 0.7363249063491821,
       "learning_rate": 0.00022679237151169054,
-      "loss": 3.2719,
+      "loss": 3.2737,
       "step": 57850
     },
     {
       "epoch": 6.231837261866322,
-      "grad_norm": 0.7447527647018433,
+      "grad_norm": 0.7528419494628906,
       "learning_rate": 0.0002264691304816291,
-      "loss": 3.2606,
+      "loss": 3.2685,
       "step": 57900
     },
     {
       "epoch": 6.237218813905931,
-      "grad_norm": 0.7365005016326904,
+      "grad_norm": 0.7231560945510864,
       "learning_rate": 0.00022614588945156768,
-      "loss": 3.2832,
+      "loss": 3.276,
       "step": 57950
     },
     {
       "epoch": 6.242600365945538,
-      "grad_norm": 0.7803393006324768,
+      "grad_norm": 0.7695891261100769,
       "learning_rate": 0.0002258226484215063,
-      "loss": 3.2897,
+      "loss": 3.2641,
       "step": 58000
     },
     {
       "epoch": 6.242600365945538,
-      "eval_accuracy": 0.38468098588588123,
-      "eval_loss": 3.381647825241089,
-      "eval_runtime": 201.6412,
-      "eval_samples_per_second": 89.322,
-      "eval_steps_per_second": 5.584,
+      "eval_accuracy": 0.38499488407877724,
+      "eval_loss": 3.3782927989959717,
+      "eval_runtime": 181.7534,
+      "eval_samples_per_second": 99.096,
+      "eval_steps_per_second": 6.195,
       "step": 58000
     },
     {
       "epoch": 6.247981917985147,
-      "grad_norm": 0.7322475910186768,
+      "grad_norm": 0.6853750348091125,
       "learning_rate": 0.00022549940739144487,
-      "loss": 3.2811,
+      "loss": 3.2834,
       "step": 58050
     },
     {
       "epoch": 6.253363470024755,
-      "grad_norm": 0.7434014081954956,
+      "grad_norm": 0.7114074230194092,
       "learning_rate": 0.00022517616636138344,
-      "loss": 3.2893,
+      "loss": 3.2764,
       "step": 58100
     },
     {
       "epoch": 6.258745022064363,
-      "grad_norm": 0.7110326290130615,
+      "grad_norm": 0.7310791015625,
       "learning_rate": 0.00022485292533132203,
-      "loss": 3.2929,
+      "loss": 3.265,
       "step": 58150
     },
     {
       "epoch": 6.264126574103972,
-      "grad_norm": 0.7830908298492432,
+      "grad_norm": 0.7508020401000977,
       "learning_rate": 0.00022452968430126063,
-      "loss": 3.2784,
+      "loss": 3.278,
       "step": 58200
     },
     {
       "epoch": 6.26950812614358,
-      "grad_norm": 0.7722041606903076,
+      "grad_norm": 0.722253680229187,
       "learning_rate": 0.00022420644327119922,
-      "loss": 3.2607,
+      "loss": 3.273,
       "step": 58250
     },
     {
       "epoch": 6.274889678183188,
-      "grad_norm": 0.7421392202377319,
+      "grad_norm": 0.7455535531044006,
       "learning_rate": 0.0002238832022411378,
-      "loss": 3.2634,
+      "loss": 3.2699,
       "step": 58300
     },
     {
       "epoch": 6.280271230222796,
-      "grad_norm": 0.7481209635734558,
+      "grad_norm": 0.7685554027557373,
       "learning_rate": 0.00022355996121107636,
-      "loss": 3.2692,
+      "loss": 3.2679,
       "step": 58350
     },
     {
       "epoch": 6.285652782262405,
-      "grad_norm": 0.7708833813667297,
+      "grad_norm": 0.7710944414138794,
       "learning_rate": 0.00022323672018101498,
-      "loss": 3.2706,
+      "loss": 3.2684,
       "step": 58400
     },
     {
       "epoch": 6.2910343343020125,
-      "grad_norm": 0.8033352494239807,
+      "grad_norm": 0.7518188953399658,
       "learning_rate": 0.00022291347915095355,
-      "loss": 3.2784,
+      "loss": 3.2723,
       "step": 58450
     },
     {
       "epoch": 6.296415886341621,
-      "grad_norm": 0.7255500555038452,
+      "grad_norm": 0.7095345854759216,
       "learning_rate": 0.0002225902381208921,
-      "loss": 3.2862,
+      "loss": 3.2769,
       "step": 58500
     },
     {
       "epoch": 6.301797438381229,
-      "grad_norm": 0.7180657386779785,
+      "grad_norm": 0.6977404356002808,
       "learning_rate": 0.00022226699709083073,
-      "loss": 3.2808,
+      "loss": 3.2736,
       "step": 58550
     },
     {
       "epoch": 6.307178990420837,
-      "grad_norm": 0.8310112953186035,
+      "grad_norm": 0.7841971516609192,
       "learning_rate": 0.0002219437560607693,
-      "loss": 3.2954,
+      "loss": 3.2648,
       "step": 58600
     },
     {
       "epoch": 6.312560542460446,
-      "grad_norm": 0.7167147994041443,
+      "grad_norm": 0.7735727429389954,
       "learning_rate": 0.00022162051503070787,
-      "loss": 3.2825,
+      "loss": 3.2796,
       "step": 58650
     },
     {
       "epoch": 6.3179420945000535,
-      "grad_norm": 0.8282204866409302,
+      "grad_norm": 0.7525529861450195,
       "learning_rate": 0.00022129727400064646,
-      "loss": 3.2813,
+      "loss": 3.2736,
       "step": 58700
     },
     {
       "epoch": 6.323323646539662,
-      "grad_norm": 0.7656278014183044,
+      "grad_norm": 0.7922319173812866,
       "learning_rate": 0.00022097403297058506,
-      "loss": 3.282,
+      "loss": 3.2759,
       "step": 58750
     },
     {
       "epoch": 6.328705198579271,
-      "grad_norm": 0.729318380355835,
-      "learning_rate": 0.00022065079194052363,
-      "loss": 3.2716,
+      "grad_norm": 0.7689076662063599,
+      "learning_rate": 0.00022065725676112487,
+      "loss": 3.2805,
       "step": 58800
     },
     {
       "epoch": 6.334086750618878,
-      "grad_norm": 0.7712202072143555,
-      "learning_rate": 0.00022032755091046222,
-      "loss": 3.2681,
+      "grad_norm": 0.8025597929954529,
+      "learning_rate": 0.00022033401573106343,
+      "loss": 3.2693,
       "step": 58850
     },
     {
       "epoch": 6.339468302658487,
-      "grad_norm": 0.7325397729873657,
-      "learning_rate": 0.0002200043098804008,
-      "loss": 3.2774,
+      "grad_norm": 0.7327874898910522,
+      "learning_rate": 0.000220010774701002,
+      "loss": 3.2623,
       "step": 58900
     },
     {
       "epoch": 6.344849854698095,
-      "grad_norm": 0.7850790619850159,
-      "learning_rate": 0.00021968106885033938,
-      "loss": 3.2693,
+      "grad_norm": 0.7330926060676575,
+      "learning_rate": 0.00021968753367094062,
+      "loss": 3.2857,
       "step": 58950
     },
     {
       "epoch": 6.350231406737703,
-      "grad_norm": 0.7271026968955994,
-      "learning_rate": 0.00021935782782027798,
-      "loss": 3.2645,
+      "grad_norm": 0.7519079446792603,
+      "learning_rate": 0.0002193642926408792,
+      "loss": 3.2616,
       "step": 59000
     },
     {
       "epoch": 6.350231406737703,
-      "eval_accuracy": 0.384902529124135,
-      "eval_loss": 3.378819704055786,
-      "eval_runtime": 223.2378,
-      "eval_samples_per_second": 80.681,
-      "eval_steps_per_second": 5.044,
+      "eval_accuracy": 0.38532497155195766,
+      "eval_loss": 3.372903823852539,
+      "eval_runtime": 181.846,
+      "eval_samples_per_second": 99.045,
+      "eval_steps_per_second": 6.192,
       "step": 59000
     },
     {
       "epoch": 6.355612958777312,
-      "grad_norm": 0.7542627453804016,
-      "learning_rate": 0.00021903458679021655,
-      "loss": 3.2872,
+      "grad_norm": 0.7668281197547913,
+      "learning_rate": 0.00021904105161081778,
+      "loss": 3.2668,
       "step": 59050
     },
     {
       "epoch": 6.360994510816919,
-      "grad_norm": 0.7405682802200317,
-      "learning_rate": 0.0002187113457601551,
-      "loss": 3.266,
+      "grad_norm": 0.7630317211151123,
+      "learning_rate": 0.00021871781058075638,
+      "loss": 3.2703,
       "step": 59100
     },
     {
       "epoch": 6.366376062856528,
-      "grad_norm": 0.7460401654243469,
-      "learning_rate": 0.00021838810473009373,
-      "loss": 3.2759,
+      "grad_norm": 0.7405530214309692,
+      "learning_rate": 0.00021839456955069495,
+      "loss": 3.2873,
       "step": 59150
     },
     {
       "epoch": 6.371757614896136,
-      "grad_norm": 0.7826977372169495,
-      "learning_rate": 0.0002180648637000323,
-      "loss": 3.2753,
+      "grad_norm": 0.7828007936477661,
+      "learning_rate": 0.00021807132852063354,
+      "loss": 3.2628,
       "step": 59200
     },
     {
       "epoch": 6.377139166935744,
-      "grad_norm": 0.7382274270057678,
-      "learning_rate": 0.0002177416226699709,
-      "loss": 3.2868,
+      "grad_norm": 0.7717626690864563,
+      "learning_rate": 0.0002177480874905721,
+      "loss": 3.2743,
       "step": 59250
     },
     {
       "epoch": 6.382520718975353,
-      "grad_norm": 0.7684206962585449,
-      "learning_rate": 0.00021741838163990946,
-      "loss": 3.2841,
+      "grad_norm": 0.7008639574050903,
+      "learning_rate": 0.0002174248464605107,
+      "loss": 3.2804,
       "step": 59300
     },
     {
       "epoch": 6.387902271014961,
-      "grad_norm": 0.7826603055000305,
-      "learning_rate": 0.00021709514060984806,
-      "loss": 3.2647,
+      "grad_norm": 0.762617290019989,
+      "learning_rate": 0.0002171016054304493,
+      "loss": 3.2635,
       "step": 59350
     },
     {
       "epoch": 6.393283823054569,
-      "grad_norm": 0.760877788066864,
-      "learning_rate": 0.00021677189957978665,
-      "loss": 3.2792,
+      "grad_norm": 0.7396647930145264,
+      "learning_rate": 0.00021677836440038787,
+      "loss": 3.2816,
       "step": 59400
     },
     {
       "epoch": 6.398665375094177,
-      "grad_norm": 0.7618247866630554,
-      "learning_rate": 0.00021644865854972522,
-      "loss": 3.2909,
+      "grad_norm": 0.7557615637779236,
+      "learning_rate": 0.00021645512337032643,
+      "loss": 3.281,
       "step": 59450
     },
     {
       "epoch": 6.404046927133785,
-      "grad_norm": 0.9166077971458435,
-      "learning_rate": 0.0002161254175196638,
-      "loss": 3.2866,
+      "grad_norm": 0.9398220181465149,
+      "learning_rate": 0.00021613188234026506,
+      "loss": 3.29,
       "step": 59500
     },
     {
       "epoch": 6.4094284791733935,
-      "grad_norm": 0.764710009098053,
-      "learning_rate": 0.0002158021764896024,
-      "loss": 3.2599,
+      "grad_norm": 0.7071760892868042,
+      "learning_rate": 0.00021580864131020362,
+      "loss": 3.2918,
       "step": 59550
     },
     {
       "epoch": 6.414810031213002,
-      "grad_norm": 0.7640218734741211,
-      "learning_rate": 0.00021547893545954098,
-      "loss": 3.2676,
+      "grad_norm": 0.7526212334632874,
+      "learning_rate": 0.0002154854002801422,
+      "loss": 3.2764,
       "step": 59600
     },
     {
       "epoch": 6.42019158325261,
-      "grad_norm": 0.7581861615180969,
-      "learning_rate": 0.00021515569442947955,
-      "loss": 3.3134,
+      "grad_norm": 0.7543957829475403,
+      "learning_rate": 0.0002151621592500808,
+      "loss": 3.2705,
       "step": 59650
     },
     {
       "epoch": 6.425573135292218,
-      "grad_norm": 0.7885519862174988,
-      "learning_rate": 0.00021483245339941817,
-      "loss": 3.2905,
+      "grad_norm": 0.7652724385261536,
+      "learning_rate": 0.00021483891822001938,
+      "loss": 3.2656,
       "step": 59700
     },
     {
       "epoch": 6.430954687331827,
-      "grad_norm": 0.7909292578697205,
-      "learning_rate": 0.00021450921236935674,
-      "loss": 3.2853,
+      "grad_norm": 0.7655626535415649,
+      "learning_rate": 0.00021451567718995795,
+      "loss": 3.3013,
       "step": 59750
     },
     {
       "epoch": 6.4363362393714345,
-      "grad_norm": 0.7698927521705627,
-      "learning_rate": 0.0002141859713392953,
-      "loss": 3.2799,
+      "grad_norm": 0.7473166584968567,
+      "learning_rate": 0.00021419243615989654,
+      "loss": 3.2817,
       "step": 59800
     },
     {
       "epoch": 6.441717791411043,
-      "grad_norm": 0.7979402542114258,
-      "learning_rate": 0.0002138627303092339,
-      "loss": 3.3018,
+      "grad_norm": 0.7694827318191528,
+      "learning_rate": 0.00021386919512983514,
+      "loss": 3.2955,
       "step": 59850
     },
     {
       "epoch": 6.447099343450651,
-      "grad_norm": 0.7837560772895813,
-      "learning_rate": 0.0002135394892791725,
-      "loss": 3.2739,
+      "grad_norm": 0.7287412285804749,
+      "learning_rate": 0.0002135459540997737,
+      "loss": 3.2759,
       "step": 59900
     },
     {
       "epoch": 6.452480895490259,
-      "grad_norm": 0.7438477873802185,
-      "learning_rate": 0.00021321624824911106,
-      "loss": 3.2845,
+      "grad_norm": 0.7668284177780151,
+      "learning_rate": 0.0002132227130697123,
+      "loss": 3.2828,
       "step": 59950
     },
     {
       "epoch": 6.457862447529868,
-      "grad_norm": 0.7494663596153259,
-      "learning_rate": 0.00021289300721904965,
-      "loss": 3.2801,
+      "grad_norm": 0.731171190738678,
+      "learning_rate": 0.00021289947203965087,
+      "loss": 3.2897,
       "step": 60000
     },
     {
       "epoch": 6.457862447529868,
-      "eval_accuracy": 0.38531291108141025,
-      "eval_loss": 3.374072313308716,
-      "eval_runtime": 209.5252,
-      "eval_samples_per_second": 85.961,
-      "eval_steps_per_second": 5.374,
+      "eval_accuracy": 0.3859030049151307,
+      "eval_loss": 3.3710451126098633,
+      "eval_runtime": 181.3667,
+      "eval_samples_per_second": 99.307,
+      "eval_steps_per_second": 6.208,
       "step": 60000
     },
     {
       "epoch": 6.4632439995694755,
-      "grad_norm": 0.7274054288864136,
-      "learning_rate": 0.00021256976618898822,
-      "loss": 3.2882,
+      "grad_norm": 0.7099169492721558,
+      "learning_rate": 0.0002125762310095895,
+      "loss": 3.2892,
       "step": 60050
     },
     {
       "epoch": 6.468625551609084,
-      "grad_norm": 0.7407465577125549,
+      "grad_norm": 0.7312466502189636,
       "learning_rate": 0.00021225298997952806,
-      "loss": 3.2824,
+      "loss": 3.2709,
       "step": 60100
     },
     {
       "epoch": 6.474007103648693,
-      "grad_norm": 0.7144964933395386,
+      "grad_norm": 0.7752417325973511,
       "learning_rate": 0.00021192974894946662,
-      "loss": 3.2957,
+      "loss": 3.2954,
       "step": 60150
     },
     {
       "epoch": 6.4793886556883,
-      "grad_norm": 0.7461879849433899,
+      "grad_norm": 0.7163131833076477,
       "learning_rate": 0.00021160650791940524,
-      "loss": 3.2838,
+      "loss": 3.2764,
       "step": 60200
     },
     {
       "epoch": 6.484770207727909,
-      "grad_norm": 0.755248486995697,
+      "grad_norm": 0.7041307687759399,
       "learning_rate": 0.0002112832668893438,
-      "loss": 3.2862,
+      "loss": 3.2711,
       "step": 60250
     },
     {
       "epoch": 6.490151759767517,
-      "grad_norm": 0.7556350231170654,
+      "grad_norm": 0.7462310791015625,
       "learning_rate": 0.00021096002585928238,
-      "loss": 3.2679,
+      "loss": 3.2699,
       "step": 60300
     },
     {
       "epoch": 6.495533311807125,
-      "grad_norm": 0.7320284843444824,
+      "grad_norm": 0.7815905809402466,
       "learning_rate": 0.00021063678482922097,
-      "loss": 3.2899,
+      "loss": 3.2678,
       "step": 60350
     },
     {
       "epoch": 6.500914863846734,
-      "grad_norm": 0.7573622465133667,
+      "grad_norm": 0.7336490154266357,
       "learning_rate": 0.00021031354379915957,
-      "loss": 3.2869,
+      "loss": 3.2772,
       "step": 60400
     },
     {
       "epoch": 6.506296415886341,
-      "grad_norm": 0.7960693836212158,
+      "grad_norm": 0.7608310580253601,
       "learning_rate": 0.00020999030276909814,
-      "loss": 3.2784,
+      "loss": 3.2814,
       "step": 60450
     },
     {
       "epoch": 6.51167796792595,
-      "grad_norm": 0.8241477608680725,
+      "grad_norm": 0.7783084511756897,
       "learning_rate": 0.00020966706173903673,
-      "loss": 3.2832,
+      "loss": 3.2857,
       "step": 60500
     },
     {
       "epoch": 6.517059519965558,
-      "grad_norm": 0.773271381855011,
+      "grad_norm": 0.7255032658576965,
       "learning_rate": 0.0002093438207089753,
-      "loss": 3.2755,
+      "loss": 3.266,
       "step": 60550
     },
     {
       "epoch": 6.522441072005166,
-      "grad_norm": 0.7589271664619446,
+      "grad_norm": 0.7634445428848267,
       "learning_rate": 0.00020902057967891387,
-      "loss": 3.2864,
+      "loss": 3.2791,
       "step": 60600
     },
     {
       "epoch": 6.5278226240447745,
-      "grad_norm": 0.735211968421936,
+      "grad_norm": 0.7686166763305664,
       "learning_rate": 0.0002086973386488525,
-      "loss": 3.2909,
+      "loss": 3.2847,
       "step": 60650
     },
     {
       "epoch": 6.533204176084383,
-      "grad_norm": 0.767329752445221,
+      "grad_norm": 0.8355399966239929,
       "learning_rate": 0.00020837409761879106,
-      "loss": 3.2849,
+      "loss": 3.2915,
       "step": 60700
     },
     {
       "epoch": 6.538585728123991,
-      "grad_norm": 0.7533854842185974,
+      "grad_norm": 0.7668452262878418,
       "learning_rate": 0.00020805085658872962,
-      "loss": 3.2829,
+      "loss": 3.2779,
       "step": 60750
     },
     {
       "epoch": 6.543967280163599,
-      "grad_norm": 0.8554604649543762,
+      "grad_norm": 0.7561519742012024,
       "learning_rate": 0.00020772761555866825,
-      "loss": 3.284,
+      "loss": 3.272,
       "step": 60800
     },
     {
       "epoch": 6.549348832203208,
-      "grad_norm": 0.7716811895370483,
+      "grad_norm": 0.7628666758537292,
       "learning_rate": 0.0002074043745286068,
-      "loss": 3.2828,
+      "loss": 3.2855,
       "step": 60850
     },
     {
       "epoch": 6.5547303842428155,
-      "grad_norm": 0.7305426597595215,
+      "grad_norm": 0.71573406457901,
       "learning_rate": 0.00020708113349854538,
-      "loss": 3.3054,
+      "loss": 3.2881,
       "step": 60900
     },
     {
       "epoch": 6.560111936282424,
-      "grad_norm": 0.8045239448547363,
+      "grad_norm": 0.7471049427986145,
       "learning_rate": 0.00020675789246848397,
-      "loss": 3.2869,
+      "loss": 3.2862,
       "step": 60950
     },
     {
       "epoch": 6.565493488322032,
-      "grad_norm": 0.7284274697303772,
+      "grad_norm": 0.7021371722221375,
       "learning_rate": 0.00020643465143842257,
-      "loss": 3.2836,
+      "loss": 3.2636,
       "step": 61000
     },
     {
       "epoch": 6.565493488322032,
-      "eval_accuracy": 0.38564528026523476,
-      "eval_loss": 3.370778799057007,
-      "eval_runtime": 201.2966,
-      "eval_samples_per_second": 89.475,
-      "eval_steps_per_second": 5.594,
+      "eval_accuracy": 0.3862979581623363,
+      "eval_loss": 3.3656346797943115,
+      "eval_runtime": 181.6152,
+      "eval_samples_per_second": 99.171,
+      "eval_steps_per_second": 6.2,
       "step": 61000
     },
     {
       "epoch": 6.57087504036164,
-      "grad_norm": 0.7565922737121582,
+      "grad_norm": 0.7603363990783691,
       "learning_rate": 0.00020611141040836116,
-      "loss": 3.285,
+      "loss": 3.2882,
       "step": 61050
     },
     {
       "epoch": 6.576256592401249,
-      "grad_norm": 0.8001964092254639,
+      "grad_norm": 0.6989268660545349,
       "learning_rate": 0.00020578816937829973,
-      "loss": 3.2914,
+      "loss": 3.2779,
       "step": 61100
     },
     {
       "epoch": 6.5816381444408565,
-      "grad_norm": 0.7406817078590393,
+      "grad_norm": 0.7628535628318787,
       "learning_rate": 0.0002054649283482383,
-      "loss": 3.2947,
+      "loss": 3.2753,
       "step": 61150
     },
     {
       "epoch": 6.587019696480465,
-      "grad_norm": 0.7439241409301758,
+      "grad_norm": 0.767864465713501,
       "learning_rate": 0.00020514168731817692,
-      "loss": 3.2742,
+      "loss": 3.2874,
       "step": 61200
     },
     {
       "epoch": 6.592401248520073,
-      "grad_norm": 0.7465095520019531,
+      "grad_norm": 0.7803367972373962,
       "learning_rate": 0.0002048184462881155,
-      "loss": 3.2787,
+      "loss": 3.2884,
       "step": 61250
     },
     {
       "epoch": 6.597782800559681,
-      "grad_norm": 0.7967016100883484,
+      "grad_norm": 0.8632737994194031,
       "learning_rate": 0.00020449520525805406,
-      "loss": 3.2923,
+      "loss": 3.2735,
       "step": 61300
     },
     {
       "epoch": 6.60316435259929,
-      "grad_norm": 0.7551931142807007,
+      "grad_norm": 0.7356277704238892,
       "learning_rate": 0.00020417196422799268,
-      "loss": 3.2902,
+      "loss": 3.284,
       "step": 61350
     },
     {
       "epoch": 6.608545904638898,
-      "grad_norm": 0.7699846029281616,
+      "grad_norm": 0.7249445915222168,
       "learning_rate": 0.00020384872319793125,
-      "loss": 3.2934,
+      "loss": 3.2915,
       "step": 61400
     },
     {
       "epoch": 6.613927456678506,
-      "grad_norm": 0.7986684441566467,
+      "grad_norm": 0.7576565742492676,
       "learning_rate": 0.0002035254821678698,
-      "loss": 3.2858,
+      "loss": 3.2708,
       "step": 61450
     },
     {
       "epoch": 6.619309008718115,
-      "grad_norm": 0.7962992787361145,
+      "grad_norm": 0.7591082453727722,
       "learning_rate": 0.0002032022411378084,
-      "loss": 3.2827,
+      "loss": 3.2874,
       "step": 61500
     },
     {
       "epoch": 6.624690560757722,
-      "grad_norm": 0.7751727104187012,
+      "grad_norm": 0.7346405982971191,
       "learning_rate": 0.000202879000107747,
-      "loss": 3.2885,
+      "loss": 3.2892,
       "step": 61550
     },
     {
       "epoch": 6.630072112797331,
-      "grad_norm": 0.7956772446632385,
+      "grad_norm": 0.803242564201355,
       "learning_rate": 0.00020255575907768557,
-      "loss": 3.3027,
+      "loss": 3.2873,
       "step": 61600
     },
     {
       "epoch": 6.635453664836939,
-      "grad_norm": 0.7654341459274292,
+      "grad_norm": 0.7800670266151428,
       "learning_rate": 0.00020223251804762416,
-      "loss": 3.2883,
+      "loss": 3.2796,
       "step": 61650
     },
     {
       "epoch": 6.640835216876547,
-      "grad_norm": 0.7234878540039062,
+      "grad_norm": 0.7379322648048401,
       "learning_rate": 0.00020190927701756273,
-      "loss": 3.3,
+      "loss": 3.289,
       "step": 61700
     },
     {
       "epoch": 6.6462167689161555,
-      "grad_norm": 0.7403562068939209,
+      "grad_norm": 0.7375909686088562,
       "learning_rate": 0.00020158603598750133,
-      "loss": 3.2763,
+      "loss": 3.2602,
       "step": 61750
     },
     {
       "epoch": 6.651598320955763,
-      "grad_norm": 0.7899567484855652,
+      "grad_norm": 0.7885672450065613,
       "learning_rate": 0.00020126279495743992,
-      "loss": 3.2967,
+      "loss": 3.2871,
       "step": 61800
     },
     {
       "epoch": 6.656979872995372,
-      "grad_norm": 0.7615472674369812,
+      "grad_norm": 0.7855244874954224,
       "learning_rate": 0.0002009395539273785,
-      "loss": 3.2782,
+      "loss": 3.2683,
       "step": 61850
     },
     {
       "epoch": 6.66236142503498,
-      "grad_norm": 0.7665838599205017,
+      "grad_norm": 0.7652177214622498,
       "learning_rate": 0.00020061631289731706,
-      "loss": 3.2865,
+      "loss": 3.3011,
       "step": 61900
     },
     {
       "epoch": 6.667742977074588,
-      "grad_norm": 0.769633948802948,
+      "grad_norm": 0.7672140598297119,
       "learning_rate": 0.00020029307186725568,
-      "loss": 3.3017,
+      "loss": 3.2708,
       "step": 61950
     },
     {
       "epoch": 6.6731245291141965,
-      "grad_norm": 0.7775278687477112,
+      "grad_norm": 0.7493014931678772,
       "learning_rate": 0.00019996983083719425,
-      "loss": 3.2808,
+      "loss": 3.2753,
       "step": 62000
     },
     {
       "epoch": 6.6731245291141965,
-      "eval_accuracy": 0.3862611248333672,
-      "eval_loss": 3.3647584915161133,
-      "eval_runtime": 210.4271,
-      "eval_samples_per_second": 85.593,
-      "eval_steps_per_second": 5.351,
+      "eval_accuracy": 0.38660707562816854,
+      "eval_loss": 3.36199688911438,
+      "eval_runtime": 181.6046,
+      "eval_samples_per_second": 99.177,
+      "eval_steps_per_second": 6.2,
       "step": 62000
     },
     {
       "epoch": 6.678506081153805,
-      "grad_norm": 0.7849041223526001,
+      "grad_norm": 0.786527693271637,
       "learning_rate": 0.00019964658980713284,
-      "loss": 3.2781,
+      "loss": 3.2959,
       "step": 62050
     },
     {
       "epoch": 6.683887633193413,
-      "grad_norm": 0.8136065602302551,
+      "grad_norm": 0.784377932548523,
       "learning_rate": 0.00019932334877707144,
-      "loss": 3.3005,
+      "loss": 3.3004,
       "step": 62100
     },
     {
       "epoch": 6.689269185233021,
-      "grad_norm": 0.7949984073638916,
-      "learning_rate": 0.00019900657256761124,
-      "loss": 3.2789,
+      "grad_norm": 0.7290779948234558,
+      "learning_rate": 0.00019900010774701,
+      "loss": 3.2987,
       "step": 62150
     },
     {
       "epoch": 6.69465073727263,
-      "grad_norm": 0.7674379348754883,
-      "learning_rate": 0.0001986833315375498,
-      "loss": 3.2832,
+      "grad_norm": 0.7574355602264404,
+      "learning_rate": 0.0001986768667169486,
+      "loss": 3.2978,
       "step": 62200
     },
     {
       "epoch": 6.7000322893122375,
-      "grad_norm": 0.8072303533554077,
-      "learning_rate": 0.00019836009050748838,
-      "loss": 3.2916,
+      "grad_norm": 0.7647578120231628,
+      "learning_rate": 0.00019835362568688716,
+      "loss": 3.2844,
       "step": 62250
     },
     {
       "epoch": 6.705413841351846,
-      "grad_norm": 0.7814190983772278,
-      "learning_rate": 0.000198036849477427,
-      "loss": 3.3209,
+      "grad_norm": 0.7932471036911011,
+      "learning_rate": 0.00019803038465682576,
+      "loss": 3.2911,
       "step": 62300
     },
     {
       "epoch": 6.710795393391454,
-      "grad_norm": 0.788261890411377,
-      "learning_rate": 0.00019772007326796678,
-      "loss": 3.2981,
+      "grad_norm": 0.7346493005752563,
+      "learning_rate": 0.00019770714362676435,
+      "loss": 3.2736,
       "step": 62350
     },
     {
       "epoch": 6.716176945431062,
-      "grad_norm": 0.7617114186286926,
-      "learning_rate": 0.0001973968322379054,
-      "loss": 3.2803,
+      "grad_norm": 0.7071163654327393,
+      "learning_rate": 0.00019738390259670292,
+      "loss": 3.2841,
       "step": 62400
     },
     {
       "epoch": 6.721558497470671,
-      "grad_norm": 0.7262803316116333,
-      "learning_rate": 0.00019707359120784397,
-      "loss": 3.2805,
+      "grad_norm": 0.7414072155952454,
+      "learning_rate": 0.0001970606615666415,
+      "loss": 3.2847,
       "step": 62450
     },
     {
       "epoch": 6.7269400495102785,
-      "grad_norm": 0.7623103260993958,
-      "learning_rate": 0.00019675035017778253,
-      "loss": 3.2771,
+      "grad_norm": 0.7445050477981567,
+      "learning_rate": 0.0001967374205365801,
+      "loss": 3.2787,
       "step": 62500
     },
     {
       "epoch": 6.732321601549887,
-      "grad_norm": 0.7562897801399231,
-      "learning_rate": 0.00019642710914772113,
-      "loss": 3.292,
+      "grad_norm": 0.7484267354011536,
+      "learning_rate": 0.00019641417950651868,
+      "loss": 3.2728,
       "step": 62550
     },
     {
       "epoch": 6.737703153589496,
-      "grad_norm": 0.7772451043128967,
-      "learning_rate": 0.00019610386811765972,
-      "loss": 3.2792,
+      "grad_norm": 0.707564651966095,
+      "learning_rate": 0.00019609093847645725,
+      "loss": 3.2817,
       "step": 62600
     },
     {
       "epoch": 6.743084705629103,
-      "grad_norm": 0.8128804564476013,
-      "learning_rate": 0.00019578062708759832,
-      "loss": 3.3054,
+      "grad_norm": 0.7779157757759094,
+      "learning_rate": 0.00019576769744639587,
+      "loss": 3.3052,
       "step": 62650
     },
     {
       "epoch": 6.748466257668712,
-      "grad_norm": 0.8137348294258118,
-      "learning_rate": 0.00019545738605753689,
-      "loss": 3.2941,
+      "grad_norm": 0.7554226517677307,
+      "learning_rate": 0.00019544445641633444,
+      "loss": 3.2896,
       "step": 62700
     },
     {
       "epoch": 6.75384780970832,
-      "grad_norm": 0.7956899404525757,
-      "learning_rate": 0.00019513414502747545,
-      "loss": 3.2922,
+      "grad_norm": 0.7871652841567993,
+      "learning_rate": 0.000195121215386273,
+      "loss": 3.28,
       "step": 62750
     },
     {
       "epoch": 6.759229361747928,
-      "grad_norm": 0.7582367062568665,
-      "learning_rate": 0.00019481090399741408,
-      "loss": 3.2699,
+      "grad_norm": 0.8141342997550964,
+      "learning_rate": 0.0001947979743562116,
+      "loss": 3.2777,
       "step": 62800
     },
     {
       "epoch": 6.7646109137875365,
-      "grad_norm": 1.1021339893341064,
-      "learning_rate": 0.00019448766296735264,
-      "loss": 3.2984,
+      "grad_norm": 0.748104453086853,
+      "learning_rate": 0.00019448119814675143,
+      "loss": 3.2805,
       "step": 62850
     },
     {
       "epoch": 6.769992465827144,
-      "grad_norm": 0.8446829319000244,
-      "learning_rate": 0.0001941644219372912,
-      "loss": 3.2944,
+      "grad_norm": 0.8079434633255005,
+      "learning_rate": 0.00019415795711669,
+      "loss": 3.2851,
       "step": 62900
     },
     {
       "epoch": 6.775374017866753,
-      "grad_norm": 0.7830290794372559,
-      "learning_rate": 0.00019384118090722983,
-      "loss": 3.2988,
+      "grad_norm": 0.787795901298523,
+      "learning_rate": 0.00019383471608662857,
+      "loss": 3.2978,
       "step": 62950
     },
     {
       "epoch": 6.780755569906361,
-      "grad_norm": 0.7751957178115845,
-      "learning_rate": 0.0001935179398771684,
-      "loss": 3.2776,
+      "grad_norm": 0.7400677800178528,
+      "learning_rate": 0.0001935114750565672,
+      "loss": 3.2841,
       "step": 63000
     },
     {
       "epoch": 6.780755569906361,
-      "eval_accuracy": 0.38648049501386467,
-      "eval_loss": 3.3618505001068115,
-      "eval_runtime": 180.2451,
-      "eval_samples_per_second": 99.925,
-      "eval_steps_per_second": 6.247,
+      "eval_accuracy": 0.3866848710958437,
+      "eval_loss": 3.358719825744629,
+      "eval_runtime": 181.8686,
+      "eval_samples_per_second": 99.033,
+      "eval_steps_per_second": 6.191,
       "step": 63000
     },
     {
       "epoch": 6.786137121945969,
-      "grad_norm": 0.8304117321968079,
-      "learning_rate": 0.00019319469884710697,
-      "loss": 3.2927,
+      "grad_norm": 0.7799237966537476,
+      "learning_rate": 0.00019318823402650576,
+      "loss": 3.2945,
       "step": 63050
     },
     {
       "epoch": 6.7915186739855775,
-      "grad_norm": 0.7430444955825806,
-      "learning_rate": 0.00019287145781704556,
-      "loss": 3.2934,
+      "grad_norm": 0.7586919069290161,
+      "learning_rate": 0.00019286499299644432,
+      "loss": 3.2758,
       "step": 63100
     },
     {
       "epoch": 6.796900226025185,
-      "grad_norm": 0.756223738193512,
-      "learning_rate": 0.00019254821678698416,
-      "loss": 3.2953,
+      "grad_norm": 0.7611883282661438,
+      "learning_rate": 0.00019254175196638292,
+      "loss": 3.2795,
       "step": 63150
     },
     {
       "epoch": 6.802281778064794,
-      "grad_norm": 0.7750756144523621,
-      "learning_rate": 0.00019222497575692272,
-      "loss": 3.2911,
+      "grad_norm": 0.7690772414207458,
+      "learning_rate": 0.0001922185109363215,
+      "loss": 3.2882,
       "step": 63200
     },
     {
       "epoch": 6.807663330104402,
-      "grad_norm": 0.7659454941749573,
-      "learning_rate": 0.00019190173472686132,
-      "loss": 3.2952,
+      "grad_norm": 0.7428942322731018,
+      "learning_rate": 0.00019189526990626008,
+      "loss": 3.2859,
       "step": 63250
     },
     {
       "epoch": 6.813044882144011,
-      "grad_norm": 0.7663837671279907,
-      "learning_rate": 0.0001915784936967999,
-      "loss": 3.2953,
+      "grad_norm": 0.7428345680236816,
+      "learning_rate": 0.00019157202887619867,
+      "loss": 3.2726,
       "step": 63300
     },
     {
       "epoch": 6.8184264341836185,
-      "grad_norm": 0.7659280300140381,
-      "learning_rate": 0.00019125525266673845,
-      "loss": 3.2848,
+      "grad_norm": 0.7654890418052673,
+      "learning_rate": 0.00019124878784613724,
+      "loss": 3.278,
       "step": 63350
     },
     {
       "epoch": 6.823807986223227,
-      "grad_norm": 0.8383036255836487,
-      "learning_rate": 0.00019093201163667708,
-      "loss": 3.2811,
+      "grad_norm": 0.7603652477264404,
+      "learning_rate": 0.00019092554681607584,
+      "loss": 3.2923,
       "step": 63400
     },
     {
       "epoch": 6.829189538262835,
-      "grad_norm": 0.7594733238220215,
-      "learning_rate": 0.00019060877060661564,
-      "loss": 3.2928,
+      "grad_norm": 0.8035446405410767,
+      "learning_rate": 0.00019060230578601443,
+      "loss": 3.2921,
       "step": 63450
     },
     {
       "epoch": 6.834571090302443,
-      "grad_norm": 0.7703410983085632,
-      "learning_rate": 0.0001902855295765542,
-      "loss": 3.2949,
+      "grad_norm": 0.727232813835144,
+      "learning_rate": 0.000190279064755953,
+      "loss": 3.2733,
       "step": 63500
     },
     {
       "epoch": 6.839952642342052,
-      "grad_norm": 0.7872480154037476,
-      "learning_rate": 0.00018996228854649283,
-      "loss": 3.2945,
+      "grad_norm": 0.7339451909065247,
+      "learning_rate": 0.00018995582372589157,
+      "loss": 3.2796,
       "step": 63550
     },
     {
       "epoch": 6.8453341943816595,
-      "grad_norm": 0.7895832061767578,
-      "learning_rate": 0.0001896390475164314,
-      "loss": 3.2881,
+      "grad_norm": 0.7597670555114746,
+      "learning_rate": 0.0001896325826958302,
+      "loss": 3.2764,
       "step": 63600
     },
     {
       "epoch": 6.850715746421268,
-      "grad_norm": 0.7981587052345276,
-      "learning_rate": 0.00018931580648637,
-      "loss": 3.2887,
+      "grad_norm": 0.7778809070587158,
+      "learning_rate": 0.00018930934166576876,
+      "loss": 3.2994,
       "step": 63650
     },
     {
       "epoch": 6.856097298460876,
-      "grad_norm": 0.7759888172149658,
-      "learning_rate": 0.00018899256545630856,
-      "loss": 3.3,
+      "grad_norm": 0.7657696604728699,
+      "learning_rate": 0.00018898610063570732,
+      "loss": 3.2802,
       "step": 63700
     },
     {
       "epoch": 6.861478850500484,
-      "grad_norm": 0.7870774269104004,
-      "learning_rate": 0.00018866932442624716,
-      "loss": 3.2849,
+      "grad_norm": 0.7392223477363586,
+      "learning_rate": 0.00018866285960564595,
+      "loss": 3.2892,
       "step": 63750
     },
     {
       "epoch": 6.866860402540093,
-      "grad_norm": 0.8005082011222839,
-      "learning_rate": 0.00018834608339618575,
-      "loss": 3.2913,
+      "grad_norm": 0.7388136386871338,
+      "learning_rate": 0.0001883396185755845,
+      "loss": 3.2787,
       "step": 63800
     },
     {
       "epoch": 6.8722419545797,
-      "grad_norm": 0.8050322532653809,
-      "learning_rate": 0.00018802284236612432,
-      "loss": 3.295,
+      "grad_norm": 0.7497984170913696,
+      "learning_rate": 0.0001880163775455231,
+      "loss": 3.2846,
       "step": 63850
     },
     {
       "epoch": 6.877623506619309,
-      "grad_norm": 0.8603694438934326,
-      "learning_rate": 0.0001876996013360629,
-      "loss": 3.2935,
+      "grad_norm": 0.7447848320007324,
+      "learning_rate": 0.00018769313651546168,
+      "loss": 3.2838,
       "step": 63900
     },
     {
       "epoch": 6.8830050586589175,
-      "grad_norm": 0.8038960695266724,
-      "learning_rate": 0.0001873763603060015,
-      "loss": 3.2684,
+      "grad_norm": 0.780323326587677,
+      "learning_rate": 0.00018736989548540027,
+      "loss": 3.2782,
       "step": 63950
     },
     {
       "epoch": 6.888386610698525,
-      "grad_norm": 0.8121204376220703,
-      "learning_rate": 0.00018705311927594008,
-      "loss": 3.2847,
+      "grad_norm": 0.8068848848342896,
+      "learning_rate": 0.00018704665445533886,
+      "loss": 3.273,
       "step": 64000
     },
     {
       "epoch": 6.888386610698525,
-      "eval_accuracy": 0.3866242427844432,
-      "eval_loss": 3.361416816711426,
-      "eval_runtime": 180.531,
-      "eval_samples_per_second": 99.767,
-      "eval_steps_per_second": 6.237,
+      "eval_accuracy": 0.3873136453576261,
+      "eval_loss": 3.3533878326416016,
+      "eval_runtime": 181.6713,
+      "eval_samples_per_second": 99.141,
+      "eval_steps_per_second": 6.198,
       "step": 64000
     },
     {
       "epoch": 6.893768162738134,
-      "grad_norm": 0.8602204322814941,
-      "learning_rate": 0.00018672987824587864,
-      "loss": 3.3055,
+      "grad_norm": 0.8123435974121094,
+      "learning_rate": 0.00018672341342527743,
+      "loss": 3.2825,
       "step": 64050
     },
     {
       "epoch": 6.899149714777742,
-      "grad_norm": 0.7669270634651184,
-      "learning_rate": 0.00018640663721581727,
-      "loss": 3.2954,
+      "grad_norm": 0.7806638479232788,
+      "learning_rate": 0.000186400172395216,
+      "loss": 3.2812,
       "step": 64100
     },
     {
       "epoch": 6.90453126681735,
-      "grad_norm": 0.7600193619728088,
-      "learning_rate": 0.00018608339618575583,
-      "loss": 3.2762,
+      "grad_norm": 0.7815253138542175,
+      "learning_rate": 0.00018607693136515462,
+      "loss": 3.2877,
       "step": 64150
     },
     {
       "epoch": 6.9099128188569585,
-      "grad_norm": 0.7429041862487793,
-      "learning_rate": 0.0001857601551556944,
-      "loss": 3.3104,
+      "grad_norm": 0.7817805409431458,
+      "learning_rate": 0.0001857536903350932,
+      "loss": 3.3011,
       "step": 64200
     },
     {
       "epoch": 6.915294370896566,
-      "grad_norm": 0.8116273283958435,
-      "learning_rate": 0.000185436914125633,
-      "loss": 3.2847,
+      "grad_norm": 0.7825344204902649,
+      "learning_rate": 0.00018543044930503176,
+      "loss": 3.287,
       "step": 64250
     },
     {
       "epoch": 6.920675922936175,
-      "grad_norm": 0.7676055431365967,
-      "learning_rate": 0.0001851136730955716,
-      "loss": 3.2907,
+      "grad_norm": 0.7773126363754272,
+      "learning_rate": 0.00018510720827497035,
+      "loss": 3.3033,
       "step": 64300
     },
     {
       "epoch": 6.926057474975783,
-      "grad_norm": 0.7990931272506714,
-      "learning_rate": 0.00018479043206551016,
-      "loss": 3.2927,
+      "grad_norm": 0.7567140460014343,
+      "learning_rate": 0.00018478396724490895,
+      "loss": 3.2814,
       "step": 64350
     },
     {
       "epoch": 6.931439027015391,
-      "grad_norm": 0.7669125199317932,
-      "learning_rate": 0.00018446719103544875,
-      "loss": 3.2772,
+      "grad_norm": 0.7970994114875793,
+      "learning_rate": 0.0001844607262148475,
+      "loss": 3.2942,
       "step": 64400
     },
     {
       "epoch": 6.9368205790549995,
-      "grad_norm": 0.8254827260971069,
-      "learning_rate": 0.00018414395000538732,
-      "loss": 3.2951,
+      "grad_norm": 0.8008365631103516,
+      "learning_rate": 0.0001841374851847861,
+      "loss": 3.2856,
       "step": 64450
     },
     {
       "epoch": 6.942202131094608,
-      "grad_norm": 0.7549811601638794,
-      "learning_rate": 0.00018382070897532591,
-      "loss": 3.29,
+      "grad_norm": 0.7467749118804932,
+      "learning_rate": 0.00018381424415472468,
+      "loss": 3.2803,
       "step": 64500
     },
     {
       "epoch": 6.947583683134216,
-      "grad_norm": 0.8003718256950378,
-      "learning_rate": 0.0001834974679452645,
-      "loss": 3.2858,
+      "grad_norm": 0.7841525673866272,
+      "learning_rate": 0.00018349100312466327,
+      "loss": 3.299,
       "step": 64550
     },
     {
       "epoch": 6.952965235173824,
-      "grad_norm": 0.8304224014282227,
-      "learning_rate": 0.00018318069173580434,
-      "loss": 3.2969,
+      "grad_norm": 0.7758609056472778,
+      "learning_rate": 0.00018316776209460186,
+      "loss": 3.2884,
       "step": 64600
     },
     {
       "epoch": 6.958346787213433,
-      "grad_norm": 0.7723833322525024,
-      "learning_rate": 0.0001828574507057429,
-      "loss": 3.2879,
+      "grad_norm": 0.7331240773200989,
+      "learning_rate": 0.00018284452106454043,
+      "loss": 3.2807,
       "step": 64650
     },
     {
       "epoch": 6.9637283392530405,
-      "grad_norm": 0.7792532444000244,
-      "learning_rate": 0.00018253420967568148,
-      "loss": 3.2591,
+      "grad_norm": 0.7761428952217102,
+      "learning_rate": 0.000182521280034479,
+      "loss": 3.2592,
       "step": 64700
     },
     {
       "epoch": 6.969109891292649,
-      "grad_norm": 0.8019757270812988,
-      "learning_rate": 0.00018221096864562007,
-      "loss": 3.2834,
+      "grad_norm": 0.7510924339294434,
+      "learning_rate": 0.00018219803900441762,
+      "loss": 3.2775,
       "step": 64750
     },
     {
       "epoch": 6.974491443332257,
-      "grad_norm": 0.7884219884872437,
-      "learning_rate": 0.00018188772761555867,
-      "loss": 3.2763,
+      "grad_norm": 0.7802988886833191,
+      "learning_rate": 0.00018188126279495743,
+      "loss": 3.2852,
       "step": 64800
     },
     {
       "epoch": 6.979872995371865,
-      "grad_norm": 0.7742852568626404,
-      "learning_rate": 0.00018156448658549723,
-      "loss": 3.2954,
+      "grad_norm": 0.7580111026763916,
+      "learning_rate": 0.00018155802176489602,
+      "loss": 3.2932,
       "step": 64850
     },
     {
       "epoch": 6.985254547411474,
-      "grad_norm": 0.7695539593696594,
-      "learning_rate": 0.00018124124555543583,
-      "loss": 3.2884,
+      "grad_norm": 0.7831270694732666,
+      "learning_rate": 0.0001812347807348346,
+      "loss": 3.2885,
       "step": 64900
     },
     {
       "epoch": 6.990636099451081,
-      "grad_norm": 0.8002060055732727,
-      "learning_rate": 0.0001809180045253744,
-      "loss": 3.2998,
+      "grad_norm": 0.8049140572547913,
+      "learning_rate": 0.00018091153970477319,
+      "loss": 3.2837,
       "step": 64950
     },
     {
       "epoch": 6.99601765149069,
-      "grad_norm": 0.7580960392951965,
-      "learning_rate": 0.00018059476349531296,
-      "loss": 3.2856,
+      "grad_norm": 0.7779400944709778,
+      "learning_rate": 0.00018058829867471175,
+      "loss": 3.2759,
       "step": 65000
     },
     {
       "epoch": 6.99601765149069,
-      "eval_accuracy": 0.3874277308898313,
-      "eval_loss": 3.353395938873291,
-      "eval_runtime": 180.2586,
-      "eval_samples_per_second": 99.918,
-      "eval_steps_per_second": 6.247,
+      "eval_accuracy": 0.3876208070714776,
+      "eval_loss": 3.350797176361084,
+      "eval_runtime": 181.5152,
+      "eval_samples_per_second": 99.226,
+      "eval_steps_per_second": 6.203,
       "step": 65000
     },
     {
       "epoch": 7.0013992035302985,
-      "grad_norm": 0.7535995244979858,
-      "learning_rate": 0.00018027152246525159,
-      "loss": 3.2655,
+      "grad_norm": 0.7582424283027649,
+      "learning_rate": 0.00018026505764465035,
+      "loss": 3.2703,
       "step": 65050
     },
     {
       "epoch": 7.006780755569906,
-      "grad_norm": 0.7798009514808655,
-      "learning_rate": 0.00017994828143519015,
-      "loss": 3.2142,
+      "grad_norm": 0.8157245516777039,
+      "learning_rate": 0.00017994181661458894,
+      "loss": 3.176,
       "step": 65100
     },
     {
       "epoch": 7.012162307609515,
-      "grad_norm": 0.7834371328353882,
-      "learning_rate": 0.00017962504040512872,
-      "loss": 3.1959,
+      "grad_norm": 0.7752619981765747,
+      "learning_rate": 0.0001796185755845275,
+      "loss": 3.2072,
       "step": 65150
     },
     {
       "epoch": 7.017543859649122,
-      "grad_norm": 0.7991567850112915,
-      "learning_rate": 0.00017930179937506734,
-      "loss": 3.1903,
+      "grad_norm": 0.7710782885551453,
+      "learning_rate": 0.00017929533455446608,
+      "loss": 3.1918,
       "step": 65200
     },
     {
       "epoch": 7.022925411688731,
-      "grad_norm": 0.8215781450271606,
-      "learning_rate": 0.0001789785583450059,
-      "loss": 3.2197,
+      "grad_norm": 0.7788631319999695,
+      "learning_rate": 0.0001789720935244047,
+      "loss": 3.2232,
       "step": 65250
     },
     {
       "epoch": 7.0283069637283395,
-      "grad_norm": 0.7673898935317993,
-      "learning_rate": 0.00017865531731494448,
-      "loss": 3.198,
+      "grad_norm": 0.7731738090515137,
+      "learning_rate": 0.00017864885249434327,
+      "loss": 3.1962,
       "step": 65300
     },
     {
       "epoch": 7.033688515767947,
-      "grad_norm": 0.7570087313652039,
-      "learning_rate": 0.00017833207628488307,
-      "loss": 3.1956,
+      "grad_norm": 0.768539547920227,
+      "learning_rate": 0.00017832561146428183,
+      "loss": 3.1887,
       "step": 65350
     },
     {
       "epoch": 7.039070067807556,
-      "grad_norm": 0.7777502536773682,
-      "learning_rate": 0.00017800883525482167,
-      "loss": 3.2017,
+      "grad_norm": 0.7669661045074463,
+      "learning_rate": 0.00017800237043422046,
+      "loss": 3.1966,
       "step": 65400
     },
     {
       "epoch": 7.044451619847164,
-      "grad_norm": 0.8165860176086426,
-      "learning_rate": 0.00017768559422476026,
-      "loss": 3.207,
+      "grad_norm": 0.7783058881759644,
+      "learning_rate": 0.00017767912940415902,
+      "loss": 3.1895,
       "step": 65450
     },
     {
       "epoch": 7.049833171886772,
-      "grad_norm": 0.8510911464691162,
-      "learning_rate": 0.00017736235319469883,
-      "loss": 3.2091,
+      "grad_norm": 0.7638408541679382,
+      "learning_rate": 0.0001773558883740976,
+      "loss": 3.2074,
       "step": 65500
     },
     {
       "epoch": 7.0552147239263805,
-      "grad_norm": 0.7689694762229919,
-      "learning_rate": 0.0001770391121646374,
-      "loss": 3.2019,
+      "grad_norm": 0.7813393473625183,
+      "learning_rate": 0.00017703264734403619,
+      "loss": 3.212,
       "step": 65550
     },
     {
       "epoch": 7.060596275965988,
-      "grad_norm": 0.7681092023849487,
-      "learning_rate": 0.00017671587113457602,
-      "loss": 3.2155,
+      "grad_norm": 0.7785745859146118,
+      "learning_rate": 0.00017670940631397475,
+      "loss": 3.2032,
       "step": 65600
     },
     {
       "epoch": 7.065977828005597,
-      "grad_norm": 0.7637883424758911,
-      "learning_rate": 0.0001763926301045146,
-      "loss": 3.2149,
+      "grad_norm": 0.7837199568748474,
+      "learning_rate": 0.00017638616528391337,
+      "loss": 3.1961,
       "step": 65650
     },
     {
       "epoch": 7.071359380045205,
-      "grad_norm": 0.810635507106781,
-      "learning_rate": 0.00017606938907445315,
-      "loss": 3.2153,
+      "grad_norm": 0.7816468477249146,
+      "learning_rate": 0.00017606292425385194,
+      "loss": 3.2053,
       "step": 65700
     },
     {
       "epoch": 7.076740932084813,
-      "grad_norm": 0.7918748259544373,
-      "learning_rate": 0.00017574614804439178,
-      "loss": 3.2095,
+      "grad_norm": 0.7704637050628662,
+      "learning_rate": 0.0001757396832237905,
+      "loss": 3.2102,
       "step": 65750
     },
     {
       "epoch": 7.0821224841244215,
-      "grad_norm": 0.8016531467437744,
-      "learning_rate": 0.00017542290701433034,
-      "loss": 3.211,
+      "grad_norm": 0.7501810193061829,
+      "learning_rate": 0.00017541644219372913,
+      "loss": 3.2083,
       "step": 65800
     },
     {
       "epoch": 7.08750403616403,
-      "grad_norm": 0.8359585404396057,
-      "learning_rate": 0.0001750996659842689,
-      "loss": 3.2128,
+      "grad_norm": 0.7586031556129456,
+      "learning_rate": 0.0001750932011636677,
+      "loss": 3.2133,
       "step": 65850
     },
     {
       "epoch": 7.092885588203638,
-      "grad_norm": 0.794703483581543,
-      "learning_rate": 0.0001747764249542075,
-      "loss": 3.1945,
+      "grad_norm": 0.7444468140602112,
+      "learning_rate": 0.00017476996013360627,
+      "loss": 3.1872,
       "step": 65900
     },
     {
       "epoch": 7.098267140243246,
-      "grad_norm": 0.7841687202453613,
-      "learning_rate": 0.0001744531839241461,
-      "loss": 3.2197,
+      "grad_norm": 0.83144211769104,
+      "learning_rate": 0.00017444671910354486,
+      "loss": 3.2046,
       "step": 65950
     },
     {
       "epoch": 7.103648692282855,
-      "grad_norm": 0.7183166742324829,
-      "learning_rate": 0.00017412994289408467,
-      "loss": 3.2254,
+      "grad_norm": 0.7932811379432678,
+      "learning_rate": 0.00017412347807348346,
+      "loss": 3.1971,
       "step": 66000
     },
     {
       "epoch": 7.103648692282855,
-      "eval_accuracy": 0.3870331036012891,
-      "eval_loss": 3.3611578941345215,
-      "eval_runtime": 179.9703,
-      "eval_samples_per_second": 100.078,
-      "eval_steps_per_second": 6.257,
+      "eval_accuracy": 0.38757169596618546,
+      "eval_loss": 3.3566536903381348,
+      "eval_runtime": 181.6859,
+      "eval_samples_per_second": 99.133,
+      "eval_steps_per_second": 6.198,
       "step": 66000
     },
     {
       "epoch": 7.109030244322462,
-      "grad_norm": 0.7794191837310791,
-      "learning_rate": 0.00017380670186402326,
-      "loss": 3.223,
+      "grad_norm": 0.7888944149017334,
+      "learning_rate": 0.00017380023704342202,
+      "loss": 3.2249,
       "step": 66050
     },
     {
       "epoch": 7.114411796362071,
-      "grad_norm": 0.8234367966651917,
-      "learning_rate": 0.00017348346083396183,
-      "loss": 3.1997,
+      "grad_norm": 0.7705891728401184,
+      "learning_rate": 0.00017347699601336062,
+      "loss": 3.1996,
       "step": 66100
     },
     {
       "epoch": 7.119793348401679,
-      "grad_norm": 0.8090280890464783,
-      "learning_rate": 0.00017316021980390042,
-      "loss": 3.2169,
+      "grad_norm": 0.8424685001373291,
+      "learning_rate": 0.00017315375498329919,
+      "loss": 3.2006,
       "step": 66150
     },
     {
       "epoch": 7.125174900441287,
-      "grad_norm": 0.8259748816490173,
-      "learning_rate": 0.00017283697877383902,
-      "loss": 3.2235,
+      "grad_norm": 0.7318297624588013,
+      "learning_rate": 0.00017283051395323778,
+      "loss": 3.2206,
       "step": 66200
     },
     {
       "epoch": 7.130556452480896,
-      "grad_norm": 0.8138856887817383,
-      "learning_rate": 0.0001725137377437776,
-      "loss": 3.2307,
+      "grad_norm": 0.7851969599723816,
+      "learning_rate": 0.00017250727292317638,
+      "loss": 3.232,
       "step": 66250
     },
     {
       "epoch": 7.135938004520503,
-      "grad_norm": 0.8415960669517517,
-      "learning_rate": 0.00017219049671371615,
-      "loss": 3.218,
+      "grad_norm": 0.7682739496231079,
+      "learning_rate": 0.00017218403189311494,
+      "loss": 3.2252,
       "step": 66300
     },
     {
       "epoch": 7.141319556560112,
-      "grad_norm": 0.8021517992019653,
-      "learning_rate": 0.00017186725568365478,
-      "loss": 3.2237,
+      "grad_norm": 0.8173055052757263,
+      "learning_rate": 0.0001718607908630535,
+      "loss": 3.2147,
       "step": 66350
     },
     {
       "epoch": 7.1467011085997205,
-      "grad_norm": 0.7997321486473083,
-      "learning_rate": 0.00017154401465359334,
-      "loss": 3.2073,
+      "grad_norm": 0.8225165605545044,
+      "learning_rate": 0.00017153754983299213,
+      "loss": 3.196,
       "step": 66400
     },
     {
       "epoch": 7.152082660639328,
-      "grad_norm": 0.7626723647117615,
-      "learning_rate": 0.00017122077362353194,
-      "loss": 3.2268,
+      "grad_norm": 0.7722660899162292,
+      "learning_rate": 0.0001712143088029307,
+      "loss": 3.2166,
       "step": 66450
     },
     {
       "epoch": 7.157464212678937,
-      "grad_norm": 0.8034189939498901,
-      "learning_rate": 0.00017089753259347053,
-      "loss": 3.2221,
+      "grad_norm": 0.7975138425827026,
+      "learning_rate": 0.00017089106777286927,
+      "loss": 3.2217,
       "step": 66500
     },
     {
       "epoch": 7.162845764718545,
-      "grad_norm": 0.8207605481147766,
-      "learning_rate": 0.0001705742915634091,
-      "loss": 3.2177,
+      "grad_norm": 0.7703187465667725,
+      "learning_rate": 0.0001705678267428079,
+      "loss": 3.2248,
       "step": 66550
     },
     {
       "epoch": 7.168227316758153,
-      "grad_norm": 0.8149417042732239,
-      "learning_rate": 0.0001702510505333477,
-      "loss": 3.221,
+      "grad_norm": 0.7811352610588074,
+      "learning_rate": 0.00017024458571274646,
+      "loss": 3.2359,
       "step": 66600
     },
     {
       "epoch": 7.1736088687977615,
-      "grad_norm": 0.8007292151451111,
-      "learning_rate": 0.00016992780950328626,
-      "loss": 3.2226,
+      "grad_norm": 0.8045165538787842,
+      "learning_rate": 0.00016992134468268505,
+      "loss": 3.2242,
       "step": 66650
     },
     {
       "epoch": 7.178990420837369,
-      "grad_norm": 0.7681205868721008,
-      "learning_rate": 0.00016960456847322486,
-      "loss": 3.2232,
+      "grad_norm": 0.8541814088821411,
+      "learning_rate": 0.00016959810365262362,
+      "loss": 3.2237,
       "step": 66700
     },
     {
       "epoch": 7.184371972876978,
-      "grad_norm": 0.785612940788269,
-      "learning_rate": 0.00016928132744316345,
-      "loss": 3.2141,
+      "grad_norm": 0.8304926753044128,
+      "learning_rate": 0.0001692748626225622,
+      "loss": 3.2235,
       "step": 66750
     },
     {
       "epoch": 7.189753524916586,
-      "grad_norm": 0.7706198692321777,
-      "learning_rate": 0.00016895808641310202,
-      "loss": 3.2163,
+      "grad_norm": 0.7365669012069702,
+      "learning_rate": 0.0001689516215925008,
+      "loss": 3.204,
       "step": 66800
     },
     {
       "epoch": 7.195135076956194,
-      "grad_norm": 0.8889934420585632,
-      "learning_rate": 0.0001686348453830406,
-      "loss": 3.2103,
+      "grad_norm": 0.8246890902519226,
+      "learning_rate": 0.00016862838056243938,
+      "loss": 3.2132,
       "step": 66850
     },
     {
       "epoch": 7.2005166289958025,
-      "grad_norm": 0.7743992209434509,
-      "learning_rate": 0.0001683116043529792,
-      "loss": 3.2334,
+      "grad_norm": 0.7545448541641235,
+      "learning_rate": 0.00016830513953237794,
+      "loss": 3.232,
       "step": 66900
     },
     {
       "epoch": 7.205898181035411,
-      "grad_norm": 0.7609925866127014,
-      "learning_rate": 0.00016798836332291778,
-      "loss": 3.2293,
+      "grad_norm": 0.7964430451393127,
+      "learning_rate": 0.00016798189850231657,
+      "loss": 3.2269,
       "step": 66950
     },
     {
       "epoch": 7.211279733075019,
-      "grad_norm": 0.778520941734314,
-      "learning_rate": 0.00016766512229285634,
-      "loss": 3.2413,
+      "grad_norm": 0.7150341868400574,
+      "learning_rate": 0.00016765865747225513,
+      "loss": 3.2303,
       "step": 67000
     },
     {
       "epoch": 7.211279733075019,
-      "eval_accuracy": 0.38746108732639034,
-      "eval_loss": 3.359044075012207,
-      "eval_runtime": 180.1842,
-      "eval_samples_per_second": 99.959,
-      "eval_steps_per_second": 6.249,
+      "eval_accuracy": 0.38786408088729424,
+      "eval_loss": 3.353409767150879,
+      "eval_runtime": 181.3389,
+      "eval_samples_per_second": 99.322,
+      "eval_steps_per_second": 6.209,
       "step": 67000
     },
     {
       "epoch": 7.216661285114627,
-      "grad_norm": 0.7969582080841064,
-      "learning_rate": 0.00016734188126279494,
-      "loss": 3.2454,
+      "grad_norm": 0.8423494696617126,
+      "learning_rate": 0.0001673354164421937,
+      "loss": 3.2443,
       "step": 67050
     },
     {
       "epoch": 7.222042837154235,
-      "grad_norm": 0.7877959609031677,
-      "learning_rate": 0.00016701864023273353,
-      "loss": 3.2199,
+      "grad_norm": 0.7422992587089539,
+      "learning_rate": 0.00016701217541213232,
+      "loss": 3.207,
       "step": 67100
     },
     {
       "epoch": 7.2274243891938434,
-      "grad_norm": 0.8033312559127808,
-      "learning_rate": 0.0001666953992026721,
-      "loss": 3.2453,
+      "grad_norm": 0.7926647067070007,
+      "learning_rate": 0.0001666889343820709,
+      "loss": 3.2395,
       "step": 67150
     },
     {
       "epoch": 7.232805941233452,
-      "grad_norm": 0.775666356086731,
-      "learning_rate": 0.0001663721581726107,
-      "loss": 3.2301,
+      "grad_norm": 0.8212176561355591,
+      "learning_rate": 0.00016636569335200946,
+      "loss": 3.2073,
       "step": 67200
     },
     {
       "epoch": 7.23818749327306,
-      "grad_norm": 0.851057767868042,
-      "learning_rate": 0.00016604891714254926,
-      "loss": 3.2248,
+      "grad_norm": 0.8213561773300171,
+      "learning_rate": 0.00016604245232194805,
+      "loss": 3.2357,
       "step": 67250
     },
     {
       "epoch": 7.243569045312668,
-      "grad_norm": 0.8318371176719666,
-      "learning_rate": 0.00016572567611248786,
-      "loss": 3.2132,
+      "grad_norm": 0.8193587064743042,
+      "learning_rate": 0.00016571921129188665,
+      "loss": 3.221,
       "step": 67300
     },
     {
       "epoch": 7.248950597352277,
-      "grad_norm": 0.817715585231781,
-      "learning_rate": 0.00016540243508242645,
-      "loss": 3.238,
+      "grad_norm": 0.8527218103408813,
+      "learning_rate": 0.00016539597026182521,
+      "loss": 3.2299,
       "step": 67350
     },
     {
       "epoch": 7.254332149391884,
-      "grad_norm": 0.7780318856239319,
-      "learning_rate": 0.00016507919405236502,
-      "loss": 3.2238,
+      "grad_norm": 0.7740417122840881,
+      "learning_rate": 0.0001650727292317638,
+      "loss": 3.2189,
       "step": 67400
     },
     {
       "epoch": 7.259713701431493,
-      "grad_norm": 0.7914072275161743,
-      "learning_rate": 0.00016475595302230364,
-      "loss": 3.2442,
+      "grad_norm": 0.7607315182685852,
+      "learning_rate": 0.00016474948820170238,
+      "loss": 3.2363,
       "step": 67450
     },
     {
       "epoch": 7.265095253471101,
-      "grad_norm": 0.7719249129295349,
-      "learning_rate": 0.0001644327119922422,
-      "loss": 3.2172,
+      "grad_norm": 0.7882745265960693,
+      "learning_rate": 0.00016442624717164094,
+      "loss": 3.2082,
       "step": 67500
     },
     {
       "epoch": 7.270476805510709,
-      "grad_norm": 0.7739912271499634,
-      "learning_rate": 0.00016410947096218078,
-      "loss": 3.2197,
+      "grad_norm": 0.8261680006980896,
+      "learning_rate": 0.00016410300614157957,
+      "loss": 3.2368,
       "step": 67550
     },
     {
       "epoch": 7.275858357550318,
-      "grad_norm": 0.8210995197296143,
-      "learning_rate": 0.00016378622993211937,
-      "loss": 3.228,
+      "grad_norm": 0.7855998277664185,
+      "learning_rate": 0.00016377976511151813,
+      "loss": 3.2237,
       "step": 67600
     },
     {
       "epoch": 7.281239909589925,
-      "grad_norm": 0.8359017968177795,
-      "learning_rate": 0.00016346298890205797,
-      "loss": 3.2404,
+      "grad_norm": 0.8033557534217834,
+      "learning_rate": 0.00016345652408145675,
+      "loss": 3.2331,
       "step": 67650
     },
     {
       "epoch": 7.286621461629534,
-      "grad_norm": 0.8332840204238892,
-      "learning_rate": 0.00016313974787199653,
-      "loss": 3.2456,
+      "grad_norm": 0.7629213929176331,
+      "learning_rate": 0.00016313328305139532,
+      "loss": 3.2268,
       "step": 67700
     },
     {
       "epoch": 7.2920030136691425,
-      "grad_norm": 0.8179646730422974,
-      "learning_rate": 0.00016281650684193513,
-      "loss": 3.2327,
+      "grad_norm": 0.7886566519737244,
+      "learning_rate": 0.0001628100420213339,
+      "loss": 3.2256,
       "step": 67750
     },
     {
       "epoch": 7.29738456570875,
-      "grad_norm": 0.7919309139251709,
-      "learning_rate": 0.0001624932658118737,
-      "loss": 3.2263,
+      "grad_norm": 0.8063637614250183,
+      "learning_rate": 0.00016248680099127248,
+      "loss": 3.2256,
       "step": 67800
     },
     {
       "epoch": 7.302766117748359,
-      "grad_norm": 0.8252838850021362,
-      "learning_rate": 0.0001621700247818123,
-      "loss": 3.2371,
+      "grad_norm": 0.7654820680618286,
+      "learning_rate": 0.00016216355996121105,
+      "loss": 3.2192,
       "step": 67850
     },
     {
       "epoch": 7.308147669787967,
-      "grad_norm": 0.7692036032676697,
-      "learning_rate": 0.00016184678375175089,
-      "loss": 3.2357,
+      "grad_norm": 0.7816736698150635,
+      "learning_rate": 0.00016184031893114965,
+      "loss": 3.2137,
       "step": 67900
     },
     {
       "epoch": 7.313529221827575,
-      "grad_norm": 0.8547636866569519,
-      "learning_rate": 0.00016153000754229067,
-      "loss": 3.2201,
+      "grad_norm": 0.7907631993293762,
+      "learning_rate": 0.00016151707790108824,
+      "loss": 3.2417,
       "step": 67950
     },
     {
       "epoch": 7.3189107738671835,
-      "grad_norm": 0.8004331588745117,
-      "learning_rate": 0.0001612067665122293,
-      "loss": 3.2095,
+      "grad_norm": 0.793752908706665,
+      "learning_rate": 0.0001611938368710268,
+      "loss": 3.2096,
       "step": 68000
     },
     {
       "epoch": 7.3189107738671835,
-      "eval_accuracy": 0.38778693733694597,
-      "eval_loss": 3.356480598449707,
-      "eval_runtime": 180.04,
-      "eval_samples_per_second": 100.039,
-      "eval_steps_per_second": 6.254,
+      "eval_accuracy": 0.38831336057840715,
+      "eval_loss": 3.350975513458252,
+      "eval_runtime": 181.5869,
+      "eval_samples_per_second": 99.187,
+      "eval_steps_per_second": 6.201,
       "step": 68000
     },
     {
       "epoch": 7.324292325906791,
-      "grad_norm": 0.811409592628479,
-      "learning_rate": 0.00016088352548216785,
-      "loss": 3.2466,
+      "grad_norm": 0.760727047920227,
+      "learning_rate": 0.00016087059584096538,
+      "loss": 3.2186,
       "step": 68050
     },
     {
       "epoch": 7.3296738779464,
-      "grad_norm": 0.8523727655410767,
-      "learning_rate": 0.00016056028445210642,
-      "loss": 3.2269,
+      "grad_norm": 0.7698805928230286,
+      "learning_rate": 0.000160547354810904,
+      "loss": 3.2293,
       "step": 68100
     },
     {
       "epoch": 7.335055429986008,
-      "grad_norm": 0.809126615524292,
-      "learning_rate": 0.00016023704342204504,
-      "loss": 3.2503,
+      "grad_norm": 0.7481377720832825,
+      "learning_rate": 0.00016022411378084257,
+      "loss": 3.2303,
       "step": 68150
     },
     {
       "epoch": 7.340436982025616,
-      "grad_norm": 0.8085951209068298,
-      "learning_rate": 0.0001599138023919836,
-      "loss": 3.2459,
+      "grad_norm": 0.7590205669403076,
+      "learning_rate": 0.00015990087275078113,
+      "loss": 3.2357,
       "step": 68200
     },
     {
       "epoch": 7.3458185340652244,
-      "grad_norm": 0.7844497561454773,
-      "learning_rate": 0.0001595905613619222,
-      "loss": 3.236,
+      "grad_norm": 0.8002303242683411,
+      "learning_rate": 0.00015957763172071976,
+      "loss": 3.2209,
       "step": 68250
     },
     {
       "epoch": 7.351200086104833,
-      "grad_norm": 0.8168286681175232,
-      "learning_rate": 0.00015926732033186077,
-      "loss": 3.2236,
+      "grad_norm": 0.796108067035675,
+      "learning_rate": 0.00015925439069065832,
+      "loss": 3.2304,
       "step": 68300
     },
     {
       "epoch": 7.356581638144441,
-      "grad_norm": 0.8217577338218689,
-      "learning_rate": 0.00015894407930179934,
-      "loss": 3.226,
+      "grad_norm": 0.8114858865737915,
+      "learning_rate": 0.0001589311496605969,
+      "loss": 3.2392,
       "step": 68350
     },
     {
       "epoch": 7.361963190184049,
-      "grad_norm": 0.8433511853218079,
-      "learning_rate": 0.00015862083827173796,
-      "loss": 3.2294,
+      "grad_norm": 0.7842608094215393,
+      "learning_rate": 0.00015860790863053548,
+      "loss": 3.2327,
       "step": 68400
     },
     {
       "epoch": 7.367344742223658,
-      "grad_norm": 0.8781927227973938,
-      "learning_rate": 0.00015829759724167653,
-      "loss": 3.2264,
+      "grad_norm": 0.7788580060005188,
+      "learning_rate": 0.00015828466760047408,
+      "loss": 3.2271,
       "step": 68450
     },
     {
       "epoch": 7.372726294263265,
-      "grad_norm": 0.8104392290115356,
-      "learning_rate": 0.0001579743562116151,
-      "loss": 3.2114,
+      "grad_norm": 0.7798500657081604,
+      "learning_rate": 0.00015796142657041265,
+      "loss": 3.2234,
       "step": 68500
     },
     {
       "epoch": 7.378107846302874,
-      "grad_norm": 0.8192629218101501,
-      "learning_rate": 0.00015765111518155372,
-      "loss": 3.2297,
+      "grad_norm": 0.796565055847168,
+      "learning_rate": 0.00015763818554035124,
+      "loss": 3.2084,
       "step": 68550
     },
     {
       "epoch": 7.383489398342482,
-      "grad_norm": 0.7943025231361389,
-      "learning_rate": 0.0001573278741514923,
-      "loss": 3.2313,
+      "grad_norm": 0.8083150386810303,
+      "learning_rate": 0.0001573149445102898,
+      "loss": 3.227,
       "step": 68600
     },
     {
       "epoch": 7.38887095038209,
-      "grad_norm": 0.858521580696106,
-      "learning_rate": 0.00015700463312143085,
-      "loss": 3.2268,
+      "grad_norm": 0.8706159591674805,
+      "learning_rate": 0.00015699170348022843,
+      "loss": 3.2317,
       "step": 68650
     },
     {
       "epoch": 7.394252502421699,
-      "grad_norm": 0.804550290107727,
-      "learning_rate": 0.00015668139209136945,
-      "loss": 3.2131,
+      "grad_norm": 0.7665033340454102,
+      "learning_rate": 0.000156668462450167,
+      "loss": 3.2073,
       "step": 68700
     },
     {
       "epoch": 7.399634054461306,
-      "grad_norm": 0.8339864611625671,
-      "learning_rate": 0.00015635815106130804,
-      "loss": 3.2387,
+      "grad_norm": 0.8318244814872742,
+      "learning_rate": 0.00015634522142010557,
+      "loss": 3.2267,
       "step": 68750
     },
     {
       "epoch": 7.405015606500915,
-      "grad_norm": 0.7839356064796448,
-      "learning_rate": 0.0001560349100312466,
-      "loss": 3.2269,
+      "grad_norm": 0.8506566882133484,
+      "learning_rate": 0.0001560284452106454,
+      "loss": 3.2304,
       "step": 68800
     },
     {
       "epoch": 7.4103971585405235,
-      "grad_norm": 0.8349343538284302,
-      "learning_rate": 0.0001557116690011852,
-      "loss": 3.2374,
+      "grad_norm": 0.8108691573143005,
+      "learning_rate": 0.00015570520418058397,
+      "loss": 3.2248,
       "step": 68850
     },
     {
       "epoch": 7.415778710580131,
-      "grad_norm": 0.8796477317810059,
-      "learning_rate": 0.00015538842797112377,
-      "loss": 3.2449,
+      "grad_norm": 0.7729952335357666,
+      "learning_rate": 0.00015538196315052256,
+      "loss": 3.241,
       "step": 68900
     },
     {
       "epoch": 7.42116026261974,
-      "grad_norm": 0.8090130686759949,
-      "learning_rate": 0.00015506518694106237,
-      "loss": 3.2412,
+      "grad_norm": 0.762362539768219,
+      "learning_rate": 0.00015505872212046113,
+      "loss": 3.2391,
       "step": 68950
     },
     {
       "epoch": 7.426541814659347,
-      "grad_norm": 0.8464242219924927,
-      "learning_rate": 0.00015474194591100096,
-      "loss": 3.2314,
+      "grad_norm": 0.7620248794555664,
+      "learning_rate": 0.00015473548109039972,
+      "loss": 3.2144,
       "step": 69000
     },
     {
       "epoch": 7.426541814659347,
-      "eval_accuracy": 0.38814625243694845,
-      "eval_loss": 3.3514676094055176,
-      "eval_runtime": 180.0963,
-      "eval_samples_per_second": 100.008,
-      "eval_steps_per_second": 6.252,
+      "eval_accuracy": 0.38857619191403037,
+      "eval_loss": 3.348639488220215,
+      "eval_runtime": 181.642,
+      "eval_samples_per_second": 99.157,
+      "eval_steps_per_second": 6.199,
       "step": 69000
     },
     {
       "epoch": 7.431923366698956,
-      "grad_norm": 0.848106324672699,
-      "learning_rate": 0.00015441870488093953,
-      "loss": 3.2432,
+      "grad_norm": 0.7892707586288452,
+      "learning_rate": 0.00015441224006033832,
+      "loss": 3.2208,
       "step": 69050
     },
     {
       "epoch": 7.4373049187385645,
-      "grad_norm": 0.8055359125137329,
-      "learning_rate": 0.0001540954638508781,
-      "loss": 3.2373,
+      "grad_norm": 0.8356212973594666,
+      "learning_rate": 0.00015408899903027689,
+      "loss": 3.2334,
       "step": 69100
     },
     {
       "epoch": 7.442686470778172,
-      "grad_norm": 0.8940873742103577,
-      "learning_rate": 0.00015377222282081672,
-      "loss": 3.2389,
+      "grad_norm": 0.810688853263855,
+      "learning_rate": 0.00015376575800021545,
+      "loss": 3.2326,
       "step": 69150
     },
     {
       "epoch": 7.448068022817781,
-      "grad_norm": 0.7894365787506104,
-      "learning_rate": 0.0001534489817907553,
-      "loss": 3.237,
+      "grad_norm": 0.8027896285057068,
+      "learning_rate": 0.00015344251697015408,
+      "loss": 3.2156,
       "step": 69200
     },
     {
       "epoch": 7.453449574857389,
-      "grad_norm": 0.8256045579910278,
-      "learning_rate": 0.00015312574076069388,
-      "loss": 3.2281,
+      "grad_norm": 0.8302093148231506,
+      "learning_rate": 0.00015311927594009264,
+      "loss": 3.2403,
       "step": 69250
     },
     {
       "epoch": 7.458831126896997,
-      "grad_norm": 0.7994895577430725,
-      "learning_rate": 0.00015280249973063248,
-      "loss": 3.2525,
+      "grad_norm": 0.7733965516090393,
+      "learning_rate": 0.0001527960349100312,
+      "loss": 3.2319,
       "step": 69300
     },
     {
       "epoch": 7.4642126789366054,
-      "grad_norm": 0.8331586718559265,
-      "learning_rate": 0.00015247925870057104,
-      "loss": 3.2337,
+      "grad_norm": 0.8160131573677063,
+      "learning_rate": 0.00015247279387996983,
+      "loss": 3.234,
       "step": 69350
     },
     {
       "epoch": 7.469594230976213,
-      "grad_norm": 0.7736141085624695,
-      "learning_rate": 0.00015215601767050964,
-      "loss": 3.2536,
+      "grad_norm": 0.8102094531059265,
+      "learning_rate": 0.0001521495528499084,
+      "loss": 3.2424,
       "step": 69400
     },
     {
       "epoch": 7.474975783015822,
-      "grad_norm": 0.8336132764816284,
-      "learning_rate": 0.0001518327766404482,
-      "loss": 3.2434,
+      "grad_norm": 0.8183262944221497,
+      "learning_rate": 0.000151826311819847,
+      "loss": 3.2325,
       "step": 69450
     },
     {
       "epoch": 7.48035733505543,
-      "grad_norm": 0.8452787399291992,
-      "learning_rate": 0.0001515095356103868,
-      "loss": 3.2308,
+      "grad_norm": 0.8657780289649963,
+      "learning_rate": 0.00015150307078978556,
+      "loss": 3.2311,
       "step": 69500
     },
     {
       "epoch": 7.485738887095038,
-      "grad_norm": 0.7975885272026062,
-      "learning_rate": 0.0001511862945803254,
-      "loss": 3.2414,
+      "grad_norm": 0.7791820764541626,
+      "learning_rate": 0.00015117982975972416,
+      "loss": 3.2286,
       "step": 69550
     },
     {
       "epoch": 7.491120439134646,
-      "grad_norm": 0.7722816467285156,
-      "learning_rate": 0.00015086305355026396,
-      "loss": 3.2329,
+      "grad_norm": 0.8065508604049683,
+      "learning_rate": 0.00015085658872966275,
+      "loss": 3.2361,
       "step": 69600
     },
     {
       "epoch": 7.496501991174255,
-      "grad_norm": 0.7675669193267822,
-      "learning_rate": 0.00015053981252020253,
-      "loss": 3.2375,
+      "grad_norm": 0.8331541419029236,
+      "learning_rate": 0.00015053334769960132,
+      "loss": 3.2505,
       "step": 69650
     },
     {
       "epoch": 7.501883543213863,
-      "grad_norm": 0.7885386347770691,
-      "learning_rate": 0.00015021657149014115,
-      "loss": 3.2363,
+      "grad_norm": 0.8051180243492126,
+      "learning_rate": 0.0001502101066695399,
+      "loss": 3.2229,
       "step": 69700
     },
     {
       "epoch": 7.507265095253471,
-      "grad_norm": 0.792849063873291,
-      "learning_rate": 0.00014989333046007972,
-      "loss": 3.2484,
+      "grad_norm": 0.7906857132911682,
+      "learning_rate": 0.00014988686563947848,
+      "loss": 3.242,
       "step": 69750
     },
     {
       "epoch": 7.51264664729308,
-      "grad_norm": 0.837902843952179,
-      "learning_rate": 0.00014957008943001832,
-      "loss": 3.2335,
+      "grad_norm": 0.799919605255127,
+      "learning_rate": 0.00014956362460941708,
+      "loss": 3.234,
       "step": 69800
     },
     {
       "epoch": 7.518028199332687,
-      "grad_norm": 0.8501192331314087,
-      "learning_rate": 0.00014924684839995688,
-      "loss": 3.2218,
+      "grad_norm": 0.7783071994781494,
+      "learning_rate": 0.00014924038357935567,
+      "loss": 3.2514,
       "step": 69850
     },
     {
       "epoch": 7.523409751372296,
-      "grad_norm": 0.8147408962249756,
-      "learning_rate": 0.00014892360736989548,
-      "loss": 3.2178,
+      "grad_norm": 0.818239152431488,
+      "learning_rate": 0.00014891714254929424,
+      "loss": 3.2355,
       "step": 69900
     },
     {
       "epoch": 7.528791303411904,
-      "grad_norm": 0.7939187288284302,
-      "learning_rate": 0.00014860036633983407,
-      "loss": 3.2485,
+      "grad_norm": 0.7930513620376587,
+      "learning_rate": 0.00014859390151923283,
+      "loss": 3.2317,
       "step": 69950
     },
     {
       "epoch": 7.534172855451512,
-      "grad_norm": 0.846481204032898,
-      "learning_rate": 0.00014827712530977264,
-      "loss": 3.2351,
+      "grad_norm": 0.8230518698692322,
+      "learning_rate": 0.00014827066048917143,
+      "loss": 3.2082,
       "step": 70000
     },
     {
       "epoch": 7.534172855451512,
-      "eval_accuracy": 0.38860226860710584,
-      "eval_loss": 3.34824800491333,
-      "eval_runtime": 180.1854,
-      "eval_samples_per_second": 99.958,
-      "eval_steps_per_second": 6.249,
+      "eval_accuracy": 0.3889937449619014,
+      "eval_loss": 3.3427162170410156,
+      "eval_runtime": 181.5664,
+      "eval_samples_per_second": 99.198,
+      "eval_steps_per_second": 6.202,
       "step": 70000
     },
     {
       "epoch": 7.539554407491121,
-      "grad_norm": 0.822512686252594,
-      "learning_rate": 0.00014795388427971123,
-      "loss": 3.2308,
+      "grad_norm": 0.7911315560340881,
+      "learning_rate": 0.00014794741945911,
+      "loss": 3.2252,
       "step": 70050
     },
     {
       "epoch": 7.544935959530728,
-      "grad_norm": 0.8041254878044128,
-      "learning_rate": 0.0001476306432496498,
-      "loss": 3.2348,
+      "grad_norm": 0.7912471890449524,
+      "learning_rate": 0.0001476241784290486,
+      "loss": 3.2372,
       "step": 70100
     },
     {
       "epoch": 7.550317511570337,
-      "grad_norm": 0.8251069784164429,
-      "learning_rate": 0.0001473074022195884,
-      "loss": 3.2205,
+      "grad_norm": 0.813069224357605,
+      "learning_rate": 0.00014730093739898716,
+      "loss": 3.2288,
       "step": 70150
     },
     {
       "epoch": 7.5556990636099455,
-      "grad_norm": 0.7714356184005737,
-      "learning_rate": 0.00014698416118952696,
-      "loss": 3.2205,
+      "grad_norm": 0.8153437972068787,
+      "learning_rate": 0.00014697769636892575,
+      "loss": 3.2109,
       "step": 70200
     },
     {
       "epoch": 7.561080615649553,
-      "grad_norm": 0.8164318203926086,
-      "learning_rate": 0.00014666092015946556,
-      "loss": 3.2301,
+      "grad_norm": 0.8853901028633118,
+      "learning_rate": 0.00014665445533886432,
+      "loss": 3.2344,
       "step": 70250
     },
     {
       "epoch": 7.566462167689162,
-      "grad_norm": 0.8093104958534241,
-      "learning_rate": 0.00014633767912940415,
-      "loss": 3.2324,
+      "grad_norm": 0.817044734954834,
+      "learning_rate": 0.00014633121430880291,
+      "loss": 3.2314,
       "step": 70300
     },
     {
       "epoch": 7.57184371972877,
-      "grad_norm": 0.822935938835144,
-      "learning_rate": 0.00014601443809934272,
-      "loss": 3.2378,
+      "grad_norm": 0.8000847101211548,
+      "learning_rate": 0.0001460079732787415,
+      "loss": 3.2389,
       "step": 70350
     },
     {
       "epoch": 7.577225271768378,
-      "grad_norm": 0.8421322703361511,
-      "learning_rate": 0.00014569119706928132,
-      "loss": 3.2322,
+      "grad_norm": 0.8437427282333374,
+      "learning_rate": 0.00014568473224868008,
+      "loss": 3.2303,
       "step": 70400
     },
     {
       "epoch": 7.5826068238079865,
-      "grad_norm": 0.7766196727752686,
-      "learning_rate": 0.0001453679560392199,
-      "loss": 3.245,
+      "grad_norm": 0.8098722100257874,
+      "learning_rate": 0.00014536149121861867,
+      "loss": 3.2411,
       "step": 70450
     },
     {
       "epoch": 7.587988375847594,
-      "grad_norm": 0.8198171854019165,
-      "learning_rate": 0.00014504471500915848,
-      "loss": 3.2239,
+      "grad_norm": 0.8720215559005737,
+      "learning_rate": 0.00014503825018855727,
+      "loss": 3.2363,
       "step": 70500
     },
     {
       "epoch": 7.593369927887203,
-      "grad_norm": 0.7970831394195557,
-      "learning_rate": 0.00014472147397909707,
-      "loss": 3.2244,
+      "grad_norm": 0.8183484673500061,
+      "learning_rate": 0.00014471500915849583,
+      "loss": 3.2384,
       "step": 70550
     },
     {
       "epoch": 7.598751479926811,
-      "grad_norm": 0.8099634051322937,
-      "learning_rate": 0.00014439823294903564,
-      "loss": 3.239,
+      "grad_norm": 0.7885457873344421,
+      "learning_rate": 0.00014439176812843443,
+      "loss": 3.2395,
       "step": 70600
     },
     {
       "epoch": 7.604133031966419,
-      "grad_norm": 0.7905451655387878,
-      "learning_rate": 0.00014407499191897423,
-      "loss": 3.2474,
+      "grad_norm": 0.8109983205795288,
+      "learning_rate": 0.00014406852709837302,
+      "loss": 3.2247,
       "step": 70650
     },
     {
       "epoch": 7.609514584006027,
-      "grad_norm": 0.8306140303611755,
-      "learning_rate": 0.0001437517508889128,
-      "loss": 3.2549,
+      "grad_norm": 0.8180787563323975,
+      "learning_rate": 0.0001437452860683116,
+      "loss": 3.2441,
       "step": 70700
     },
     {
       "epoch": 7.614896136045635,
-      "grad_norm": 0.8547305464744568,
-      "learning_rate": 0.0001434285098588514,
-      "loss": 3.245,
+      "grad_norm": 0.8033555150032043,
+      "learning_rate": 0.00014342204503825018,
+      "loss": 3.2429,
       "step": 70750
     },
     {
       "epoch": 7.620277688085244,
-      "grad_norm": 0.8246287107467651,
-      "learning_rate": 0.00014311173364939123,
-      "loss": 3.236,
+      "grad_norm": 0.7792641520500183,
+      "learning_rate": 0.00014310526882879,
+      "loss": 3.225,
       "step": 70800
     },
     {
       "epoch": 7.625659240124852,
-      "grad_norm": 0.7652173638343811,
-      "learning_rate": 0.0001427884926193298,
-      "loss": 3.2216,
+      "grad_norm": 0.7894015908241272,
+      "learning_rate": 0.00014278202779872856,
+      "loss": 3.2136,
       "step": 70850
     },
     {
       "epoch": 7.63104079216446,
-      "grad_norm": 0.8844686150550842,
-      "learning_rate": 0.0001424652515892684,
-      "loss": 3.2323,
+      "grad_norm": 0.8328763246536255,
+      "learning_rate": 0.00014245878676866715,
+      "loss": 3.2249,
       "step": 70900
     },
     {
       "epoch": 7.636422344204068,
-      "grad_norm": 0.8048098087310791,
-      "learning_rate": 0.000142142010559207,
-      "loss": 3.2566,
+      "grad_norm": 0.7897523641586304,
+      "learning_rate": 0.00014213554573860575,
+      "loss": 3.2324,
       "step": 70950
     },
     {
       "epoch": 7.641803896243677,
-      "grad_norm": 0.7921103835105896,
-      "learning_rate": 0.00014181876952914555,
-      "loss": 3.2345,
+      "grad_norm": 0.8640326261520386,
+      "learning_rate": 0.00014181230470854434,
+      "loss": 3.22,
       "step": 71000
     },
     {
       "epoch": 7.641803896243677,
-      "eval_accuracy": 0.3890942488831298,
-      "eval_loss": 3.343825101852417,
-      "eval_runtime": 180.1847,
-      "eval_samples_per_second": 99.959,
-      "eval_steps_per_second": 6.249,
+      "eval_accuracy": 0.3896137183397708,
+      "eval_loss": 3.338975667953491,
+      "eval_runtime": 181.6579,
+      "eval_samples_per_second": 99.148,
+      "eval_steps_per_second": 6.198,
       "step": 71000
     },
     {
       "epoch": 7.647185448283285,
-      "grad_norm": 0.799433171749115,
-      "learning_rate": 0.00014149552849908415,
-      "loss": 3.2373,
+      "grad_norm": 0.8010990023612976,
+      "learning_rate": 0.0001414890636784829,
+      "loss": 3.239,
       "step": 71050
     },
     {
       "epoch": 7.652567000322893,
-      "grad_norm": 0.8794057369232178,
-      "learning_rate": 0.00014117228746902272,
-      "loss": 3.2458,
+      "grad_norm": 0.7835224866867065,
+      "learning_rate": 0.0001411658226484215,
+      "loss": 3.23,
       "step": 71100
     },
     {
       "epoch": 7.657948552362502,
-      "grad_norm": 0.8924712538719177,
-      "learning_rate": 0.0001408490464389613,
-      "loss": 3.2278,
+      "grad_norm": 0.8282446265220642,
+      "learning_rate": 0.00014084258161836007,
+      "loss": 3.2146,
       "step": 71150
     },
     {
       "epoch": 7.663330104402109,
-      "grad_norm": 0.8285935521125793,
-      "learning_rate": 0.00014052580540889988,
-      "loss": 3.2317,
+      "grad_norm": 0.870155394077301,
+      "learning_rate": 0.00014051934058829867,
+      "loss": 3.2181,
       "step": 71200
     },
     {
       "epoch": 7.668711656441718,
-      "grad_norm": 0.8792213201522827,
-      "learning_rate": 0.00014020256437883847,
-      "loss": 3.2425,
+      "grad_norm": 0.8047078847885132,
+      "learning_rate": 0.00014019609955823723,
+      "loss": 3.2365,
       "step": 71250
     },
     {
       "epoch": 7.674093208481326,
-      "grad_norm": 0.8292412161827087,
-      "learning_rate": 0.00013987932334877707,
-      "loss": 3.2234,
+      "grad_norm": 0.8340807557106018,
+      "learning_rate": 0.00013987285852817583,
+      "loss": 3.2256,
       "step": 71300
     },
     {
       "epoch": 7.679474760520934,
-      "grad_norm": 0.7959591150283813,
-      "learning_rate": 0.00013955608231871564,
-      "loss": 3.2474,
+      "grad_norm": 0.8595308065414429,
+      "learning_rate": 0.0001395496174981144,
+      "loss": 3.2493,
       "step": 71350
     },
     {
       "epoch": 7.684856312560543,
-      "grad_norm": 0.821670651435852,
-      "learning_rate": 0.00013923284128865423,
-      "loss": 3.2452,
+      "grad_norm": 0.8378287553787231,
+      "learning_rate": 0.000139226376468053,
+      "loss": 3.2326,
       "step": 71400
     },
     {
       "epoch": 7.69023786460015,
-      "grad_norm": 0.8048251271247864,
-      "learning_rate": 0.00013890960025859283,
-      "loss": 3.2417,
+      "grad_norm": 0.8430325984954834,
+      "learning_rate": 0.00013890313543799159,
+      "loss": 3.2475,
       "step": 71450
     },
     {
       "epoch": 7.695619416639759,
-      "grad_norm": 0.846571147441864,
-      "learning_rate": 0.0001385863592285314,
-      "loss": 3.2356,
+      "grad_norm": 0.8614859580993652,
+      "learning_rate": 0.00013857989440793018,
+      "loss": 3.244,
       "step": 71500
     },
     {
       "epoch": 7.7010009686793675,
-      "grad_norm": 0.8968475461006165,
-      "learning_rate": 0.00013826311819847,
-      "loss": 3.2225,
+      "grad_norm": 0.7920451164245605,
+      "learning_rate": 0.00013825665337786875,
+      "loss": 3.2181,
       "step": 71550
     },
     {
       "epoch": 7.706382520718975,
-      "grad_norm": 0.8345704674720764,
-      "learning_rate": 0.00013793987716840858,
-      "loss": 3.2244,
+      "grad_norm": 0.8030318021774292,
+      "learning_rate": 0.00013793341234780734,
+      "loss": 3.2142,
       "step": 71600
     },
     {
       "epoch": 7.711764072758584,
-      "grad_norm": 0.7907124757766724,
-      "learning_rate": 0.00013761663613834715,
-      "loss": 3.2141,
+      "grad_norm": 0.8280240297317505,
+      "learning_rate": 0.00013761017131774594,
+      "loss": 3.2039,
       "step": 71650
     },
     {
       "epoch": 7.717145624798192,
-      "grad_norm": 0.8455408811569214,
-      "learning_rate": 0.00013729339510828572,
-      "loss": 3.2401,
+      "grad_norm": 0.8104244470596313,
+      "learning_rate": 0.0001372869302876845,
+      "loss": 3.2386,
       "step": 71700
     },
     {
       "epoch": 7.7225271768378,
-      "grad_norm": 0.8723199963569641,
-      "learning_rate": 0.0001369701540782243,
-      "loss": 3.241,
+      "grad_norm": 0.797492265701294,
+      "learning_rate": 0.0001369636892576231,
+      "loss": 3.2491,
       "step": 71750
     },
     {
       "epoch": 7.727908728877408,
-      "grad_norm": 0.8454627990722656,
-      "learning_rate": 0.0001366469130481629,
-      "loss": 3.2365,
+      "grad_norm": 0.825414776802063,
+      "learning_rate": 0.00013664044822756167,
+      "loss": 3.217,
       "step": 71800
     },
     {
       "epoch": 7.733290280917016,
-      "grad_norm": 0.8864803314208984,
-      "learning_rate": 0.00013632367201810147,
-      "loss": 3.2511,
+      "grad_norm": 0.8557885885238647,
+      "learning_rate": 0.00013631720719750026,
+      "loss": 3.2477,
       "step": 71850
     },
     {
       "epoch": 7.738671832956625,
-      "grad_norm": 0.8611536026000977,
-      "learning_rate": 0.00013600043098804007,
-      "loss": 3.2308,
+      "grad_norm": 0.7807618975639343,
+      "learning_rate": 0.00013599396616743883,
+      "loss": 3.2249,
       "step": 71900
     },
     {
       "epoch": 7.744053384996233,
-      "grad_norm": 0.8165392875671387,
-      "learning_rate": 0.00013567718995797866,
-      "loss": 3.2312,
+      "grad_norm": 0.8000689148902893,
+      "learning_rate": 0.00013567072513737742,
+      "loss": 3.2333,
       "step": 71950
     },
     {
       "epoch": 7.749434937035841,
-      "grad_norm": 0.8212934732437134,
-      "learning_rate": 0.00013535394892791723,
-      "loss": 3.2414,
+      "grad_norm": 0.8142040371894836,
+      "learning_rate": 0.00013534748410731602,
+      "loss": 3.2323,
       "step": 72000
     },
     {
       "epoch": 7.749434937035841,
-      "eval_accuracy": 0.38907805960284547,
-      "eval_loss": 3.3423426151275635,
-      "eval_runtime": 180.2997,
-      "eval_samples_per_second": 99.895,
-      "eval_steps_per_second": 6.245,
+      "eval_accuracy": 0.3897280211777517,
+      "eval_loss": 3.3353919982910156,
+      "eval_runtime": 181.8559,
+      "eval_samples_per_second": 99.04,
+      "eval_steps_per_second": 6.192,
       "step": 72000
     },
     {
       "epoch": 7.754816489075449,
-      "grad_norm": 0.7960253953933716,
-      "learning_rate": 0.00013503070789785583,
-      "loss": 3.2214,
+      "grad_norm": 0.7958768606185913,
+      "learning_rate": 0.0001350242430772546,
+      "loss": 3.2188,
       "step": 72050
     },
     {
       "epoch": 7.760198041115058,
-      "grad_norm": 0.8525424003601074,
+      "grad_norm": 0.8156412243843079,
       "learning_rate": 0.00013470746686779442,
-      "loss": 3.2315,
+      "loss": 3.2383,
       "step": 72100
     },
     {
       "epoch": 7.765579593154666,
-      "grad_norm": 0.7953471541404724,
+      "grad_norm": 0.8011084794998169,
       "learning_rate": 0.000134384225837733,
-      "loss": 3.2473,
+      "loss": 3.2353,
       "step": 72150
     },
     {
       "epoch": 7.770961145194274,
-      "grad_norm": 0.8736206293106079,
+      "grad_norm": 0.7789442539215088,
       "learning_rate": 0.00013406098480767158,
-      "loss": 3.241,
+      "loss": 3.2372,
       "step": 72200
     },
     {
       "epoch": 7.776342697233883,
-      "grad_norm": 0.794816255569458,
+      "grad_norm": 0.8180352449417114,
       "learning_rate": 0.00013373774377761015,
-      "loss": 3.2199,
+      "loss": 3.2418,
       "step": 72250
     },
     {
       "epoch": 7.78172424927349,
-      "grad_norm": 0.8048193454742432,
+      "grad_norm": 0.799294650554657,
       "learning_rate": 0.00013341450274754874,
-      "loss": 3.2403,
+      "loss": 3.232,
       "step": 72300
     },
     {
       "epoch": 7.787105801313099,
-      "grad_norm": 0.8669823408126831,
+      "grad_norm": 0.8253015279769897,
       "learning_rate": 0.0001330912617174873,
-      "loss": 3.2492,
+      "loss": 3.2542,
       "step": 72350
     },
     {
       "epoch": 7.792487353352707,
-      "grad_norm": 0.8714507818222046,
+      "grad_norm": 0.8519314527511597,
       "learning_rate": 0.0001327680206874259,
-      "loss": 3.2362,
+      "loss": 3.2296,
       "step": 72400
     },
     {
       "epoch": 7.797868905392315,
-      "grad_norm": 0.7854097485542297,
+      "grad_norm": 0.8551774621009827,
       "learning_rate": 0.0001324447796573645,
-      "loss": 3.2506,
+      "loss": 3.2499,
       "step": 72450
     },
     {
       "epoch": 7.803250457431924,
-      "grad_norm": 0.8088889122009277,
+      "grad_norm": 0.8259185552597046,
       "learning_rate": 0.00013212153862730307,
-      "loss": 3.2229,
+      "loss": 3.2461,
       "step": 72500
     },
     {
       "epoch": 7.808632009471531,
-      "grad_norm": 0.8250226378440857,
+      "grad_norm": 0.8493896126747131,
       "learning_rate": 0.00013179829759724166,
-      "loss": 3.2271,
+      "loss": 3.2342,
       "step": 72550
     },
     {
       "epoch": 7.81401356151114,
-      "grad_norm": 0.7839990854263306,
+      "grad_norm": 0.8167123794555664,
       "learning_rate": 0.00013147505656718026,
-      "loss": 3.2476,
+      "loss": 3.2211,
       "step": 72600
     },
     {
       "epoch": 7.819395113550748,
-      "grad_norm": 0.8592643141746521,
+      "grad_norm": 0.9556056261062622,
       "learning_rate": 0.00013115181553711883,
-      "loss": 3.2563,
+      "loss": 3.2345,
       "step": 72650
     },
     {
       "epoch": 7.824776665590356,
-      "grad_norm": 0.834892749786377,
+      "grad_norm": 0.809105634689331,
       "learning_rate": 0.00013082857450705742,
-      "loss": 3.2486,
+      "loss": 3.2488,
       "step": 72700
     },
     {
       "epoch": 7.830158217629965,
-      "grad_norm": 0.8037497401237488,
+      "grad_norm": 0.8134852647781372,
       "learning_rate": 0.00013050533347699602,
-      "loss": 3.2666,
+      "loss": 3.2357,
       "step": 72750
     },
     {
       "epoch": 7.835539769669572,
-      "grad_norm": 0.7950841188430786,
+      "grad_norm": 0.8138014674186707,
       "learning_rate": 0.00013018209244693458,
-      "loss": 3.2504,
+      "loss": 3.2466,
       "step": 72800
     },
     {
       "epoch": 7.840921321709181,
-      "grad_norm": 0.868694007396698,
+      "grad_norm": 0.9054873585700989,
       "learning_rate": 0.00012985885141687318,
-      "loss": 3.2259,
+      "loss": 3.2214,
       "step": 72850
     },
     {
       "epoch": 7.846302873748789,
-      "grad_norm": 0.8432836532592773,
+      "grad_norm": 0.8494545221328735,
       "learning_rate": 0.00012953561038681175,
-      "loss": 3.2525,
+      "loss": 3.2695,
       "step": 72900
     },
     {
       "epoch": 7.851684425788397,
-      "grad_norm": 0.8124203681945801,
+      "grad_norm": 0.8666238784790039,
       "learning_rate": 0.00012921236935675034,
-      "loss": 3.2448,
+      "loss": 3.2267,
       "step": 72950
     },
     {
       "epoch": 7.857065977828006,
-      "grad_norm": 0.8090877532958984,
+      "grad_norm": 0.8144758939743042,
       "learning_rate": 0.0001288891283266889,
-      "loss": 3.2356,
+      "loss": 3.2271,
       "step": 73000
     },
     {
       "epoch": 7.857065977828006,
-      "eval_accuracy": 0.38977637171282914,
-      "eval_loss": 3.336695671081543,
-      "eval_runtime": 180.1915,
-      "eval_samples_per_second": 99.955,
-      "eval_steps_per_second": 6.249,
+      "eval_accuracy": 0.39005289331231696,
+      "eval_loss": 3.331833839416504,
+      "eval_runtime": 181.3725,
+      "eval_samples_per_second": 99.304,
+      "eval_steps_per_second": 6.208,
       "step": 73000
     },
     {
       "epoch": 7.862447529867614,
-      "grad_norm": 0.8222204446792603,
+      "grad_norm": 0.7968391180038452,
       "learning_rate": 0.0001285658872966275,
-      "loss": 3.2259,
+      "loss": 3.2291,
       "step": 73050
     },
     {
       "epoch": 7.867829081907222,
-      "grad_norm": 0.8175197839736938,
+      "grad_norm": 0.8024161458015442,
       "learning_rate": 0.0001282426462665661,
-      "loss": 3.2454,
+      "loss": 3.2354,
       "step": 73100
     },
     {
       "epoch": 7.87321063394683,
-      "grad_norm": 0.7853533625602722,
+      "grad_norm": 0.7787240743637085,
       "learning_rate": 0.00012791940523650466,
-      "loss": 3.2267,
+      "loss": 3.2297,
       "step": 73150
     },
     {
       "epoch": 7.878592185986438,
-      "grad_norm": 0.8725240230560303,
+      "grad_norm": 0.852200984954834,
       "learning_rate": 0.00012759616420644326,
-      "loss": 3.2317,
+      "loss": 3.2137,
       "step": 73200
     },
     {
       "epoch": 7.883973738026047,
-      "grad_norm": 0.8071550726890564,
+      "grad_norm": 0.820639967918396,
       "learning_rate": 0.00012727292317638185,
-      "loss": 3.2218,
+      "loss": 3.2247,
       "step": 73250
     },
     {
       "epoch": 7.889355290065655,
-      "grad_norm": 0.8475930094718933,
+      "grad_norm": 0.810571551322937,
       "learning_rate": 0.00012694968214632045,
-      "loss": 3.2259,
+      "loss": 3.2165,
       "step": 73300
     },
     {
       "epoch": 7.894736842105263,
-      "grad_norm": 0.8381208181381226,
+      "grad_norm": 0.8381794095039368,
       "learning_rate": 0.00012662644111625902,
-      "loss": 3.2366,
+      "loss": 3.2344,
       "step": 73350
     },
     {
       "epoch": 7.900118394144871,
-      "grad_norm": 0.809328019618988,
+      "grad_norm": 0.8724242448806763,
       "learning_rate": 0.0001263032000861976,
-      "loss": 3.2301,
+      "loss": 3.2473,
       "step": 73400
     },
     {
       "epoch": 7.90549994618448,
-      "grad_norm": 0.8766582608222961,
+      "grad_norm": 0.8407784104347229,
       "learning_rate": 0.00012597995905613618,
-      "loss": 3.2601,
+      "loss": 3.2332,
       "step": 73450
     },
     {
       "epoch": 7.910881498224088,
-      "grad_norm": 0.8222476840019226,
+      "grad_norm": 0.7963847517967224,
       "learning_rate": 0.00012565671802607477,
-      "loss": 3.2429,
+      "loss": 3.2548,
       "step": 73500
     },
     {
       "epoch": 7.916263050263696,
-      "grad_norm": 0.8451269268989563,
+      "grad_norm": 0.8477411270141602,
       "learning_rate": 0.00012533347699601334,
-      "loss": 3.2484,
+      "loss": 3.2359,
       "step": 73550
     },
     {
       "epoch": 7.921644602303305,
-      "grad_norm": 0.8155948519706726,
+      "grad_norm": 0.808516800403595,
       "learning_rate": 0.00012501023596595193,
-      "loss": 3.2388,
+      "loss": 3.2396,
       "step": 73600
     },
     {
       "epoch": 7.927026154342912,
-      "grad_norm": 0.8491314053535461,
+      "grad_norm": 0.8295881748199463,
       "learning_rate": 0.0001246869949358905,
-      "loss": 3.2469,
+      "loss": 3.2476,
       "step": 73650
     },
     {
       "epoch": 7.932407706382521,
-      "grad_norm": 0.8263399600982666,
+      "grad_norm": 0.850948691368103,
       "learning_rate": 0.0001243637539058291,
-      "loss": 3.2575,
+      "loss": 3.2378,
       "step": 73700
     },
     {
       "epoch": 7.937789258422129,
-      "grad_norm": 0.8386456966400146,
+      "grad_norm": 0.8757145404815674,
       "learning_rate": 0.0001240405128757677,
-      "loss": 3.2509,
+      "loss": 3.2401,
       "step": 73750
     },
     {
       "epoch": 7.943170810461737,
-      "grad_norm": 0.7784790396690369,
+      "grad_norm": 0.8073046207427979,
       "learning_rate": 0.00012371727184570629,
-      "loss": 3.2274,
+      "loss": 3.223,
       "step": 73800
     },
     {
       "epoch": 7.948552362501346,
-      "grad_norm": 0.8531802892684937,
+      "grad_norm": 0.8661218285560608,
       "learning_rate": 0.00012339403081564485,
-      "loss": 3.2205,
+      "loss": 3.2168,
       "step": 73850
     },
     {
       "epoch": 7.953933914540953,
-      "grad_norm": 0.8842747211456299,
+      "grad_norm": 0.8399752974510193,
       "learning_rate": 0.00012307078978558345,
-      "loss": 3.2316,
+      "loss": 3.2398,
       "step": 73900
     },
     {
       "epoch": 7.959315466580562,
-      "grad_norm": 0.8715245723724365,
+      "grad_norm": 0.8319146037101746,
       "learning_rate": 0.00012274754875552202,
-      "loss": 3.2406,
+      "loss": 3.2288,
       "step": 73950
     },
     {
       "epoch": 7.96469701862017,
-      "grad_norm": 0.8610970973968506,
+      "grad_norm": 0.8275104761123657,
       "learning_rate": 0.0001224243077254606,
-      "loss": 3.2443,
+      "loss": 3.2537,
       "step": 74000
     },
     {
       "epoch": 7.96469701862017,
-      "eval_accuracy": 0.39020728906590135,
-      "eval_loss": 3.335205554962158,
-      "eval_runtime": 180.1463,
-      "eval_samples_per_second": 99.98,
-      "eval_steps_per_second": 6.25,
+      "eval_accuracy": 0.39038547980191707,
+      "eval_loss": 3.3268861770629883,
+      "eval_runtime": 181.6459,
+      "eval_samples_per_second": 99.154,
+      "eval_steps_per_second": 6.199,
       "step": 74000
     },
     {
       "epoch": 7.970078570659778,
-      "grad_norm": 0.8783667087554932,
+      "grad_norm": 0.9003944993019104,
       "learning_rate": 0.00012210106669539918,
-      "loss": 3.239,
+      "loss": 3.2313,
       "step": 74050
     },
     {
       "epoch": 7.975460122699387,
-      "grad_norm": 0.8203068971633911,
+      "grad_norm": 0.8478912711143494,
       "learning_rate": 0.00012177782566533779,
-      "loss": 3.2352,
+      "loss": 3.2065,
       "step": 74100
     },
     {
       "epoch": 7.980841674738995,
-      "grad_norm": 0.8619286417961121,
+      "grad_norm": 0.8998082280158997,
       "learning_rate": 0.00012145458463527635,
-      "loss": 3.2383,
+      "loss": 3.2452,
       "step": 74150
     },
     {
       "epoch": 7.986223226778603,
-      "grad_norm": 0.8609179854393005,
+      "grad_norm": 0.8041027784347534,
       "learning_rate": 0.00012113134360521495,
-      "loss": 3.248,
+      "loss": 3.2296,
       "step": 74200
     },
     {
       "epoch": 7.991604778818211,
-      "grad_norm": 0.8277957439422607,
-      "learning_rate": 0.00012080810257515353,
-      "loss": 3.2689,
+      "grad_norm": 0.8321233987808228,
+      "learning_rate": 0.00012081456739575477,
+      "loss": 3.2551,
       "step": 74250
     },
     {
       "epoch": 7.996986330857819,
-      "grad_norm": 0.8138071894645691,
-      "learning_rate": 0.00012048486154509212,
-      "loss": 3.2372,
+      "grad_norm": 0.839951753616333,
+      "learning_rate": 0.00012049132636569334,
+      "loss": 3.2304,
       "step": 74300
     },
     {
       "epoch": 8.002367882897428,
-      "grad_norm": 0.8171045184135437,
-      "learning_rate": 0.00012016162051503069,
-      "loss": 3.2023,
+      "grad_norm": 0.8819961547851562,
+      "learning_rate": 0.00012016808533563193,
+      "loss": 3.1913,
       "step": 74350
     },
     {
       "epoch": 8.007749434937036,
-      "grad_norm": 0.884177029132843,
-      "learning_rate": 0.00011983837948496929,
-      "loss": 3.1379,
+      "grad_norm": 0.8040395379066467,
+      "learning_rate": 0.00011984484430557051,
+      "loss": 3.1398,
       "step": 74400
     },
     {
       "epoch": 8.013130986976645,
-      "grad_norm": 0.8275094032287598,
+      "grad_norm": 0.8497294187545776,
       "learning_rate": 0.0001195216032755091,
-      "loss": 3.1535,
+      "loss": 3.1587,
       "step": 74450
     },
     {
       "epoch": 8.018512539016251,
-      "grad_norm": 0.8046960830688477,
+      "grad_norm": 0.8274456262588501,
       "learning_rate": 0.00011919836224544767,
-      "loss": 3.1526,
+      "loss": 3.1537,
       "step": 74500
     },
     {
       "epoch": 8.02389409105586,
-      "grad_norm": 0.8428935408592224,
+      "grad_norm": 0.8324849009513855,
       "learning_rate": 0.00011887512121538627,
-      "loss": 3.1598,
+      "loss": 3.1642,
       "step": 74550
     },
     {
       "epoch": 8.029275643095469,
-      "grad_norm": 0.8413782715797424,
+      "grad_norm": 0.7917704582214355,
       "learning_rate": 0.00011855188018532485,
-      "loss": 3.1582,
+      "loss": 3.1591,
       "step": 74600
     },
     {
       "epoch": 8.034657195135077,
-      "grad_norm": 0.8058221340179443,
+      "grad_norm": 0.8039134740829468,
       "learning_rate": 0.00011822863915526343,
-      "loss": 3.1563,
+      "loss": 3.1394,
       "step": 74650
     },
     {
       "epoch": 8.040038747174686,
-      "grad_norm": 0.8850978016853333,
+      "grad_norm": 0.879476010799408,
       "learning_rate": 0.00011790539812520201,
-      "loss": 3.1662,
+      "loss": 3.176,
       "step": 74700
     },
     {
       "epoch": 8.045420299214294,
-      "grad_norm": 0.8446381688117981,
+      "grad_norm": 0.8532137870788574,
       "learning_rate": 0.00011758215709514061,
-      "loss": 3.1713,
+      "loss": 3.1676,
       "step": 74750
     },
     {
       "epoch": 8.050801851253901,
-      "grad_norm": 0.8259122967720032,
+      "grad_norm": 0.7909590601921082,
       "learning_rate": 0.00011725891606507917,
-      "loss": 3.1745,
+      "loss": 3.1762,
       "step": 74800
     },
     {
       "epoch": 8.05618340329351,
-      "grad_norm": 0.8464698195457458,
+      "grad_norm": 0.850517988204956,
       "learning_rate": 0.00011693567503501777,
-      "loss": 3.1503,
+      "loss": 3.1247,
       "step": 74850
     },
     {
       "epoch": 8.061564955333118,
-      "grad_norm": 0.841551661491394,
+      "grad_norm": 0.8225530385971069,
       "learning_rate": 0.00011661243400495635,
-      "loss": 3.1633,
+      "loss": 3.1594,
       "step": 74900
     },
     {
       "epoch": 8.066946507372727,
-      "grad_norm": 0.8218307495117188,
+      "grad_norm": 0.8351697325706482,
       "learning_rate": 0.00011628919297489493,
-      "loss": 3.1706,
+      "loss": 3.1921,
       "step": 74950
     },
     {
       "epoch": 8.072328059412335,
-      "grad_norm": 0.8135493993759155,
+      "grad_norm": 0.8439159393310547,
       "learning_rate": 0.00011596595194483351,
-      "loss": 3.1627,
+      "loss": 3.1659,
       "step": 75000
     },
     {
       "epoch": 8.072328059412335,
-      "eval_accuracy": 0.3899024090626939,
-      "eval_loss": 3.3387045860290527,
-      "eval_runtime": 180.3237,
-      "eval_samples_per_second": 99.881,
-      "eval_steps_per_second": 6.244,
+      "eval_accuracy": 0.39026357126178923,
+      "eval_loss": 3.333638906478882,
+      "eval_runtime": 181.7813,
+      "eval_samples_per_second": 99.081,
+      "eval_steps_per_second": 6.194,
       "step": 75000
     },
     {
       "epoch": 8.077709611451942,
-      "grad_norm": 0.8384975790977478,
+      "grad_norm": 0.8241596817970276,
       "learning_rate": 0.00011564271091477211,
-      "loss": 3.1987,
+      "loss": 3.1838,
       "step": 75050
     },
     {
       "epoch": 8.08309116349155,
-      "grad_norm": 0.8627749681472778,
+      "grad_norm": 0.8463574051856995,
       "learning_rate": 0.0001153194698847107,
-      "loss": 3.1645,
+      "loss": 3.1655,
       "step": 75100
     },
     {
       "epoch": 8.088472715531159,
-      "grad_norm": 0.8553587794303894,
+      "grad_norm": 0.810834527015686,
       "learning_rate": 0.00011499622885464927,
-      "loss": 3.1827,
+      "loss": 3.1723,
       "step": 75150
     },
     {
       "epoch": 8.093854267570768,
-      "grad_norm": 0.8322249054908752,
+      "grad_norm": 0.914654016494751,
       "learning_rate": 0.00011467298782458786,
-      "loss": 3.1613,
+      "loss": 3.1619,
       "step": 75200
     },
     {
       "epoch": 8.099235819610376,
-      "grad_norm": 0.858957827091217,
+      "grad_norm": 0.8331172466278076,
       "learning_rate": 0.00011434974679452645,
-      "loss": 3.156,
+      "loss": 3.1524,
       "step": 75250
     },
     {
       "epoch": 8.104617371649983,
-      "grad_norm": 0.8273110389709473,
+      "grad_norm": 0.8197857737541199,
       "learning_rate": 0.00011402650576446503,
-      "loss": 3.1704,
+      "loss": 3.1628,
       "step": 75300
     },
     {
       "epoch": 8.109998923689592,
-      "grad_norm": 0.829047441482544,
+      "grad_norm": 0.8444637060165405,
       "learning_rate": 0.00011370326473440361,
-      "loss": 3.1814,
+      "loss": 3.1808,
       "step": 75350
     },
     {
       "epoch": 8.1153804757292,
-      "grad_norm": 0.8540376424789429,
+      "grad_norm": 0.8376138806343079,
       "learning_rate": 0.0001133800237043422,
-      "loss": 3.172,
+      "loss": 3.1719,
       "step": 75400
     },
     {
       "epoch": 8.120762027768809,
-      "grad_norm": 0.8101136684417725,
+      "grad_norm": 0.7956539988517761,
       "learning_rate": 0.00011305678267428077,
-      "loss": 3.17,
+      "loss": 3.1568,
       "step": 75450
     },
     {
       "epoch": 8.126143579808417,
-      "grad_norm": 0.8184477686882019,
+      "grad_norm": 0.8370229005813599,
       "learning_rate": 0.00011273354164421936,
-      "loss": 3.1818,
+      "loss": 3.1747,
       "step": 75500
     },
     {
       "epoch": 8.131525131848026,
-      "grad_norm": 0.8428223133087158,
+      "grad_norm": 0.8129817247390747,
       "learning_rate": 0.00011241030061415795,
-      "loss": 3.1718,
+      "loss": 3.1657,
       "step": 75550
     },
     {
       "epoch": 8.136906683887632,
-      "grad_norm": 0.8273821473121643,
+      "grad_norm": 0.835895836353302,
       "learning_rate": 0.00011208705958409654,
-      "loss": 3.1799,
+      "loss": 3.1671,
       "step": 75600
     },
     {
       "epoch": 8.142288235927241,
-      "grad_norm": 0.8538662195205688,
+      "grad_norm": 0.8597838878631592,
       "learning_rate": 0.00011176381855403511,
-      "loss": 3.1665,
+      "loss": 3.161,
       "step": 75650
     },
     {
       "epoch": 8.14766978796685,
-      "grad_norm": 0.8688125014305115,
+      "grad_norm": 0.8241186141967773,
       "learning_rate": 0.0001114405775239737,
-      "loss": 3.17,
+      "loss": 3.1771,
       "step": 75700
     },
     {
       "epoch": 8.153051340006458,
-      "grad_norm": 0.8661782145500183,
+      "grad_norm": 0.8637716770172119,
       "learning_rate": 0.00011111733649391228,
-      "loss": 3.174,
+      "loss": 3.1776,
       "step": 75750
     },
     {
       "epoch": 8.158432892046067,
-      "grad_norm": 0.8403778672218323,
+      "grad_norm": 0.79430091381073,
       "learning_rate": 0.00011079409546385086,
-      "loss": 3.1963,
+      "loss": 3.1832,
       "step": 75800
     },
     {
       "epoch": 8.163814444085673,
-      "grad_norm": 0.8399979472160339,
+      "grad_norm": 0.810520589351654,
       "learning_rate": 0.00011047085443378945,
-      "loss": 3.1708,
+      "loss": 3.1742,
       "step": 75850
     },
     {
       "epoch": 8.169195996125282,
-      "grad_norm": 0.8141344785690308,
+      "grad_norm": 0.8526862263679504,
       "learning_rate": 0.00011014761340372804,
-      "loss": 3.1854,
+      "loss": 3.1835,
       "step": 75900
     },
     {
       "epoch": 8.17457754816489,
-      "grad_norm": 0.8441236019134521,
+      "grad_norm": 0.8866553902626038,
       "learning_rate": 0.00010982437237366661,
-      "loss": 3.1724,
+      "loss": 3.166,
       "step": 75950
     },
     {
       "epoch": 8.1799591002045,
-      "grad_norm": 0.8193464875221252,
+      "grad_norm": 0.8259860277175903,
       "learning_rate": 0.0001095011313436052,
-      "loss": 3.1778,
+      "loss": 3.1683,
       "step": 76000
     },
     {
       "epoch": 8.1799591002045,
-      "eval_accuracy": 0.3904291582628185,
-      "eval_loss": 3.335801839828491,
-      "eval_runtime": 180.1521,
-      "eval_samples_per_second": 99.977,
-      "eval_steps_per_second": 6.25,
+      "eval_accuracy": 0.390622125791577,
+      "eval_loss": 3.3311233520507812,
+      "eval_runtime": 181.5127,
+      "eval_samples_per_second": 99.227,
+      "eval_steps_per_second": 6.203,
       "step": 76000
     },
     {
       "epoch": 8.185340652244108,
-      "grad_norm": 0.8355197906494141,
+      "grad_norm": 0.8608167171478271,
       "learning_rate": 0.0001091778903135438,
-      "loss": 3.1686,
+      "loss": 3.1721,
       "step": 76050
     },
     {
       "epoch": 8.190722204283716,
-      "grad_norm": 0.8153267502784729,
+      "grad_norm": 0.8262567520141602,
       "learning_rate": 0.00010885464928348238,
-      "loss": 3.1838,
+      "loss": 3.1615,
       "step": 76100
     },
     {
       "epoch": 8.196103756323323,
-      "grad_norm": 0.8740131258964539,
+      "grad_norm": 0.8556413054466248,
       "learning_rate": 0.00010853140825342096,
-      "loss": 3.1696,
+      "loss": 3.1743,
       "step": 76150
     },
     {
       "epoch": 8.201485308362932,
-      "grad_norm": 0.8934776782989502,
+      "grad_norm": 0.8432217240333557,
       "learning_rate": 0.00010820816722335954,
-      "loss": 3.1742,
+      "loss": 3.156,
       "step": 76200
     },
     {
       "epoch": 8.20686686040254,
-      "grad_norm": 0.9179877042770386,
+      "grad_norm": 0.8617128133773804,
       "learning_rate": 0.00010788492619329814,
-      "loss": 3.1695,
+      "loss": 3.1779,
       "step": 76250
     },
     {
       "epoch": 8.212248412442149,
-      "grad_norm": 0.8545275926589966,
+      "grad_norm": 0.8448604941368103,
       "learning_rate": 0.0001075616851632367,
-      "loss": 3.172,
+      "loss": 3.1801,
       "step": 76300
     },
     {
       "epoch": 8.217629964481757,
-      "grad_norm": 0.8500077724456787,
+      "grad_norm": 0.8647608160972595,
       "learning_rate": 0.0001072384441331753,
-      "loss": 3.1938,
+      "loss": 3.1902,
       "step": 76350
     },
     {
       "epoch": 8.223011516521364,
-      "grad_norm": 0.8435236811637878,
+      "grad_norm": 0.8533148169517517,
       "learning_rate": 0.00010691520310311388,
-      "loss": 3.1834,
+      "loss": 3.1765,
       "step": 76400
     },
     {
       "epoch": 8.228393068560973,
-      "grad_norm": 0.8567854762077332,
+      "grad_norm": 0.8908019065856934,
       "learning_rate": 0.00010659196207305246,
-      "loss": 3.169,
+      "loss": 3.174,
       "step": 76450
     },
     {
       "epoch": 8.233774620600581,
-      "grad_norm": 0.8797134757041931,
+      "grad_norm": 0.8582380414009094,
       "learning_rate": 0.00010626872104299104,
-      "loss": 3.1859,
+      "loss": 3.1617,
       "step": 76500
     },
     {
       "epoch": 8.23915617264019,
-      "grad_norm": 0.8667067289352417,
+      "grad_norm": 0.8442745804786682,
       "learning_rate": 0.00010594548001292964,
-      "loss": 3.1973,
+      "loss": 3.1729,
       "step": 76550
     },
     {
       "epoch": 8.244537724679798,
-      "grad_norm": 0.8613109588623047,
+      "grad_norm": 0.8726164102554321,
       "learning_rate": 0.00010562223898286823,
-      "loss": 3.1941,
+      "loss": 3.1923,
       "step": 76600
     },
     {
       "epoch": 8.249919276719407,
-      "grad_norm": 0.8120648264884949,
+      "grad_norm": 0.8388357758522034,
       "learning_rate": 0.0001052989979528068,
-      "loss": 3.1899,
+      "loss": 3.1809,
       "step": 76650
     },
     {
       "epoch": 8.255300828759013,
-      "grad_norm": 0.8621121048927307,
+      "grad_norm": 0.8465825915336609,
       "learning_rate": 0.00010497575692274539,
-      "loss": 3.1836,
+      "loss": 3.1812,
       "step": 76700
     },
     {
       "epoch": 8.260682380798622,
-      "grad_norm": 0.8764901757240295,
+      "grad_norm": 0.8217101693153381,
       "learning_rate": 0.00010465251589268397,
-      "loss": 3.2011,
+      "loss": 3.1871,
       "step": 76750
     },
     {
       "epoch": 8.26606393283823,
-      "grad_norm": 0.8889468312263489,
+      "grad_norm": 0.9329084753990173,
       "learning_rate": 0.00010432927486262254,
-      "loss": 3.1871,
+      "loss": 3.1706,
       "step": 76800
     },
     {
       "epoch": 8.27144548487784,
-      "grad_norm": 0.8703168630599976,
+      "grad_norm": 0.860842764377594,
       "learning_rate": 0.00010400603383256114,
-      "loss": 3.1714,
+      "loss": 3.168,
       "step": 76850
     },
     {
       "epoch": 8.276827036917448,
-      "grad_norm": 0.8331544399261475,
+      "grad_norm": 0.8332845568656921,
       "learning_rate": 0.00010368279280249973,
-      "loss": 3.1679,
+      "loss": 3.1895,
       "step": 76900
     },
     {
       "epoch": 8.282208588957054,
-      "grad_norm": 0.8356614708900452,
+      "grad_norm": 0.8840993642807007,
       "learning_rate": 0.0001033595517724383,
-      "loss": 3.187,
+      "loss": 3.188,
       "step": 76950
     },
     {
       "epoch": 8.287590140996663,
-      "grad_norm": 0.9109562039375305,
+      "grad_norm": 0.9360073804855347,
       "learning_rate": 0.00010303631074237689,
-      "loss": 3.1902,
+      "loss": 3.1794,
       "step": 77000
     },
     {
       "epoch": 8.287590140996663,
-      "eval_accuracy": 0.3904268765521744,
-      "eval_loss": 3.3348872661590576,
-      "eval_runtime": 180.23,
-      "eval_samples_per_second": 99.933,
-      "eval_steps_per_second": 6.248,
+      "eval_accuracy": 0.3908713755162235,
+      "eval_loss": 3.3284499645233154,
+      "eval_runtime": 181.5852,
+      "eval_samples_per_second": 99.188,
+      "eval_steps_per_second": 6.201,
       "step": 77000
     },
     {
       "epoch": 8.292971693036272,
-      "grad_norm": 0.857496440410614,
+      "grad_norm": 0.8750916123390198,
       "learning_rate": 0.00010271306971231547,
-      "loss": 3.1796,
+      "loss": 3.1811,
       "step": 77050
     },
     {
       "epoch": 8.29835324507588,
-      "grad_norm": 0.8897091746330261,
+      "grad_norm": 0.8605511784553528,
       "learning_rate": 0.00010238982868225407,
-      "loss": 3.195,
+      "loss": 3.1687,
       "step": 77100
     },
     {
       "epoch": 8.303734797115489,
-      "grad_norm": 0.8627259135246277,
+      "grad_norm": 0.8794077038764954,
       "learning_rate": 0.00010206658765219264,
-      "loss": 3.1633,
+      "loss": 3.1757,
       "step": 77150
     },
     {
       "epoch": 8.309116349155097,
-      "grad_norm": 0.8576809167861938,
+      "grad_norm": 0.8809419274330139,
       "learning_rate": 0.00010174334662213123,
-      "loss": 3.178,
+      "loss": 3.1749,
       "step": 77200
     },
     {
       "epoch": 8.314497901194704,
-      "grad_norm": 0.913425862789154,
+      "grad_norm": 0.8522841930389404,
       "learning_rate": 0.00010142010559206981,
-      "loss": 3.2039,
+      "loss": 3.1894,
       "step": 77250
     },
     {
       "epoch": 8.319879453234313,
-      "grad_norm": 0.8628541231155396,
+      "grad_norm": 0.843936026096344,
       "learning_rate": 0.00010109686456200839,
-      "loss": 3.1867,
+      "loss": 3.1796,
       "step": 77300
     },
     {
       "epoch": 8.325261005273921,
-      "grad_norm": 0.8340932726860046,
+      "grad_norm": 0.900905191898346,
       "learning_rate": 0.00010077362353194697,
-      "loss": 3.1896,
+      "loss": 3.1674,
       "step": 77350
     },
     {
       "epoch": 8.33064255731353,
-      "grad_norm": 0.8404185175895691,
+      "grad_norm": 0.8734228610992432,
       "learning_rate": 0.00010045038250188557,
-      "loss": 3.1841,
+      "loss": 3.1932,
       "step": 77400
     },
     {
       "epoch": 8.336024109353138,
-      "grad_norm": 0.863701343536377,
+      "grad_norm": 0.8321887850761414,
       "learning_rate": 0.00010012714147182414,
-      "loss": 3.1791,
+      "loss": 3.1787,
       "step": 77450
     },
     {
       "epoch": 8.341405661392745,
-      "grad_norm": 0.859533965587616,
+      "grad_norm": 0.8625351786613464,
       "learning_rate": 9.980390044176273e-05,
-      "loss": 3.1825,
+      "loss": 3.1798,
       "step": 77500
     },
     {
       "epoch": 8.346787213432354,
-      "grad_norm": 0.8829686641693115,
+      "grad_norm": 0.8561045527458191,
       "learning_rate": 9.948065941170133e-05,
-      "loss": 3.1839,
+      "loss": 3.1823,
       "step": 77550
     },
     {
       "epoch": 8.352168765471962,
-      "grad_norm": 0.8447333574295044,
+      "grad_norm": 0.8517137169837952,
       "learning_rate": 9.91574183816399e-05,
-      "loss": 3.1852,
+      "loss": 3.1843,
       "step": 77600
     },
     {
       "epoch": 8.35755031751157,
-      "grad_norm": 0.8658083081245422,
+      "grad_norm": 0.852493405342102,
       "learning_rate": 9.883417735157849e-05,
-      "loss": 3.1787,
+      "loss": 3.1831,
       "step": 77650
     },
     {
       "epoch": 8.36293186955118,
-      "grad_norm": 0.8201994895935059,
+      "grad_norm": 0.8284958600997925,
       "learning_rate": 9.851093632151707e-05,
-      "loss": 3.187,
+      "loss": 3.1779,
       "step": 77700
     },
     {
       "epoch": 8.368313421590786,
-      "grad_norm": 0.8592942953109741,
+      "grad_norm": 0.8786062002182007,
       "learning_rate": 9.818769529145566e-05,
-      "loss": 3.1745,
+      "loss": 3.1756,
       "step": 77750
     },
     {
       "epoch": 8.373694973630395,
-      "grad_norm": 0.8522955775260925,
+      "grad_norm": 0.8092980980873108,
       "learning_rate": 9.786445426139423e-05,
-      "loss": 3.1748,
+      "loss": 3.176,
       "step": 77800
     },
     {
       "epoch": 8.379076525670003,
-      "grad_norm": 0.8451487421989441,
+      "grad_norm": 0.8757535815238953,
       "learning_rate": 9.754121323133283e-05,
-      "loss": 3.1835,
+      "loss": 3.1847,
       "step": 77850
     },
     {
       "epoch": 8.384458077709612,
-      "grad_norm": 0.8497421145439148,
+      "grad_norm": 0.8386533856391907,
       "learning_rate": 9.72179722012714e-05,
-      "loss": 3.186,
+      "loss": 3.1662,
       "step": 77900
     },
     {
       "epoch": 8.38983962974922,
-      "grad_norm": 0.8562092781066895,
+      "grad_norm": 0.8449146151542664,
       "learning_rate": 9.689473117120999e-05,
-      "loss": 3.1989,
+      "loss": 3.1819,
       "step": 77950
     },
     {
       "epoch": 8.395221181788829,
-      "grad_norm": 0.8583953380584717,
+      "grad_norm": 0.9429228901863098,
       "learning_rate": 9.657149014114857e-05,
-      "loss": 3.1921,
+      "loss": 3.1874,
       "step": 78000
     },
     {
       "epoch": 8.395221181788829,
-      "eval_accuracy": 0.3905158632672945,
-      "eval_loss": 3.331059694290161,
-      "eval_runtime": 180.1209,
-      "eval_samples_per_second": 99.994,
-      "eval_steps_per_second": 6.251,
+      "eval_accuracy": 0.39114072602511557,
+      "eval_loss": 3.326591968536377,
+      "eval_runtime": 181.7422,
+      "eval_samples_per_second": 99.102,
+      "eval_steps_per_second": 6.196,
       "step": 78000
     },
     {
       "epoch": 8.400602733828435,
-      "grad_norm": 0.860522449016571,
+      "grad_norm": 0.8306030631065369,
       "learning_rate": 9.624824911108716e-05,
-      "loss": 3.1781,
+      "loss": 3.1811,
       "step": 78050
     },
     {
       "epoch": 8.405984285868044,
-      "grad_norm": 0.875423014163971,
+      "grad_norm": 0.8511099815368652,
       "learning_rate": 9.592500808102574e-05,
-      "loss": 3.172,
+      "loss": 3.1617,
       "step": 78100
     },
     {
       "epoch": 8.411365837907653,
-      "grad_norm": 0.8359448313713074,
+      "grad_norm": 0.8214622735977173,
       "learning_rate": 9.560176705096433e-05,
-      "loss": 3.1911,
+      "loss": 3.1762,
       "step": 78150
     },
     {
       "epoch": 8.416747389947261,
-      "grad_norm": 0.8645302653312683,
+      "grad_norm": 0.9175541400909424,
       "learning_rate": 9.52785260209029e-05,
-      "loss": 3.1773,
+      "loss": 3.1681,
       "step": 78200
     },
     {
       "epoch": 8.42212894198687,
-      "grad_norm": 0.8695240616798401,
-      "learning_rate": 9.49552849908415e-05,
-      "loss": 3.1858,
+      "grad_norm": 0.8537863492965698,
+      "learning_rate": 9.496174981144271e-05,
+      "loss": 3.1746,
       "step": 78250
     },
     {
       "epoch": 8.427510494026476,
-      "grad_norm": 0.8338247537612915,
-      "learning_rate": 9.463204396078007e-05,
-      "loss": 3.1981,
+      "grad_norm": 0.8496671319007874,
+      "learning_rate": 9.463850878138131e-05,
+      "loss": 3.1936,
       "step": 78300
     },
     {
       "epoch": 8.432892046066085,
-      "grad_norm": 0.9113908410072327,
-      "learning_rate": 9.430880293071866e-05,
-      "loss": 3.1838,
+      "grad_norm": 0.9404656887054443,
+      "learning_rate": 9.431526775131989e-05,
+      "loss": 3.1868,
       "step": 78350
     },
     {
       "epoch": 8.438273598105694,
-      "grad_norm": 0.853624165058136,
-      "learning_rate": 9.398556190065726e-05,
-      "loss": 3.1826,
+      "grad_norm": 0.8471515774726868,
+      "learning_rate": 9.399202672125848e-05,
+      "loss": 3.1784,
       "step": 78400
     },
     {
       "epoch": 8.443655150145302,
-      "grad_norm": 0.8597407341003418,
+      "grad_norm": 0.8557966947555542,
       "learning_rate": 9.366878569119705e-05,
-      "loss": 3.1836,
+      "loss": 3.1839,
       "step": 78450
     },
     {
       "epoch": 8.44903670218491,
-      "grad_norm": 0.8677853941917419,
+      "grad_norm": 0.8111828565597534,
       "learning_rate": 9.334554466113565e-05,
-      "loss": 3.1936,
+      "loss": 3.1989,
       "step": 78500
     },
     {
       "epoch": 8.45441825422452,
-      "grad_norm": 0.938667893409729,
+      "grad_norm": 0.8373623490333557,
       "learning_rate": 9.302230363107424e-05,
-      "loss": 3.194,
+      "loss": 3.1803,
       "step": 78550
     },
     {
       "epoch": 8.459799806264126,
-      "grad_norm": 0.8398536443710327,
+      "grad_norm": 0.8651726841926575,
       "learning_rate": 9.269906260101281e-05,
-      "loss": 3.1795,
+      "loss": 3.1937,
       "step": 78600
     },
     {
       "epoch": 8.465181358303735,
-      "grad_norm": 0.8896383047103882,
+      "grad_norm": 0.9093789458274841,
       "learning_rate": 9.23758215709514e-05,
-      "loss": 3.2054,
+      "loss": 3.1907,
       "step": 78650
     },
     {
       "epoch": 8.470562910343343,
-      "grad_norm": 0.8669857382774353,
+      "grad_norm": 0.8739199042320251,
       "learning_rate": 9.205258054088998e-05,
-      "loss": 3.1779,
+      "loss": 3.1824,
       "step": 78700
     },
     {
       "epoch": 8.475944462382952,
-      "grad_norm": 0.8286911249160767,
+      "grad_norm": 0.8636928796768188,
       "learning_rate": 9.172933951082856e-05,
-      "loss": 3.1926,
+      "loss": 3.1903,
       "step": 78750
     },
     {
       "epoch": 8.48132601442256,
-      "grad_norm": 0.9176493883132935,
+      "grad_norm": 0.8505300283432007,
       "learning_rate": 9.140609848076715e-05,
-      "loss": 3.174,
+      "loss": 3.1837,
       "step": 78800
     },
     {
       "epoch": 8.486707566462167,
-      "grad_norm": 0.8651835322380066,
+      "grad_norm": 0.9065614938735962,
       "learning_rate": 9.108285745070574e-05,
-      "loss": 3.1803,
+      "loss": 3.164,
       "step": 78850
     },
     {
       "epoch": 8.492089118501776,
-      "grad_norm": 0.8289731740951538,
+      "grad_norm": 0.8492125868797302,
       "learning_rate": 9.075961642064432e-05,
-      "loss": 3.1996,
+      "loss": 3.1778,
       "step": 78900
     },
     {
       "epoch": 8.497470670541384,
-      "grad_norm": 0.9024319052696228,
+      "grad_norm": 0.868232011795044,
       "learning_rate": 9.04363753905829e-05,
-      "loss": 3.1928,
+      "loss": 3.1872,
       "step": 78950
     },
     {
       "epoch": 8.502852222580993,
-      "grad_norm": 0.8506482243537903,
+      "grad_norm": 0.8927959203720093,
       "learning_rate": 9.011313436052148e-05,
-      "loss": 3.1977,
+      "loss": 3.185,
       "step": 79000
     },
     {
       "epoch": 8.502852222580993,
-      "eval_accuracy": 0.3912251493189474,
-      "eval_loss": 3.3283579349517822,
-      "eval_runtime": 180.3098,
-      "eval_samples_per_second": 99.889,
-      "eval_steps_per_second": 6.245,
+      "eval_accuracy": 0.39166584543192295,
+      "eval_loss": 3.32340145111084,
+      "eval_runtime": 181.4161,
+      "eval_samples_per_second": 99.28,
+      "eval_steps_per_second": 6.207,
       "step": 79000
     },
     {
       "epoch": 8.508233774620601,
-      "grad_norm": 0.8655647039413452,
+      "grad_norm": 0.8536961078643799,
       "learning_rate": 8.978989333046008e-05,
-      "loss": 3.1843,
+      "loss": 3.1806,
       "step": 79050
     },
     {
       "epoch": 8.513615326660208,
-      "grad_norm": 0.8432492613792419,
+      "grad_norm": 0.8980720043182373,
       "learning_rate": 8.946665230039865e-05,
-      "loss": 3.1584,
+      "loss": 3.1779,
       "step": 79100
     },
     {
       "epoch": 8.518996878699816,
-      "grad_norm": 0.817113995552063,
+      "grad_norm": 0.8239961862564087,
       "learning_rate": 8.914341127033724e-05,
-      "loss": 3.2022,
+      "loss": 3.1751,
       "step": 79150
     },
     {
       "epoch": 8.524378430739425,
-      "grad_norm": 0.8752052783966064,
+      "grad_norm": 0.8525621891021729,
       "learning_rate": 8.882017024027582e-05,
-      "loss": 3.1791,
+      "loss": 3.1839,
       "step": 79200
     },
     {
       "epoch": 8.529759982779034,
-      "grad_norm": 0.8376759886741638,
+      "grad_norm": 0.8827507495880127,
       "learning_rate": 8.84969292102144e-05,
-      "loss": 3.1689,
+      "loss": 3.1723,
       "step": 79250
     },
     {
       "epoch": 8.535141534818642,
-      "grad_norm": 0.8467958569526672,
+      "grad_norm": 0.8329736590385437,
       "learning_rate": 8.817368818015298e-05,
-      "loss": 3.1819,
+      "loss": 3.1767,
       "step": 79300
     },
     {
       "epoch": 8.54052308685825,
-      "grad_norm": 0.842562735080719,
+      "grad_norm": 0.8325408101081848,
       "learning_rate": 8.785044715009158e-05,
-      "loss": 3.1974,
+      "loss": 3.196,
       "step": 79350
     },
     {
       "epoch": 8.545904638897857,
-      "grad_norm": 0.9210129976272583,
+      "grad_norm": 0.8350241780281067,
       "learning_rate": 8.752720612003017e-05,
-      "loss": 3.191,
+      "loss": 3.1698,
       "step": 79400
     },
     {
       "epoch": 8.551286190937466,
-      "grad_norm": 0.8453426361083984,
+      "grad_norm": 0.8397579789161682,
       "learning_rate": 8.720396508996874e-05,
-      "loss": 3.1839,
+      "loss": 3.1807,
       "step": 79450
     },
     {
       "epoch": 8.556667742977075,
-      "grad_norm": 0.8634918928146362,
+      "grad_norm": 0.8875340819358826,
       "learning_rate": 8.688072405990734e-05,
-      "loss": 3.1739,
+      "loss": 3.1636,
       "step": 79500
     },
     {
       "epoch": 8.562049295016683,
-      "grad_norm": 0.8293809294700623,
+      "grad_norm": 0.8244848847389221,
       "learning_rate": 8.655748302984592e-05,
-      "loss": 3.169,
+      "loss": 3.1692,
       "step": 79550
     },
     {
       "epoch": 8.567430847056292,
-      "grad_norm": 0.8583179116249084,
+      "grad_norm": 0.8657687306404114,
       "learning_rate": 8.62342419997845e-05,
-      "loss": 3.1982,
+      "loss": 3.1805,
       "step": 79600
     },
     {
       "epoch": 8.572812399095898,
-      "grad_norm": 0.8895404934883118,
+      "grad_norm": 0.8618715405464172,
       "learning_rate": 8.591100096972308e-05,
-      "loss": 3.1812,
+      "loss": 3.1668,
       "step": 79650
     },
     {
       "epoch": 8.578193951135507,
-      "grad_norm": 0.820783257484436,
+      "grad_norm": 0.8069872260093689,
       "learning_rate": 8.558775993966167e-05,
-      "loss": 3.1779,
+      "loss": 3.1839,
       "step": 79700
     },
     {
       "epoch": 8.583575503175116,
-      "grad_norm": 0.9021645188331604,
+      "grad_norm": 0.9038444757461548,
       "learning_rate": 8.526451890960024e-05,
-      "loss": 3.1975,
+      "loss": 3.1845,
       "step": 79750
     },
     {
       "epoch": 8.588957055214724,
-      "grad_norm": 0.8516561985015869,
+      "grad_norm": 0.8421767950057983,
       "learning_rate": 8.494127787953884e-05,
-      "loss": 3.1713,
+      "loss": 3.1754,
       "step": 79800
     },
     {
       "epoch": 8.594338607254333,
-      "grad_norm": 0.8891592621803284,
+      "grad_norm": 0.8555297255516052,
       "learning_rate": 8.461803684947742e-05,
-      "loss": 3.2062,
+      "loss": 3.188,
       "step": 79850
     },
     {
       "epoch": 8.599720159293941,
-      "grad_norm": 0.891211211681366,
+      "grad_norm": 0.8646808862686157,
       "learning_rate": 8.429479581941601e-05,
-      "loss": 3.1899,
+      "loss": 3.1851,
       "step": 79900
     },
     {
       "epoch": 8.605101711333548,
-      "grad_norm": 0.8360291719436646,
+      "grad_norm": 0.8445534110069275,
       "learning_rate": 8.397155478935458e-05,
-      "loss": 3.1905,
+      "loss": 3.193,
       "step": 79950
     },
     {
       "epoch": 8.610483263373157,
-      "grad_norm": 0.8511280417442322,
+      "grad_norm": 0.827599287033081,
       "learning_rate": 8.364831375929317e-05,
-      "loss": 3.1876,
+      "loss": 3.1825,
       "step": 80000
     },
     {
       "epoch": 8.610483263373157,
-      "eval_accuracy": 0.39159695950104856,
-      "eval_loss": 3.3249058723449707,
-      "eval_runtime": 180.4737,
-      "eval_samples_per_second": 99.798,
-      "eval_steps_per_second": 6.239,
+      "eval_accuracy": 0.3919550794192851,
+      "eval_loss": 3.3201775550842285,
+      "eval_runtime": 181.6456,
+      "eval_samples_per_second": 99.155,
+      "eval_steps_per_second": 6.199,
       "step": 80000
     },
     {
       "epoch": 8.615864815412765,
-      "grad_norm": 0.9268209934234619,
+      "grad_norm": 0.916170060634613,
       "learning_rate": 8.332507272923177e-05,
-      "loss": 3.2012,
+      "loss": 3.1778,
       "step": 80050
     },
     {
       "epoch": 8.621246367452374,
-      "grad_norm": 0.865637481212616,
+      "grad_norm": 0.9013429284095764,
       "learning_rate": 8.300183169917034e-05,
-      "loss": 3.1838,
+      "loss": 3.1902,
       "step": 80100
     },
     {
       "epoch": 8.626627919491982,
-      "grad_norm": 0.8277395367622375,
+      "grad_norm": 0.9239352345466614,
       "learning_rate": 8.267859066910892e-05,
-      "loss": 3.1959,
+      "loss": 3.1885,
       "step": 80150
     },
     {
       "epoch": 8.632009471531589,
-      "grad_norm": 0.8716737627983093,
+      "grad_norm": 0.8812062740325928,
       "learning_rate": 8.235534963904751e-05,
-      "loss": 3.1944,
+      "loss": 3.1702,
       "step": 80200
     },
     {
       "epoch": 8.637391023571197,
-      "grad_norm": 0.8669558763504028,
-      "learning_rate": 8.203210860898608e-05,
-      "loss": 3.2017,
+      "grad_norm": 0.9011082053184509,
+      "learning_rate": 8.203857342958732e-05,
+      "loss": 3.1868,
       "step": 80250
     },
     {
       "epoch": 8.642772575610806,
-      "grad_norm": 0.842424213886261,
-      "learning_rate": 8.170886757892467e-05,
-      "loss": 3.158,
+      "grad_norm": 0.8420903086662292,
+      "learning_rate": 8.171533239952591e-05,
+      "loss": 3.1706,
       "step": 80300
     },
     {
       "epoch": 8.648154127650415,
-      "grad_norm": 0.8678565621376038,
-      "learning_rate": 8.138562654886327e-05,
-      "loss": 3.1759,
+      "grad_norm": 0.874015212059021,
+      "learning_rate": 8.13920913694645e-05,
+      "loss": 3.1717,
       "step": 80350
     },
     {
       "epoch": 8.653535679690023,
-      "grad_norm": 0.8407251834869385,
-      "learning_rate": 8.106238551880185e-05,
-      "loss": 3.1856,
+      "grad_norm": 0.8691864013671875,
+      "learning_rate": 8.106885033940308e-05,
+      "loss": 3.1816,
       "step": 80400
     },
     {
       "epoch": 8.658917231729632,
-      "grad_norm": 0.8746572732925415,
-      "learning_rate": 8.073914448874043e-05,
-      "loss": 3.1802,
+      "grad_norm": 0.916719913482666,
+      "learning_rate": 8.074560930934166e-05,
+      "loss": 3.1747,
       "step": 80450
     },
     {
       "epoch": 8.664298783769238,
-      "grad_norm": 0.827447235584259,
+      "grad_norm": 0.8649615049362183,
       "learning_rate": 8.042236827928025e-05,
-      "loss": 3.1879,
+      "loss": 3.1747,
       "step": 80500
     },
     {
       "epoch": 8.669680335808847,
-      "grad_norm": 0.8395963907241821,
+      "grad_norm": 0.8532657027244568,
       "learning_rate": 8.009912724921882e-05,
-      "loss": 3.1832,
+      "loss": 3.1879,
       "step": 80550
     },
     {
       "epoch": 8.675061887848456,
-      "grad_norm": 0.8884485363960266,
+      "grad_norm": 0.8832079768180847,
       "learning_rate": 7.977588621915741e-05,
-      "loss": 3.1769,
+      "loss": 3.1847,
       "step": 80600
     },
     {
       "epoch": 8.680443439888064,
-      "grad_norm": 0.8452661633491516,
+      "grad_norm": 0.8887848854064941,
       "learning_rate": 7.9452645189096e-05,
-      "loss": 3.21,
+      "loss": 3.2009,
       "step": 80650
     },
     {
       "epoch": 8.685824991927673,
-      "grad_norm": 0.9012867212295532,
+      "grad_norm": 0.9165250062942505,
       "learning_rate": 7.912940415903459e-05,
-      "loss": 3.1787,
+      "loss": 3.1614,
       "step": 80700
     },
     {
       "epoch": 8.69120654396728,
-      "grad_norm": 0.8785222172737122,
+      "grad_norm": 0.8537759780883789,
       "learning_rate": 7.880616312897316e-05,
-      "loss": 3.1909,
+      "loss": 3.1853,
       "step": 80750
     },
     {
       "epoch": 8.696588096006888,
-      "grad_norm": 0.9251953959465027,
+      "grad_norm": 0.9272660613059998,
       "learning_rate": 7.848292209891175e-05,
-      "loss": 3.1933,
+      "loss": 3.1878,
       "step": 80800
     },
     {
       "epoch": 8.701969648046497,
-      "grad_norm": 0.8459815382957458,
+      "grad_norm": 0.8552669882774353,
       "learning_rate": 7.815968106885033e-05,
-      "loss": 3.1805,
+      "loss": 3.1851,
       "step": 80850
     },
     {
       "epoch": 8.707351200086105,
-      "grad_norm": 0.8944352865219116,
+      "grad_norm": 0.9528095722198486,
       "learning_rate": 7.783644003878891e-05,
-      "loss": 3.1875,
+      "loss": 3.184,
       "step": 80900
     },
     {
       "epoch": 8.712732752125714,
-      "grad_norm": 0.8620814681053162,
+      "grad_norm": 0.9369159936904907,
       "learning_rate": 7.75131990087275e-05,
-      "loss": 3.1919,
+      "loss": 3.1934,
       "step": 80950
     },
     {
       "epoch": 8.718114304165322,
-      "grad_norm": 0.8819995522499084,
+      "grad_norm": 0.8817811012268066,
       "learning_rate": 7.718995797866609e-05,
-      "loss": 3.1825,
+      "loss": 3.1896,
       "step": 81000
     },
     {
       "epoch": 8.718114304165322,
-      "eval_accuracy": 0.3917101758101512,
-      "eval_loss": 3.322167158126831,
-      "eval_runtime": 180.0608,
-      "eval_samples_per_second": 100.027,
-      "eval_steps_per_second": 6.253,
+      "eval_accuracy": 0.39230820130468214,
+      "eval_loss": 3.3151938915252686,
+      "eval_runtime": 181.5727,
+      "eval_samples_per_second": 99.194,
+      "eval_steps_per_second": 6.201,
       "step": 81000
     },
     {
       "epoch": 8.723495856204929,
-      "grad_norm": 0.8823017477989197,
+      "grad_norm": 0.8849422335624695,
       "learning_rate": 7.686671694860466e-05,
-      "loss": 3.1884,
+      "loss": 3.1841,
       "step": 81050
     },
     {
       "epoch": 8.728877408244538,
-      "grad_norm": 0.8862807750701904,
+      "grad_norm": 0.8850445747375488,
       "learning_rate": 7.654347591854325e-05,
-      "loss": 3.1889,
+      "loss": 3.1767,
       "step": 81100
     },
     {
       "epoch": 8.734258960284146,
-      "grad_norm": 0.9089580774307251,
+      "grad_norm": 0.8567271828651428,
       "learning_rate": 7.622023488848185e-05,
-      "loss": 3.1832,
+      "loss": 3.1919,
       "step": 81150
     },
     {
       "epoch": 8.739640512323755,
-      "grad_norm": 0.8399238586425781,
+      "grad_norm": 0.8971793055534363,
       "learning_rate": 7.589699385842043e-05,
-      "loss": 3.1812,
+      "loss": 3.175,
       "step": 81200
     },
     {
       "epoch": 8.745022064363363,
-      "grad_norm": 0.868499755859375,
+      "grad_norm": 0.8772340416908264,
       "learning_rate": 7.557375282835901e-05,
-      "loss": 3.1851,
+      "loss": 3.1798,
       "step": 81250
     },
     {
       "epoch": 8.75040361640297,
-      "grad_norm": 0.8629152774810791,
+      "grad_norm": 0.8921447992324829,
       "learning_rate": 7.525051179829759e-05,
-      "loss": 3.18,
+      "loss": 3.1749,
       "step": 81300
     },
     {
       "epoch": 8.755785168442578,
-      "grad_norm": 0.8513661623001099,
+      "grad_norm": 0.8602461218833923,
       "learning_rate": 7.492727076823617e-05,
-      "loss": 3.1762,
+      "loss": 3.1827,
       "step": 81350
     },
     {
       "epoch": 8.761166720482187,
-      "grad_norm": 0.9037156105041504,
+      "grad_norm": 0.8485262393951416,
       "learning_rate": 7.460402973817477e-05,
-      "loss": 3.2112,
+      "loss": 3.1808,
       "step": 81400
     },
     {
       "epoch": 8.766548272521796,
-      "grad_norm": 0.8453684449195862,
+      "grad_norm": 0.8867838978767395,
       "learning_rate": 7.428078870811335e-05,
-      "loss": 3.182,
+      "loss": 3.1814,
       "step": 81450
     },
     {
       "epoch": 8.771929824561404,
-      "grad_norm": 0.831931471824646,
+      "grad_norm": 0.8461852669715881,
       "learning_rate": 7.395754767805193e-05,
-      "loss": 3.1843,
+      "loss": 3.1889,
       "step": 81500
     },
     {
       "epoch": 8.777311376601011,
-      "grad_norm": 0.8770779967308044,
+      "grad_norm": 0.8726191520690918,
       "learning_rate": 7.363430664799051e-05,
-      "loss": 3.1861,
+      "loss": 3.1805,
       "step": 81550
     },
     {
       "epoch": 8.78269292864062,
-      "grad_norm": 0.8554553985595703,
-      "learning_rate": 7.331753043853033e-05,
-      "loss": 3.1837,
+      "grad_norm": 0.8757393956184387,
+      "learning_rate": 7.331106561792909e-05,
+      "loss": 3.1835,
       "step": 81600
     },
     {
       "epoch": 8.788074480680228,
-      "grad_norm": 0.8347432613372803,
-      "learning_rate": 7.299428940846891e-05,
-      "loss": 3.1902,
+      "grad_norm": 0.8656665682792664,
+      "learning_rate": 7.298782458786768e-05,
+      "loss": 3.2026,
       "step": 81650
     },
     {
       "epoch": 8.793456032719837,
-      "grad_norm": 0.8451250195503235,
-      "learning_rate": 7.267104837840749e-05,
-      "loss": 3.1677,
+      "grad_norm": 0.8687547445297241,
+      "learning_rate": 7.266458355780627e-05,
+      "loss": 3.1732,
       "step": 81700
     },
     {
       "epoch": 8.798837584759445,
-      "grad_norm": 0.8608439564704895,
+      "grad_norm": 0.8750776648521423,
       "learning_rate": 7.234780734834607e-05,
-      "loss": 3.1794,
+      "loss": 3.1843,
       "step": 81750
     },
     {
       "epoch": 8.804219136799054,
-      "grad_norm": 0.8560290336608887,
+      "grad_norm": 0.8352978825569153,
       "learning_rate": 7.202456631828465e-05,
-      "loss": 3.1846,
+      "loss": 3.1906,
       "step": 81800
     },
     {
       "epoch": 8.80960068883866,
-      "grad_norm": 0.8419299721717834,
+      "grad_norm": 0.8741137981414795,
       "learning_rate": 7.170132528822325e-05,
-      "loss": 3.1821,
+      "loss": 3.1769,
       "step": 81850
     },
     {
       "epoch": 8.814982240878269,
-      "grad_norm": 0.8612204194068909,
+      "grad_norm": 0.8747343420982361,
       "learning_rate": 7.137808425816183e-05,
-      "loss": 3.1991,
+      "loss": 3.1883,
       "step": 81900
     },
     {
       "epoch": 8.820363792917878,
-      "grad_norm": 0.8462232947349548,
+      "grad_norm": 0.8592297434806824,
       "learning_rate": 7.105484322810041e-05,
-      "loss": 3.1852,
+      "loss": 3.1838,
       "step": 81950
     },
     {
       "epoch": 8.825745344957486,
-      "grad_norm": 0.8963456153869629,
+      "grad_norm": 0.8523854613304138,
       "learning_rate": 7.073160219803899e-05,
-      "loss": 3.2005,
+      "loss": 3.1849,
       "step": 82000
     },
     {
       "epoch": 8.825745344957486,
-      "eval_accuracy": 0.3922394240266956,
-      "eval_loss": 3.3186590671539307,
-      "eval_runtime": 180.0835,
-      "eval_samples_per_second": 100.015,
-      "eval_steps_per_second": 6.253,
+      "eval_accuracy": 0.3926516530830637,
+      "eval_loss": 3.3133463859558105,
+      "eval_runtime": 181.5153,
+      "eval_samples_per_second": 99.226,
+      "eval_steps_per_second": 6.203,
       "step": 82000
     },
     {
       "epoch": 8.831126896997095,
-      "grad_norm": 0.8609824776649475,
+      "grad_norm": 0.8775599002838135,
       "learning_rate": 7.040836116797757e-05,
-      "loss": 3.1799,
+      "loss": 3.1817,
       "step": 82050
     },
     {
       "epoch": 8.836508449036701,
-      "grad_norm": 0.839068591594696,
+      "grad_norm": 0.8804963827133179,
       "learning_rate": 7.008512013791617e-05,
-      "loss": 3.1796,
+      "loss": 3.1742,
       "step": 82100
     },
     {
       "epoch": 8.84189000107631,
-      "grad_norm": 0.850020170211792,
+      "grad_norm": 0.9261247515678406,
       "learning_rate": 6.976187910785475e-05,
-      "loss": 3.19,
+      "loss": 3.1863,
       "step": 82150
     },
     {
       "epoch": 8.847271553115919,
-      "grad_norm": 0.8698260188102722,
+      "grad_norm": 0.8318873643875122,
       "learning_rate": 6.943863807779334e-05,
-      "loss": 3.1946,
+      "loss": 3.1732,
       "step": 82200
     },
     {
       "epoch": 8.852653105155527,
-      "grad_norm": 0.8751007914543152,
+      "grad_norm": 0.9031708836555481,
       "learning_rate": 6.911539704773192e-05,
-      "loss": 3.1885,
+      "loss": 3.185,
       "step": 82250
     },
     {
       "epoch": 8.858034657195136,
-      "grad_norm": 0.8869954943656921,
+      "grad_norm": 0.9135169386863708,
       "learning_rate": 6.87921560176705e-05,
-      "loss": 3.1783,
+      "loss": 3.1703,
       "step": 82300
     },
     {
       "epoch": 8.863416209234742,
-      "grad_norm": 0.8722847104072571,
+      "grad_norm": 0.9164227247238159,
       "learning_rate": 6.846891498760909e-05,
-      "loss": 3.1916,
+      "loss": 3.1781,
       "step": 82350
     },
     {
       "epoch": 8.868797761274351,
-      "grad_norm": 0.8729155659675598,
+      "grad_norm": 0.8673744797706604,
       "learning_rate": 6.814567395754767e-05,
-      "loss": 3.1889,
+      "loss": 3.1872,
       "step": 82400
     },
     {
       "epoch": 8.87417931331396,
-      "grad_norm": 0.8469229936599731,
+      "grad_norm": 0.8416364192962646,
       "learning_rate": 6.782243292748626e-05,
-      "loss": 3.188,
+      "loss": 3.183,
       "step": 82450
     },
     {
       "epoch": 8.879560865353568,
-      "grad_norm": 0.8862374424934387,
+      "grad_norm": 0.8786758184432983,
       "learning_rate": 6.749919189742484e-05,
-      "loss": 3.1697,
+      "loss": 3.1816,
       "step": 82500
     },
     {
       "epoch": 8.884942417393177,
-      "grad_norm": 0.8705187439918518,
+      "grad_norm": 0.947847843170166,
       "learning_rate": 6.717595086736342e-05,
-      "loss": 3.1786,
+      "loss": 3.1807,
       "step": 82550
     },
     {
       "epoch": 8.890323969432785,
-      "grad_norm": 0.8562043309211731,
+      "grad_norm": 0.8385459780693054,
       "learning_rate": 6.6852709837302e-05,
-      "loss": 3.1752,
+      "loss": 3.1775,
       "step": 82600
     },
     {
       "epoch": 8.895705521472392,
-      "grad_norm": 0.9573428630828857,
+      "grad_norm": 0.8831902146339417,
       "learning_rate": 6.652946880724059e-05,
-      "loss": 3.1849,
+      "loss": 3.1715,
       "step": 82650
     },
     {
       "epoch": 8.901087073512,
-      "grad_norm": 0.8421146869659424,
+      "grad_norm": 0.9066095352172852,
       "learning_rate": 6.620622777717918e-05,
-      "loss": 3.1855,
+      "loss": 3.1895,
       "step": 82700
     },
     {
       "epoch": 8.906468625551609,
-      "grad_norm": 0.8492905497550964,
+      "grad_norm": 0.8783186078071594,
       "learning_rate": 6.588298674711776e-05,
-      "loss": 3.1856,
+      "loss": 3.1663,
       "step": 82750
     },
     {
       "epoch": 8.911850177591218,
-      "grad_norm": 0.8871941566467285,
+      "grad_norm": 0.8438040614128113,
       "learning_rate": 6.555974571705636e-05,
-      "loss": 3.2015,
+      "loss": 3.1909,
       "step": 82800
     },
     {
       "epoch": 8.917231729630826,
-      "grad_norm": 0.8534443378448486,
+      "grad_norm": 0.9270029664039612,
       "learning_rate": 6.523650468699494e-05,
-      "loss": 3.168,
+      "loss": 3.1663,
       "step": 82850
     },
     {
       "epoch": 8.922613281670433,
-      "grad_norm": 0.8974738717079163,
+      "grad_norm": 0.8941572308540344,
       "learning_rate": 6.491326365693352e-05,
-      "loss": 3.1728,
+      "loss": 3.1874,
       "step": 82900
     },
     {
       "epoch": 8.927994833710041,
-      "grad_norm": 0.8600006103515625,
+      "grad_norm": 0.8892514705657959,
       "learning_rate": 6.45900226268721e-05,
-      "loss": 3.1972,
+      "loss": 3.1893,
       "step": 82950
     },
     {
       "epoch": 8.93337638574965,
-      "grad_norm": 0.8655703663825989,
+      "grad_norm": 0.8542389273643494,
       "learning_rate": 6.426678159681068e-05,
-      "loss": 3.1977,
+      "loss": 3.1758,
       "step": 83000
     },
     {
       "epoch": 8.93337638574965,
-      "eval_accuracy": 0.39260102083734216,
-      "eval_loss": 3.314582586288452,
-      "eval_runtime": 180.2494,
-      "eval_samples_per_second": 99.923,
-      "eval_steps_per_second": 6.247,
+      "eval_accuracy": 0.39286808963559017,
+      "eval_loss": 3.3102710247039795,
+      "eval_runtime": 181.421,
+      "eval_samples_per_second": 99.277,
+      "eval_steps_per_second": 6.207,
       "step": 83000
     },
     {
       "epoch": 8.938757937789259,
-      "grad_norm": 0.8925529718399048,
+      "grad_norm": 0.9175463914871216,
       "learning_rate": 6.394354056674928e-05,
-      "loss": 3.1958,
+      "loss": 3.1839,
       "step": 83050
     },
     {
       "epoch": 8.944139489828867,
-      "grad_norm": 0.8607410788536072,
+      "grad_norm": 0.9068874716758728,
       "learning_rate": 6.362029953668786e-05,
-      "loss": 3.1825,
+      "loss": 3.1819,
       "step": 83100
     },
     {
       "epoch": 8.949521041868476,
-      "grad_norm": 0.8853652477264404,
+      "grad_norm": 0.8630077242851257,
       "learning_rate": 6.329705850662644e-05,
-      "loss": 3.1842,
+      "loss": 3.171,
       "step": 83150
     },
     {
       "epoch": 8.954902593908082,
-      "grad_norm": 0.8844691514968872,
+      "grad_norm": 0.8880189061164856,
       "learning_rate": 6.297381747656502e-05,
-      "loss": 3.1719,
+      "loss": 3.1787,
       "step": 83200
     },
     {
       "epoch": 8.960284145947691,
-      "grad_norm": 0.8798843622207642,
+      "grad_norm": 0.954484760761261,
       "learning_rate": 6.26505764465036e-05,
-      "loss": 3.1837,
+      "loss": 3.1609,
       "step": 83250
     },
     {
       "epoch": 8.9656656979873,
-      "grad_norm": 0.8720901608467102,
+      "grad_norm": 0.8992748260498047,
       "learning_rate": 6.23273354164422e-05,
-      "loss": 3.1988,
+      "loss": 3.1808,
       "step": 83300
     },
     {
       "epoch": 8.971047250026908,
-      "grad_norm": 0.8754738569259644,
+      "grad_norm": 0.9024131894111633,
       "learning_rate": 6.200409438638078e-05,
-      "loss": 3.1926,
+      "loss": 3.2001,
       "step": 83350
     },
     {
       "epoch": 8.976428802066517,
-      "grad_norm": 0.8542868494987488,
+      "grad_norm": 0.9032748937606812,
       "learning_rate": 6.168085335631936e-05,
-      "loss": 3.1732,
+      "loss": 3.1767,
       "step": 83400
     },
     {
       "epoch": 8.981810354106123,
-      "grad_norm": 0.8611506223678589,
+      "grad_norm": 0.8852294087409973,
       "learning_rate": 6.135761232625794e-05,
-      "loss": 3.1916,
+      "loss": 3.1722,
       "step": 83450
     },
     {
       "epoch": 8.987191906145732,
-      "grad_norm": 0.8679481744766235,
+      "grad_norm": 0.9219005107879639,
       "learning_rate": 6.1034371296196526e-05,
-      "loss": 3.1845,
+      "loss": 3.1974,
       "step": 83500
     },
     {
       "epoch": 8.99257345818534,
-      "grad_norm": 0.8732113838195801,
+      "grad_norm": 0.8996587991714478,
       "learning_rate": 6.0711130266135114e-05,
-      "loss": 3.1882,
+      "loss": 3.1796,
       "step": 83550
     },
     {
       "epoch": 8.997955010224949,
-      "grad_norm": 0.8252281546592712,
+      "grad_norm": 0.8689001202583313,
       "learning_rate": 6.0387889236073695e-05,
-      "loss": 3.2006,
+      "loss": 3.195,
       "step": 83600
     },
     {
       "epoch": 9.003336562264558,
-      "grad_norm": 0.8319926857948303,
+      "grad_norm": 0.8726151585578918,
       "learning_rate": 6.0064648206012276e-05,
-      "loss": 3.145,
+      "loss": 3.1485,
       "step": 83650
     },
     {
       "epoch": 9.008718114304166,
-      "grad_norm": 0.8742890357971191,
+      "grad_norm": 0.8759093880653381,
       "learning_rate": 5.9741407175950864e-05,
-      "loss": 3.1252,
+      "loss": 3.1253,
       "step": 83700
     },
     {
       "epoch": 9.014099666343773,
-      "grad_norm": 0.8541421294212341,
+      "grad_norm": 0.865600049495697,
       "learning_rate": 5.9418166145889445e-05,
-      "loss": 3.1408,
+      "loss": 3.1147,
       "step": 83750
     },
     {
       "epoch": 9.019481218383381,
-      "grad_norm": 0.9140190482139587,
-      "learning_rate": 5.910138993642926e-05,
-      "loss": 3.1268,
+      "grad_norm": 0.9205170273780823,
+      "learning_rate": 5.909492511582803e-05,
+      "loss": 3.1211,
       "step": 83800
     },
     {
       "epoch": 9.02486277042299,
-      "grad_norm": 0.8530510067939758,
-      "learning_rate": 5.877814890636784e-05,
-      "loss": 3.1121,
+      "grad_norm": 0.8818604946136475,
+      "learning_rate": 5.8771684085766614e-05,
+      "loss": 3.1101,
       "step": 83850
     },
     {
       "epoch": 9.030244322462599,
-      "grad_norm": 0.8723753094673157,
-      "learning_rate": 5.845490787630643e-05,
-      "loss": 3.1191,
+      "grad_norm": 0.860866129398346,
+      "learning_rate": 5.8448443055705195e-05,
+      "loss": 3.0984,
       "step": 83900
     },
     {
       "epoch": 9.035625874502207,
-      "grad_norm": 0.8698916435241699,
-      "learning_rate": 5.813166684624501e-05,
-      "loss": 3.1148,
+      "grad_norm": 0.8571581840515137,
+      "learning_rate": 5.812520202564378e-05,
+      "loss": 3.1228,
       "step": 83950
     },
     {
       "epoch": 9.041007426541814,
-      "grad_norm": 0.8469932675361633,
-      "learning_rate": 5.7808425816183596e-05,
-      "loss": 3.1246,
+      "grad_norm": 0.8829055428504944,
+      "learning_rate": 5.7801960995582364e-05,
+      "loss": 3.12,
       "step": 84000
     },
     {
       "epoch": 9.041007426541814,
-      "eval_accuracy": 0.392553104913816,
-      "eval_loss": 3.317844867706299,
-      "eval_runtime": 180.533,
-      "eval_samples_per_second": 99.766,
-      "eval_steps_per_second": 6.237,
+      "eval_accuracy": 0.3930165094803447,
+      "eval_loss": 3.311964273452759,
+      "eval_runtime": 181.5845,
+      "eval_samples_per_second": 99.188,
+      "eval_steps_per_second": 6.201,
       "step": 84000
     },
     {
       "epoch": 9.046388978581422,
-      "grad_norm": 0.8725888133049011,
-      "learning_rate": 5.748518478612218e-05,
-      "loss": 3.1223,
+      "grad_norm": 0.8859900832176208,
+      "learning_rate": 5.747871996552096e-05,
+      "loss": 3.126,
       "step": 84050
     },
     {
       "epoch": 9.051770530621031,
-      "grad_norm": 0.8994928598403931,
-      "learning_rate": 5.716194375606076e-05,
-      "loss": 3.1452,
+      "grad_norm": 0.8596351742744446,
+      "learning_rate": 5.715547893545954e-05,
+      "loss": 3.1255,
       "step": 84100
     },
     {
       "epoch": 9.05715208266064,
-      "grad_norm": 0.8656386137008667,
-      "learning_rate": 5.6838702725999346e-05,
-      "loss": 3.1392,
+      "grad_norm": 0.8363323211669922,
+      "learning_rate": 5.683223790539812e-05,
+      "loss": 3.1318,
       "step": 84150
     },
     {
       "epoch": 9.062533634700248,
-      "grad_norm": 0.878341794013977,
-      "learning_rate": 5.651546169593793e-05,
-      "loss": 3.1356,
+      "grad_norm": 0.8673672676086426,
+      "learning_rate": 5.650899687533671e-05,
+      "loss": 3.1238,
       "step": 84200
     },
     {
       "epoch": 9.067915186739857,
-      "grad_norm": 0.8708506226539612,
-      "learning_rate": 5.619222066587652e-05,
-      "loss": 3.1184,
+      "grad_norm": 0.9039381742477417,
+      "learning_rate": 5.618575584527529e-05,
+      "loss": 3.1144,
       "step": 84250
     },
     {
       "epoch": 9.073296738779463,
-      "grad_norm": 0.8545727133750916,
-      "learning_rate": 5.58689796358151e-05,
-      "loss": 3.1142,
+      "grad_norm": 0.9219964742660522,
+      "learning_rate": 5.586251481521388e-05,
+      "loss": 3.1259,
       "step": 84300
     },
     {
       "epoch": 9.078678290819072,
-      "grad_norm": 0.8976442217826843,
-      "learning_rate": 5.554573860575369e-05,
-      "loss": 3.107,
+      "grad_norm": 0.9105638861656189,
+      "learning_rate": 5.553927378515246e-05,
+      "loss": 3.1123,
       "step": 84350
     },
     {
       "epoch": 9.08405984285868,
-      "grad_norm": 0.8379390239715576,
-      "learning_rate": 5.522249757569227e-05,
-      "loss": 3.1199,
+      "grad_norm": 0.9005851745605469,
+      "learning_rate": 5.521603275509104e-05,
+      "loss": 3.1207,
       "step": 84400
     },
     {
       "epoch": 9.089441394898289,
-      "grad_norm": 0.8370739221572876,
-      "learning_rate": 5.489925654563085e-05,
-      "loss": 3.1285,
+      "grad_norm": 0.9091371893882751,
+      "learning_rate": 5.489279172502963e-05,
+      "loss": 3.1402,
       "step": 84450
     },
     {
       "epoch": 9.094822946937898,
-      "grad_norm": 0.8469734191894531,
-      "learning_rate": 5.457601551556944e-05,
-      "loss": 3.1455,
+      "grad_norm": 0.878474771976471,
+      "learning_rate": 5.456955069496821e-05,
+      "loss": 3.1267,
       "step": 84500
     },
     {
       "epoch": 9.100204498977504,
-      "grad_norm": 0.8684940934181213,
-      "learning_rate": 5.425277448550802e-05,
-      "loss": 3.1088,
+      "grad_norm": 0.8800078630447388,
+      "learning_rate": 5.42463096649068e-05,
+      "loss": 3.1288,
       "step": 84550
     },
     {
       "epoch": 9.105586051017113,
-      "grad_norm": 0.8705862760543823,
-      "learning_rate": 5.392953345544661e-05,
-      "loss": 3.1245,
+      "grad_norm": 0.9209677577018738,
+      "learning_rate": 5.392306863484538e-05,
+      "loss": 3.1145,
       "step": 84600
     },
     {
       "epoch": 9.110967603056721,
-      "grad_norm": 0.8357287645339966,
-      "learning_rate": 5.360629242538519e-05,
-      "loss": 3.1275,
+      "grad_norm": 0.9008042216300964,
+      "learning_rate": 5.359982760478396e-05,
+      "loss": 3.1352,
       "step": 84650
     },
     {
       "epoch": 9.11634915509633,
-      "grad_norm": 0.8421687483787537,
-      "learning_rate": 5.328305139532377e-05,
-      "loss": 3.1243,
+      "grad_norm": 0.9068533182144165,
+      "learning_rate": 5.327658657472255e-05,
+      "loss": 3.1311,
       "step": 84700
     },
     {
       "epoch": 9.121730707135939,
-      "grad_norm": 0.8550165891647339,
-      "learning_rate": 5.295981036526236e-05,
-      "loss": 3.1221,
+      "grad_norm": 0.8785526752471924,
+      "learning_rate": 5.295334554466113e-05,
+      "loss": 3.1277,
       "step": 84750
     },
     {
       "epoch": 9.127112259175545,
-      "grad_norm": 0.8496278524398804,
-      "learning_rate": 5.263656933520094e-05,
-      "loss": 3.1369,
+      "grad_norm": 0.9123873710632324,
+      "learning_rate": 5.2630104514599716e-05,
+      "loss": 3.1301,
       "step": 84800
     },
     {
       "epoch": 9.132493811215154,
-      "grad_norm": 0.8498865365982056,
-      "learning_rate": 5.231332830513953e-05,
-      "loss": 3.1421,
+      "grad_norm": 0.8630001544952393,
+      "learning_rate": 5.23068634845383e-05,
+      "loss": 3.1341,
       "step": 84850
     },
     {
       "epoch": 9.137875363254762,
-      "grad_norm": 0.905013382434845,
-      "learning_rate": 5.199008727507811e-05,
-      "loss": 3.1319,
+      "grad_norm": 0.9000571966171265,
+      "learning_rate": 5.198362245447688e-05,
+      "loss": 3.1485,
       "step": 84900
     },
     {
       "epoch": 9.143256915294371,
-      "grad_norm": 0.9200946092605591,
-      "learning_rate": 5.166684624501669e-05,
-      "loss": 3.112,
+      "grad_norm": 0.874817430973053,
+      "learning_rate": 5.166038142441547e-05,
+      "loss": 3.1079,
       "step": 84950
     },
     {
       "epoch": 9.14863846733398,
-      "grad_norm": 0.847834050655365,
-      "learning_rate": 5.1343605214955286e-05,
-      "loss": 3.1312,
+      "grad_norm": 0.9093799591064453,
+      "learning_rate": 5.1337140394354054e-05,
+      "loss": 3.1131,
       "step": 85000
     },
     {
       "epoch": 9.14863846733398,
-      "eval_accuracy": 0.39271803999751836,
-      "eval_loss": 3.315976142883301,
-      "eval_runtime": 179.964,
-      "eval_samples_per_second": 100.081,
-      "eval_steps_per_second": 6.257,
+      "eval_accuracy": 0.3931902454479601,
+      "eval_loss": 3.3117618560791016,
+      "eval_runtime": 181.6796,
+      "eval_samples_per_second": 99.136,
+      "eval_steps_per_second": 6.198,
       "step": 85000
     },
     {
       "epoch": 9.154020019373588,
-      "grad_norm": 0.8495249152183533,
-      "learning_rate": 5.102036418489387e-05,
-      "loss": 3.1226,
+      "grad_norm": 0.8656437993049622,
+      "learning_rate": 5.101389936429264e-05,
+      "loss": 3.1192,
       "step": 85050
     },
     {
       "epoch": 9.159401571413195,
-      "grad_norm": 0.9137347936630249,
-      "learning_rate": 5.0697123154832455e-05,
-      "loss": 3.1358,
+      "grad_norm": 0.88176429271698,
+      "learning_rate": 5.069065833423122e-05,
+      "loss": 3.12,
       "step": 85100
     },
     {
       "epoch": 9.164783123452803,
-      "grad_norm": 0.8848801851272583,
-      "learning_rate": 5.0373882124771036e-05,
-      "loss": 3.1319,
+      "grad_norm": 0.8807859420776367,
+      "learning_rate": 5.0367417304169804e-05,
+      "loss": 3.1302,
       "step": 85150
     },
     {
       "epoch": 9.170164675492412,
-      "grad_norm": 0.8463972806930542,
-      "learning_rate": 5.005064109470962e-05,
-      "loss": 3.1284,
+      "grad_norm": 0.8822762966156006,
+      "learning_rate": 5.004417627410839e-05,
+      "loss": 3.1281,
       "step": 85200
     },
     {
       "epoch": 9.17554622753202,
-      "grad_norm": 0.8625044822692871,
-      "learning_rate": 4.9727400064648205e-05,
-      "loss": 3.1187,
+      "grad_norm": 0.8844353556632996,
+      "learning_rate": 4.972093524404697e-05,
+      "loss": 3.1442,
       "step": 85250
     },
     {
       "epoch": 9.180927779571629,
-      "grad_norm": 0.8781099319458008,
-      "learning_rate": 4.9404159034586786e-05,
-      "loss": 3.1446,
+      "grad_norm": 0.8862656950950623,
+      "learning_rate": 4.939769421398556e-05,
+      "loss": 3.129,
       "step": 85300
     },
     {
       "epoch": 9.186309331611236,
-      "grad_norm": 0.877238392829895,
-      "learning_rate": 4.9080918004525374e-05,
-      "loss": 3.1407,
+      "grad_norm": 0.8508253693580627,
+      "learning_rate": 4.907445318392414e-05,
+      "loss": 3.1354,
       "step": 85350
     },
     {
       "epoch": 9.191690883650844,
-      "grad_norm": 0.9211892485618591,
-      "learning_rate": 4.8757676974463955e-05,
-      "loss": 3.1393,
+      "grad_norm": 0.8784798979759216,
+      "learning_rate": 4.875121215386272e-05,
+      "loss": 3.1346,
       "step": 85400
     },
     {
       "epoch": 9.197072435690453,
-      "grad_norm": 0.8986982107162476,
-      "learning_rate": 4.8434435944402536e-05,
-      "loss": 3.1425,
+      "grad_norm": 0.9343785047531128,
+      "learning_rate": 4.842797112380131e-05,
+      "loss": 3.1453,
       "step": 85450
     },
     {
       "epoch": 9.202453987730062,
-      "grad_norm": 0.8680441975593567,
-      "learning_rate": 4.8111194914341124e-05,
-      "loss": 3.1271,
+      "grad_norm": 0.9291565418243408,
+      "learning_rate": 4.810473009373989e-05,
+      "loss": 3.1097,
       "step": 85500
     },
     {
       "epoch": 9.20783553976967,
-      "grad_norm": 0.8860616683959961,
-      "learning_rate": 4.7787953884279705e-05,
-      "loss": 3.1492,
+      "grad_norm": 0.9097803831100464,
+      "learning_rate": 4.778148906367848e-05,
+      "loss": 3.1415,
       "step": 85550
     },
     {
       "epoch": 9.213217091809279,
-      "grad_norm": 0.895710289478302,
-      "learning_rate": 4.746471285421829e-05,
-      "loss": 3.1334,
+      "grad_norm": 0.847196102142334,
+      "learning_rate": 4.745824803361706e-05,
+      "loss": 3.1408,
       "step": 85600
     },
     {
       "epoch": 9.218598643848885,
-      "grad_norm": 0.8937450051307678,
-      "learning_rate": 4.7141471824156874e-05,
-      "loss": 3.1275,
+      "grad_norm": 0.930571436882019,
+      "learning_rate": 4.713500700355564e-05,
+      "loss": 3.1175,
       "step": 85650
     },
     {
       "epoch": 9.223980195888494,
-      "grad_norm": 0.8729709386825562,
-      "learning_rate": 4.6818230794095455e-05,
-      "loss": 3.1199,
+      "grad_norm": 0.8943590521812439,
+      "learning_rate": 4.681176597349424e-05,
+      "loss": 3.1091,
       "step": 85700
     },
     {
       "epoch": 9.229361747928102,
-      "grad_norm": 0.878231942653656,
+      "grad_norm": 0.8899019956588745,
       "learning_rate": 4.649498976403404e-05,
-      "loss": 3.1359,
+      "loss": 3.139,
       "step": 85750
     },
     {
       "epoch": 9.234743299967711,
-      "grad_norm": 0.8926569223403931,
+      "grad_norm": 0.8903750777244568,
       "learning_rate": 4.6171748733972624e-05,
-      "loss": 3.1282,
+      "loss": 3.1269,
       "step": 85800
     },
     {
       "epoch": 9.24012485200732,
-      "grad_norm": 0.945385754108429,
+      "grad_norm": 0.8654310703277588,
       "learning_rate": 4.584850770391122e-05,
-      "loss": 3.1277,
+      "loss": 3.1259,
       "step": 85850
     },
     {
       "epoch": 9.245506404046926,
-      "grad_norm": 0.8847855925559998,
+      "grad_norm": 0.9114632606506348,
       "learning_rate": 4.55252666738498e-05,
-      "loss": 3.1344,
+      "loss": 3.1209,
       "step": 85900
     },
     {
       "epoch": 9.250887956086535,
-      "grad_norm": 0.8700735569000244,
+      "grad_norm": 0.8892890810966492,
       "learning_rate": 4.520202564378838e-05,
-      "loss": 3.1481,
+      "loss": 3.1358,
       "step": 85950
     },
     {
       "epoch": 9.256269508126143,
-      "grad_norm": 0.8868721127510071,
+      "grad_norm": 0.9733080267906189,
       "learning_rate": 4.487878461372697e-05,
-      "loss": 3.1217,
+      "loss": 3.1451,
       "step": 86000
     },
     {
       "epoch": 9.256269508126143,
-      "eval_accuracy": 0.39288764715539676,
-      "eval_loss": 3.315528392791748,
-      "eval_runtime": 180.0796,
-      "eval_samples_per_second": 100.017,
-      "eval_steps_per_second": 6.253,
+      "eval_accuracy": 0.3934551411884518,
+      "eval_loss": 3.3093411922454834,
+      "eval_runtime": 181.5303,
+      "eval_samples_per_second": 99.218,
+      "eval_steps_per_second": 6.203,
       "step": 86000
     },
     {
       "epoch": 9.261651060165752,
-      "grad_norm": 0.8628836274147034,
+      "grad_norm": 0.8770269751548767,
       "learning_rate": 4.455554358366555e-05,
-      "loss": 3.14,
+      "loss": 3.127,
       "step": 86050
     },
     {
       "epoch": 9.26703261220536,
-      "grad_norm": 0.8455495834350586,
+      "grad_norm": 0.8930492401123047,
       "learning_rate": 4.423230255360414e-05,
-      "loss": 3.1434,
+      "loss": 3.1329,
       "step": 86100
     },
     {
       "epoch": 9.272414164244967,
-      "grad_norm": 0.8857659101486206,
+      "grad_norm": 0.8644741773605347,
       "learning_rate": 4.390906152354272e-05,
-      "loss": 3.143,
+      "loss": 3.1224,
       "step": 86150
     },
     {
       "epoch": 9.277795716284576,
-      "grad_norm": 0.8970041275024414,
+      "grad_norm": 0.8792781829833984,
       "learning_rate": 4.35858204934813e-05,
-      "loss": 3.1309,
+      "loss": 3.1197,
       "step": 86200
     },
     {
       "epoch": 9.283177268324184,
-      "grad_norm": 0.8499646186828613,
+      "grad_norm": 0.9281783103942871,
       "learning_rate": 4.326257946341989e-05,
-      "loss": 3.1318,
+      "loss": 3.1271,
       "step": 86250
     },
     {
       "epoch": 9.288558820363793,
-      "grad_norm": 0.9052905440330505,
+      "grad_norm": 0.9894546866416931,
       "learning_rate": 4.293933843335847e-05,
-      "loss": 3.1497,
+      "loss": 3.1451,
       "step": 86300
     },
     {
       "epoch": 9.293940372403402,
-      "grad_norm": 0.8457171320915222,
+      "grad_norm": 0.8942108750343323,
       "learning_rate": 4.261609740329706e-05,
-      "loss": 3.1347,
+      "loss": 3.1245,
       "step": 86350
     },
     {
       "epoch": 9.29932192444301,
-      "grad_norm": 0.9543759226799011,
+      "grad_norm": 0.9210761785507202,
       "learning_rate": 4.229285637323564e-05,
-      "loss": 3.1198,
+      "loss": 3.1245,
       "step": 86400
     },
     {
       "epoch": 9.304703476482617,
-      "grad_norm": 0.8748459219932556,
+      "grad_norm": 0.8660423159599304,
       "learning_rate": 4.196961534317422e-05,
-      "loss": 3.1473,
+      "loss": 3.1293,
       "step": 86450
     },
     {
       "epoch": 9.310085028522225,
-      "grad_norm": 0.8755471110343933,
+      "grad_norm": 0.8767070174217224,
       "learning_rate": 4.164637431311281e-05,
-      "loss": 3.1387,
+      "loss": 3.1572,
       "step": 86500
     },
     {
       "epoch": 9.315466580561834,
-      "grad_norm": 0.8886383175849915,
-      "learning_rate": 4.132313328305139e-05,
-      "loss": 3.1433,
+      "grad_norm": 0.8374525308609009,
+      "learning_rate": 4.132959810365262e-05,
+      "loss": 3.1365,
       "step": 86550
     },
     {
       "epoch": 9.320848132601443,
-      "grad_norm": 0.9134498238563538,
-      "learning_rate": 4.099989225298998e-05,
-      "loss": 3.1315,
+      "grad_norm": 0.9271016120910645,
+      "learning_rate": 4.10063570735912e-05,
+      "loss": 3.1268,
       "step": 86600
     },
     {
       "epoch": 9.326229684641051,
-      "grad_norm": 0.8910537958145142,
-      "learning_rate": 4.0676651222928564e-05,
-      "loss": 3.1212,
+      "grad_norm": 0.9330965280532837,
+      "learning_rate": 4.068311604352979e-05,
+      "loss": 3.1314,
       "step": 86650
     },
     {
       "epoch": 9.331611236680658,
-      "grad_norm": 0.8579962849617004,
-      "learning_rate": 4.035341019286714e-05,
-      "loss": 3.1215,
+      "grad_norm": 0.9366960525512695,
+      "learning_rate": 4.035987501346837e-05,
+      "loss": 3.117,
       "step": 86700
     },
     {
       "epoch": 9.336992788720266,
-      "grad_norm": 0.9000516533851624,
-      "learning_rate": 4.003016916280573e-05,
-      "loss": 3.144,
+      "grad_norm": 0.8736404180526733,
+      "learning_rate": 4.003663398340695e-05,
+      "loss": 3.1241,
       "step": 86750
     },
     {
       "epoch": 9.342374340759875,
-      "grad_norm": 0.8634475469589233,
-      "learning_rate": 3.9706928132744314e-05,
-      "loss": 3.1411,
+      "grad_norm": 0.8498652577400208,
+      "learning_rate": 3.9713392953345546e-05,
+      "loss": 3.1163,
       "step": 86800
     },
     {
       "epoch": 9.347755892799483,
-      "grad_norm": 0.8831536769866943,
-      "learning_rate": 3.93836871026829e-05,
-      "loss": 3.1445,
+      "grad_norm": 0.9030478000640869,
+      "learning_rate": 3.939015192328413e-05,
+      "loss": 3.1359,
       "step": 86850
     },
     {
       "epoch": 9.353137444839092,
-      "grad_norm": 0.8995107412338257,
-      "learning_rate": 3.906044607262148e-05,
-      "loss": 3.1412,
+      "grad_norm": 0.918379545211792,
+      "learning_rate": 3.9066910893222715e-05,
+      "loss": 3.1375,
       "step": 86900
     },
     {
       "epoch": 9.3585189968787,
-      "grad_norm": 0.8746486306190491,
-      "learning_rate": 3.8737205042560064e-05,
-      "loss": 3.1454,
+      "grad_norm": 0.9341984987258911,
+      "learning_rate": 3.8743669863161296e-05,
+      "loss": 3.1352,
       "step": 86950
     },
     {
       "epoch": 9.363900548918307,
-      "grad_norm": 0.8589686155319214,
-      "learning_rate": 3.841396401249865e-05,
-      "loss": 3.1471,
+      "grad_norm": 0.919605553150177,
+      "learning_rate": 3.842042883309988e-05,
+      "loss": 3.1218,
       "step": 87000
     },
     {
       "epoch": 9.363900548918307,
-      "eval_accuracy": 0.39334594503619824,
-      "eval_loss": 3.312450885772705,
-      "eval_runtime": 180.3192,
-      "eval_samples_per_second": 99.884,
-      "eval_steps_per_second": 6.244,
+      "eval_accuracy": 0.39360084471101103,
+      "eval_loss": 3.3079707622528076,
+      "eval_runtime": 181.5742,
+      "eval_samples_per_second": 99.194,
+      "eval_steps_per_second": 6.201,
       "step": 87000
     },
     {
       "epoch": 9.369282100957916,
-      "grad_norm": 0.8290409445762634,
-      "learning_rate": 3.809072298243723e-05,
-      "loss": 3.1332,
+      "grad_norm": 0.9246619343757629,
+      "learning_rate": 3.8097187803038465e-05,
+      "loss": 3.1167,
       "step": 87050
     },
     {
       "epoch": 9.374663652997524,
-      "grad_norm": 0.8649762272834778,
-      "learning_rate": 3.776748195237582e-05,
-      "loss": 3.1382,
+      "grad_norm": 0.9096710681915283,
+      "learning_rate": 3.7773946772977047e-05,
+      "loss": 3.1429,
       "step": 87100
     },
     {
       "epoch": 9.380045205037133,
-      "grad_norm": 0.9703359603881836,
-      "learning_rate": 3.74442409223144e-05,
-      "loss": 3.1272,
+      "grad_norm": 0.9599007964134216,
+      "learning_rate": 3.745070574291563e-05,
+      "loss": 3.1321,
       "step": 87150
     },
     {
       "epoch": 9.385426757076742,
-      "grad_norm": 0.8370105028152466,
-      "learning_rate": 3.712099989225299e-05,
-      "loss": 3.1192,
+      "grad_norm": 0.8843571543693542,
+      "learning_rate": 3.7127464712854216e-05,
+      "loss": 3.1402,
       "step": 87200
     },
     {
       "epoch": 9.390808309116348,
-      "grad_norm": 0.8700204491615295,
-      "learning_rate": 3.679775886219157e-05,
-      "loss": 3.144,
+      "grad_norm": 0.9159232378005981,
+      "learning_rate": 3.6804223682792803e-05,
+      "loss": 3.1447,
       "step": 87250
     },
     {
       "epoch": 9.396189861155957,
-      "grad_norm": 0.9093730449676514,
-      "learning_rate": 3.647451783213015e-05,
-      "loss": 3.1191,
+      "grad_norm": 0.9162116646766663,
+      "learning_rate": 3.6480982652731385e-05,
+      "loss": 3.1247,
       "step": 87300
     },
     {
       "epoch": 9.401571413195565,
-      "grad_norm": 0.8856383562088013,
-      "learning_rate": 3.615127680206874e-05,
-      "loss": 3.1242,
+      "grad_norm": 0.9041388034820557,
+      "learning_rate": 3.6157741622669966e-05,
+      "loss": 3.1341,
       "step": 87350
     },
     {
       "epoch": 9.406952965235174,
-      "grad_norm": 0.8729093670845032,
-      "learning_rate": 3.582803577200732e-05,
-      "loss": 3.1429,
+      "grad_norm": 0.952892541885376,
+      "learning_rate": 3.5834500592608554e-05,
+      "loss": 3.1213,
       "step": 87400
     },
     {
       "epoch": 9.412334517274783,
-      "grad_norm": 0.8521737456321716,
-      "learning_rate": 3.550479474194591e-05,
+      "grad_norm": 0.8928467631340027,
+      "learning_rate": 3.5511259562547135e-05,
       "loss": 3.1177,
       "step": 87450
     },
     {
       "epoch": 9.417716069314391,
-      "grad_norm": 0.8820858597755432,
-      "learning_rate": 3.518155371188449e-05,
-      "loss": 3.1373,
+      "grad_norm": 0.8981311917304993,
+      "learning_rate": 3.518801853248572e-05,
+      "loss": 3.1369,
       "step": 87500
     },
     {
       "epoch": 9.423097621353998,
-      "grad_norm": 0.8602292537689209,
-      "learning_rate": 3.485831268182308e-05,
-      "loss": 3.1505,
+      "grad_norm": 0.9102454781532288,
+      "learning_rate": 3.4864777502424304e-05,
+      "loss": 3.1391,
       "step": 87550
     },
     {
       "epoch": 9.428479173393606,
-      "grad_norm": 0.8709748387336731,
-      "learning_rate": 3.453507165176166e-05,
-      "loss": 3.1302,
+      "grad_norm": 0.8761741518974304,
+      "learning_rate": 3.4541536472362885e-05,
+      "loss": 3.1264,
       "step": 87600
     },
     {
       "epoch": 9.433860725433215,
-      "grad_norm": 0.8324317932128906,
-      "learning_rate": 3.421183062170025e-05,
-      "loss": 3.1146,
+      "grad_norm": 0.8536211252212524,
+      "learning_rate": 3.421829544230147e-05,
+      "loss": 3.1204,
       "step": 87650
     },
     {
       "epoch": 9.439242277472824,
-      "grad_norm": 0.8815144300460815,
-      "learning_rate": 3.388858959163883e-05,
-      "loss": 3.1346,
+      "grad_norm": 0.8878208994865417,
+      "learning_rate": 3.389505441224006e-05,
+      "loss": 3.151,
       "step": 87700
     },
     {
       "epoch": 9.444623829512432,
-      "grad_norm": 0.866257905960083,
-      "learning_rate": 3.3565348561577416e-05,
-      "loss": 3.1255,
+      "grad_norm": 0.8712186217308044,
+      "learning_rate": 3.357181338217864e-05,
+      "loss": 3.1276,
       "step": 87750
     },
     {
       "epoch": 9.450005381552039,
-      "grad_norm": 0.8640872240066528,
-      "learning_rate": 3.3242107531516e-05,
-      "loss": 3.1326,
+      "grad_norm": 0.9246970415115356,
+      "learning_rate": 3.324857235211722e-05,
+      "loss": 3.1342,
       "step": 87800
     },
     {
       "epoch": 9.455386933591647,
-      "grad_norm": 0.876515805721283,
-      "learning_rate": 3.291886650145458e-05,
-      "loss": 3.1376,
+      "grad_norm": 0.9588025212287903,
+      "learning_rate": 3.292533132205581e-05,
+      "loss": 3.143,
       "step": 87850
     },
     {
       "epoch": 9.460768485631256,
-      "grad_norm": 0.879612147808075,
+      "grad_norm": 0.9029521346092224,
       "learning_rate": 3.260209029199439e-05,
-      "loss": 3.1468,
+      "loss": 3.1444,
       "step": 87900
     },
     {
       "epoch": 9.466150037670864,
-      "grad_norm": 0.8799201846122742,
+      "grad_norm": 0.9064067602157593,
       "learning_rate": 3.227884926193298e-05,
-      "loss": 3.1336,
+      "loss": 3.1129,
       "step": 87950
     },
     {
       "epoch": 9.471531589710473,
-      "grad_norm": 0.8598930239677429,
+      "grad_norm": 0.9410727620124817,
       "learning_rate": 3.195560823187157e-05,
-      "loss": 3.1102,
+      "loss": 3.1126,
       "step": 88000
     },
     {
       "epoch": 9.471531589710473,
-      "eval_accuracy": 0.3935834602489607,
-      "eval_loss": 3.309974431991577,
-      "eval_runtime": 179.9534,
-      "eval_samples_per_second": 100.087,
-      "eval_steps_per_second": 6.257,
+      "eval_accuracy": 0.3940364341382593,
+      "eval_loss": 3.30414080619812,
+      "eval_runtime": 181.7772,
+      "eval_samples_per_second": 99.083,
+      "eval_steps_per_second": 6.194,
       "step": 88000
     },
     {
       "epoch": 9.476913141750082,
-      "grad_norm": 0.8778337836265564,
+      "grad_norm": 0.912320613861084,
       "learning_rate": 3.163236720181014e-05,
-      "loss": 3.1379,
+      "loss": 3.1299,
       "step": 88050
     },
     {
       "epoch": 9.482294693789688,
-      "grad_norm": 0.8822025656700134,
+      "grad_norm": 0.8634088635444641,
       "learning_rate": 3.130912617174873e-05,
-      "loss": 3.1384,
+      "loss": 3.1255,
       "step": 88100
     },
     {
       "epoch": 9.487676245829297,
-      "grad_norm": 0.927155077457428,
+      "grad_norm": 0.9067851901054382,
       "learning_rate": 3.098588514168732e-05,
-      "loss": 3.1588,
+      "loss": 3.1355,
       "step": 88150
     },
     {
       "epoch": 9.493057797868905,
-      "grad_norm": 0.8520760536193848,
+      "grad_norm": 0.940524697303772,
       "learning_rate": 3.06626441116259e-05,
-      "loss": 3.1472,
+      "loss": 3.1301,
       "step": 88200
     },
     {
       "epoch": 9.498439349908514,
-      "grad_norm": 0.8926303386688232,
+      "grad_norm": 0.909763753414154,
       "learning_rate": 3.0339403081564487e-05,
-      "loss": 3.1344,
+      "loss": 3.1408,
       "step": 88250
     },
     {
       "epoch": 9.503820901948123,
-      "grad_norm": 0.8950697779655457,
+      "grad_norm": 0.8630996942520142,
       "learning_rate": 3.0016162051503068e-05,
-      "loss": 3.135,
+      "loss": 3.1386,
       "step": 88300
     },
     {
       "epoch": 9.50920245398773,
-      "grad_norm": 0.8820440173149109,
+      "grad_norm": 0.8941014409065247,
       "learning_rate": 2.9692921021441652e-05,
-      "loss": 3.1264,
+      "loss": 3.1274,
       "step": 88350
     },
     {
       "epoch": 9.514584006027338,
-      "grad_norm": 0.8579733371734619,
+      "grad_norm": 0.901056170463562,
       "learning_rate": 2.9369679991380237e-05,
-      "loss": 3.1423,
+      "loss": 3.1236,
       "step": 88400
     },
     {
       "epoch": 9.519965558066946,
-      "grad_norm": 0.8843384385108948,
+      "grad_norm": 0.9280726909637451,
       "learning_rate": 2.904643896131882e-05,
-      "loss": 3.1311,
+      "loss": 3.1425,
       "step": 88450
     },
     {
       "epoch": 9.525347110106555,
-      "grad_norm": 0.8826112747192383,
+      "grad_norm": 0.8801893591880798,
       "learning_rate": 2.8723197931257406e-05,
-      "loss": 3.1382,
+      "loss": 3.1403,
       "step": 88500
     },
     {
       "epoch": 9.530728662146164,
-      "grad_norm": 0.8869061470031738,
+      "grad_norm": 0.8809930682182312,
       "learning_rate": 2.8399956901195987e-05,
-      "loss": 3.1228,
+      "loss": 3.1156,
       "step": 88550
     },
     {
       "epoch": 9.536110214185772,
-      "grad_norm": 0.8704985976219177,
+      "grad_norm": 0.8723519444465637,
       "learning_rate": 2.807671587113457e-05,
-      "loss": 3.1219,
+      "loss": 3.1261,
       "step": 88600
     },
     {
       "epoch": 9.541491766225379,
-      "grad_norm": 0.9074740409851074,
+      "grad_norm": 0.8852314949035645,
       "learning_rate": 2.775347484107316e-05,
-      "loss": 3.1322,
+      "loss": 3.1235,
       "step": 88650
     },
     {
       "epoch": 9.546873318264987,
-      "grad_norm": 0.8920992016792297,
+      "grad_norm": 0.9637598991394043,
       "learning_rate": 2.7430233811011744e-05,
-      "loss": 3.1456,
+      "loss": 3.1308,
       "step": 88700
     },
     {
       "epoch": 9.552254870304596,
-      "grad_norm": 0.9204345941543579,
+      "grad_norm": 0.8822471499443054,
       "learning_rate": 2.7106992780950328e-05,
-      "loss": 3.1497,
+      "loss": 3.1389,
       "step": 88750
     },
     {
       "epoch": 9.557636422344205,
-      "grad_norm": 0.8399198055267334,
+      "grad_norm": 0.8932937979698181,
       "learning_rate": 2.678375175088891e-05,
-      "loss": 3.1334,
+      "loss": 3.1301,
       "step": 88800
     },
     {
       "epoch": 9.563017974383813,
-      "grad_norm": 0.8751394748687744,
+      "grad_norm": 0.9187436103820801,
       "learning_rate": 2.6460510720827494e-05,
-      "loss": 3.1389,
+      "loss": 3.1331,
       "step": 88850
     },
     {
       "epoch": 9.56839952642342,
-      "grad_norm": 0.8489452600479126,
+      "grad_norm": 0.8509601354598999,
       "learning_rate": 2.6137269690766078e-05,
-      "loss": 3.1277,
+      "loss": 3.1255,
       "step": 88900
     },
     {
       "epoch": 9.573781078463028,
-      "grad_norm": 0.8947514295578003,
+      "grad_norm": 0.9203357696533203,
       "learning_rate": 2.5814028660704663e-05,
-      "loss": 3.1329,
+      "loss": 3.146,
       "step": 88950
     },
     {
       "epoch": 9.579162630502637,
-      "grad_norm": 0.8745483160018921,
+      "grad_norm": 0.8963664770126343,
       "learning_rate": 2.5490787630643247e-05,
       "loss": 3.1177,
       "step": 89000
     },
     {
       "epoch": 9.579162630502637,
-      "eval_accuracy": 0.3937311195235006,
-      "eval_loss": 3.307589054107666,
-      "eval_runtime": 180.1555,
-      "eval_samples_per_second": 99.975,
-      "eval_steps_per_second": 6.25,
+      "eval_accuracy": 0.39420908357699647,
+      "eval_loss": 3.302248239517212,
+      "eval_runtime": 181.7095,
+      "eval_samples_per_second": 99.12,
+      "eval_steps_per_second": 6.197,
       "step": 89000
     },
     {
       "epoch": 9.584544182542245,
-      "grad_norm": 0.9029579162597656,
+      "grad_norm": 0.9269243478775024,
       "learning_rate": 2.5167546600581828e-05,
-      "loss": 3.1263,
+      "loss": 3.1187,
       "step": 89050
     },
     {
       "epoch": 9.589925734581854,
-      "grad_norm": 0.8661422729492188,
+      "grad_norm": 0.8919320106506348,
       "learning_rate": 2.4844305570520416e-05,
-      "loss": 3.1395,
+      "loss": 3.1354,
       "step": 89100
     },
     {
       "epoch": 9.59530728662146,
-      "grad_norm": 0.8325968980789185,
+      "grad_norm": 0.9003577828407288,
       "learning_rate": 2.4521064540459e-05,
-      "loss": 3.1621,
+      "loss": 3.1393,
       "step": 89150
     },
     {
       "epoch": 9.60068883866107,
-      "grad_norm": 0.8808440566062927,
+      "grad_norm": 0.8862392902374268,
       "learning_rate": 2.4197823510397585e-05,
-      "loss": 3.1332,
+      "loss": 3.133,
       "step": 89200
     },
     {
       "epoch": 9.606070390700678,
-      "grad_norm": 0.8744810223579407,
+      "grad_norm": 0.9000269174575806,
       "learning_rate": 2.387458248033617e-05,
-      "loss": 3.1255,
+      "loss": 3.1364,
       "step": 89250
     },
     {
       "epoch": 9.611451942740286,
-      "grad_norm": 0.8575035333633423,
+      "grad_norm": 0.9127604365348816,
       "learning_rate": 2.355134145027475e-05,
-      "loss": 3.122,
+      "loss": 3.1176,
       "step": 89300
     },
     {
       "epoch": 9.616833494779895,
-      "grad_norm": 0.9315617084503174,
+      "grad_norm": 0.8685051202774048,
       "learning_rate": 2.3228100420213335e-05,
-      "loss": 3.1325,
+      "loss": 3.1393,
       "step": 89350
     },
     {
       "epoch": 9.622215046819504,
-      "grad_norm": 0.8789585828781128,
-      "learning_rate": 2.291132421075315e-05,
-      "loss": 3.1261,
+      "grad_norm": 0.8889885544776917,
+      "learning_rate": 2.290485939015192e-05,
+      "loss": 3.1275,
       "step": 89400
     },
     {
       "epoch": 9.62759659885911,
-      "grad_norm": 0.8857566714286804,
-      "learning_rate": 2.2588083180691733e-05,
-      "loss": 3.1192,
+      "grad_norm": 0.8910917639732361,
+      "learning_rate": 2.2581618360090508e-05,
+      "loss": 3.1138,
       "step": 89450
     },
     {
       "epoch": 9.632978150898719,
-      "grad_norm": 0.911800742149353,
-      "learning_rate": 2.2264842150630318e-05,
-      "loss": 3.1527,
+      "grad_norm": 0.9268326759338379,
+      "learning_rate": 2.2258377330029092e-05,
+      "loss": 3.137,
       "step": 89500
     },
     {
       "epoch": 9.638359702938327,
-      "grad_norm": 0.8703020811080933,
-      "learning_rate": 2.1941601120568905e-05,
-      "loss": 3.1369,
+      "grad_norm": 0.9023292660713196,
+      "learning_rate": 2.1935136299967673e-05,
+      "loss": 3.1344,
       "step": 89550
     },
     {
       "epoch": 9.643741254977936,
-      "grad_norm": 0.878156840801239,
-      "learning_rate": 2.1618360090507483e-05,
-      "loss": 3.1499,
+      "grad_norm": 0.8972396850585938,
+      "learning_rate": 2.1611895269906258e-05,
+      "loss": 3.1434,
       "step": 89600
     },
     {
       "epoch": 9.649122807017545,
-      "grad_norm": 0.950110912322998,
-      "learning_rate": 2.129511906044607e-05,
-      "loss": 3.1427,
+      "grad_norm": 0.881137490272522,
+      "learning_rate": 2.1288654239844842e-05,
+      "loss": 3.1103,
       "step": 89650
     },
     {
       "epoch": 9.654504359057151,
-      "grad_norm": 0.8555633425712585,
-      "learning_rate": 2.0971878030384655e-05,
-      "loss": 3.1397,
+      "grad_norm": 0.8936906456947327,
+      "learning_rate": 2.0965413209783427e-05,
+      "loss": 3.1473,
       "step": 89700
     },
     {
       "epoch": 9.65988591109676,
-      "grad_norm": 0.8625248074531555,
-      "learning_rate": 2.064863700032324e-05,
-      "loss": 3.1312,
+      "grad_norm": 0.9325161576271057,
+      "learning_rate": 2.064217217972201e-05,
+      "loss": 3.1377,
       "step": 89750
     },
     {
       "epoch": 9.665267463136368,
-      "grad_norm": 0.8741613626480103,
-      "learning_rate": 2.0325395970261824e-05,
-      "loss": 3.1403,
+      "grad_norm": 0.9079487323760986,
+      "learning_rate": 2.0318931149660596e-05,
+      "loss": 3.1253,
       "step": 89800
     },
     {
       "epoch": 9.670649015175977,
-      "grad_norm": 0.8547689914703369,
-      "learning_rate": 2.0002154940200406e-05,
-      "loss": 3.1416,
+      "grad_norm": 0.9222606420516968,
+      "learning_rate": 1.9995690119599177e-05,
+      "loss": 3.1337,
       "step": 89850
     },
     {
       "epoch": 9.676030567215586,
-      "grad_norm": 0.8787466883659363,
-      "learning_rate": 1.967891391013899e-05,
-      "loss": 3.1432,
+      "grad_norm": 0.941891074180603,
+      "learning_rate": 1.9672449089537765e-05,
+      "loss": 3.1158,
       "step": 89900
     },
     {
       "epoch": 9.681412119255192,
-      "grad_norm": 0.9100810289382935,
-      "learning_rate": 1.9355672880077575e-05,
-      "loss": 3.1335,
+      "grad_norm": 0.9089145064353943,
+      "learning_rate": 1.934920805947635e-05,
+      "loss": 3.1338,
       "step": 89950
     },
     {
       "epoch": 9.6867936712948,
-      "grad_norm": 0.8840546607971191,
-      "learning_rate": 1.9032431850016162e-05,
-      "loss": 3.1193,
+      "grad_norm": 0.9245839715003967,
+      "learning_rate": 1.9025967029414934e-05,
+      "loss": 3.1149,
       "step": 90000
     },
     {
       "epoch": 9.6867936712948,
-      "eval_accuracy": 0.3940206794695262,
-      "eval_loss": 3.3056087493896484,
-      "eval_runtime": 180.2428,
-      "eval_samples_per_second": 99.926,
-      "eval_steps_per_second": 6.247,
+      "eval_accuracy": 0.39452700192674167,
+      "eval_loss": 3.3006131649017334,
+      "eval_runtime": 181.5771,
+      "eval_samples_per_second": 99.192,
+      "eval_steps_per_second": 6.201,
       "step": 90000
     },
     {
       "epoch": 9.69217522333441,
-      "grad_norm": 0.8934172987937927,
-      "learning_rate": 1.8709190819954744e-05,
-      "loss": 3.1394,
+      "grad_norm": 0.9351643919944763,
+      "learning_rate": 1.8702725999353515e-05,
+      "loss": 3.1267,
       "step": 90050
     },
     {
       "epoch": 9.697556775374018,
-      "grad_norm": 0.8928881287574768,
-      "learning_rate": 1.8385949789893328e-05,
-      "loss": 3.1375,
+      "grad_norm": 0.9001287817955017,
+      "learning_rate": 1.8379484969292103e-05,
+      "loss": 3.1445,
       "step": 90100
     },
     {
       "epoch": 9.702938327413626,
-      "grad_norm": 0.8657145500183105,
-      "learning_rate": 1.8062708759831913e-05,
-      "loss": 3.1228,
+      "grad_norm": 1.0265345573425293,
+      "learning_rate": 1.8056243939230684e-05,
+      "loss": 3.1412,
       "step": 90150
     },
     {
       "epoch": 9.708319879453235,
-      "grad_norm": 0.8868884444236755,
-      "learning_rate": 1.7739467729770497e-05,
-      "loss": 3.1286,
+      "grad_norm": 0.8874318599700928,
+      "learning_rate": 1.7733002909169268e-05,
+      "loss": 3.1239,
       "step": 90200
     },
     {
       "epoch": 9.713701431492842,
-      "grad_norm": 0.8712571859359741,
-      "learning_rate": 1.741622669970908e-05,
-      "loss": 3.1284,
+      "grad_norm": 0.9182792901992798,
+      "learning_rate": 1.7409761879107853e-05,
+      "loss": 3.1216,
       "step": 90250
     },
     {
       "epoch": 9.71908298353245,
-      "grad_norm": 0.9264222979545593,
-      "learning_rate": 1.7092985669647666e-05,
-      "loss": 3.1212,
+      "grad_norm": 0.9692125916481018,
+      "learning_rate": 1.7086520849046437e-05,
+      "loss": 3.1229,
       "step": 90300
     },
     {
       "epoch": 9.724464535572059,
-      "grad_norm": 0.8676959276199341,
-      "learning_rate": 1.676974463958625e-05,
-      "loss": 3.1413,
+      "grad_norm": 0.8922422528266907,
+      "learning_rate": 1.6763279818985022e-05,
+      "loss": 3.1312,
       "step": 90350
     },
     {
       "epoch": 9.729846087611667,
-      "grad_norm": 0.8672192692756653,
-      "learning_rate": 1.6446503609524835e-05,
-      "loss": 3.153,
+      "grad_norm": 0.8885959386825562,
+      "learning_rate": 1.6440038788923606e-05,
+      "loss": 3.1175,
       "step": 90400
     },
     {
       "epoch": 9.735227639651276,
-      "grad_norm": 0.8709021210670471,
-      "learning_rate": 1.612326257946342e-05,
-      "loss": 3.143,
+      "grad_norm": 0.8792229890823364,
+      "learning_rate": 1.611679775886219e-05,
+      "loss": 3.1364,
       "step": 90450
     },
     {
       "epoch": 9.740609191690883,
-      "grad_norm": 0.8936007618904114,
-      "learning_rate": 1.5800021549402004e-05,
-      "loss": 3.1406,
+      "grad_norm": 0.9167876243591309,
+      "learning_rate": 1.5793556728800772e-05,
+      "loss": 3.1385,
       "step": 90500
     },
     {
       "epoch": 9.745990743730491,
-      "grad_norm": 0.9075430035591125,
+      "grad_norm": 0.9501658082008362,
       "learning_rate": 1.5476780519340585e-05,
-      "loss": 3.1166,
+      "loss": 3.1244,
       "step": 90550
     },
     {
       "epoch": 9.7513722957701,
-      "grad_norm": 0.881115198135376,
+      "grad_norm": 0.9500635266304016,
       "learning_rate": 1.5153539489279171e-05,
-      "loss": 3.1441,
+      "loss": 3.1296,
       "step": 90600
     },
     {
       "epoch": 9.756753847809708,
-      "grad_norm": 0.9102500081062317,
+      "grad_norm": 0.9250226616859436,
       "learning_rate": 1.4830298459217756e-05,
-      "loss": 3.1275,
+      "loss": 3.1216,
       "step": 90650
     },
     {
       "epoch": 9.762135399849317,
-      "grad_norm": 0.9028261303901672,
+      "grad_norm": 0.8863194584846497,
       "learning_rate": 1.4507057429156339e-05,
-      "loss": 3.1364,
+      "loss": 3.1271,
       "step": 90700
     },
     {
       "epoch": 9.767516951888926,
-      "grad_norm": 0.8845318555831909,
+      "grad_norm": 0.8879140615463257,
       "learning_rate": 1.4183816399094925e-05,
-      "loss": 3.1406,
+      "loss": 3.1415,
       "step": 90750
     },
     {
       "epoch": 9.772898503928532,
-      "grad_norm": 0.861149787902832,
+      "grad_norm": 0.8984692096710205,
       "learning_rate": 1.3860575369033508e-05,
-      "loss": 3.1309,
+      "loss": 3.127,
       "step": 90800
     },
     {
       "epoch": 9.77828005596814,
-      "grad_norm": 0.8670579195022583,
+      "grad_norm": 0.9498762488365173,
       "learning_rate": 1.3537334338972092e-05,
-      "loss": 3.143,
+      "loss": 3.147,
       "step": 90850
     },
     {
       "epoch": 9.78366160800775,
-      "grad_norm": 0.8900049924850464,
+      "grad_norm": 0.9409878253936768,
       "learning_rate": 1.3214093308910678e-05,
-      "loss": 3.1461,
+      "loss": 3.148,
       "step": 90900
     },
     {
       "epoch": 9.789043160047358,
-      "grad_norm": 0.851944625377655,
+      "grad_norm": 0.8849940896034241,
       "learning_rate": 1.2890852278849261e-05,
-      "loss": 3.1187,
+      "loss": 3.1181,
       "step": 90950
     },
     {
       "epoch": 9.794424712086967,
-      "grad_norm": 0.8535908460617065,
+      "grad_norm": 0.9464505910873413,
       "learning_rate": 1.2567611248787846e-05,
-      "loss": 3.1506,
+      "loss": 3.1401,
       "step": 91000
     },
     {
       "epoch": 9.794424712086967,
-      "eval_accuracy": 0.3942323352949888,
-      "eval_loss": 3.3036699295043945,
-      "eval_runtime": 180.5405,
-      "eval_samples_per_second": 99.762,
-      "eval_steps_per_second": 6.237,
+      "eval_accuracy": 0.3947539778093861,
+      "eval_loss": 3.298004150390625,
+      "eval_runtime": 181.4648,
+      "eval_samples_per_second": 99.253,
+      "eval_steps_per_second": 6.205,
       "step": 91000
     },
     {
       "epoch": 9.799806264126573,
-      "grad_norm": 0.8900703191757202,
+      "grad_norm": 0.9349958300590515,
       "learning_rate": 1.2244370218726428e-05,
-      "loss": 3.1451,
+      "loss": 3.1333,
       "step": 91050
     },
     {
       "epoch": 9.805187816166182,
-      "grad_norm": 0.8927872776985168,
+      "grad_norm": 0.9377318024635315,
       "learning_rate": 1.1921129188665013e-05,
-      "loss": 3.1323,
+      "loss": 3.118,
       "step": 91100
     },
     {
       "epoch": 9.81056936820579,
-      "grad_norm": 0.8900262713432312,
+      "grad_norm": 0.8938496112823486,
       "learning_rate": 1.1597888158603599e-05,
-      "loss": 3.1302,
+      "loss": 3.1345,
       "step": 91150
     },
     {
       "epoch": 9.815950920245399,
-      "grad_norm": 0.8994975090026855,
+      "grad_norm": 0.9212080240249634,
       "learning_rate": 1.1274647128542182e-05,
-      "loss": 3.1444,
+      "loss": 3.1301,
       "step": 91200
     },
     {
       "epoch": 9.821332472285007,
-      "grad_norm": 0.8664494752883911,
+      "grad_norm": 0.9028072953224182,
       "learning_rate": 1.0951406098480766e-05,
-      "loss": 3.1298,
+      "loss": 3.1017,
       "step": 91250
     },
     {
       "epoch": 9.826714024324616,
-      "grad_norm": 0.8540681004524231,
+      "grad_norm": 0.8838663697242737,
       "learning_rate": 1.0628165068419349e-05,
-      "loss": 3.1188,
+      "loss": 3.134,
       "step": 91300
     },
     {
       "epoch": 9.832095576364223,
-      "grad_norm": 0.8558421730995178,
+      "grad_norm": 0.9222512245178223,
       "learning_rate": 1.0304924038357935e-05,
-      "loss": 3.133,
+      "loss": 3.1453,
       "step": 91350
     },
     {
       "epoch": 9.837477128403831,
-      "grad_norm": 0.8718653917312622,
+      "grad_norm": 0.8636956810951233,
       "learning_rate": 9.98168300829652e-06,
-      "loss": 3.1366,
+      "loss": 3.1298,
       "step": 91400
     },
     {
       "epoch": 9.84285868044344,
-      "grad_norm": 0.8789852857589722,
+      "grad_norm": 0.9381572604179382,
       "learning_rate": 9.658441978235103e-06,
-      "loss": 3.1252,
+      "loss": 3.1341,
       "step": 91450
     },
     {
       "epoch": 9.848240232483048,
-      "grad_norm": 0.9523253440856934,
+      "grad_norm": 0.903771698474884,
       "learning_rate": 9.335200948173687e-06,
-      "loss": 3.1188,
+      "loss": 3.1135,
       "step": 91500
     },
     {
       "epoch": 9.853621784522657,
-      "grad_norm": 0.9213250279426575,
+      "grad_norm": 0.915290117263794,
       "learning_rate": 9.011959918112272e-06,
-      "loss": 3.1519,
+      "loss": 3.1265,
       "step": 91550
     },
     {
       "epoch": 9.859003336562264,
-      "grad_norm": 0.9170287251472473,
+      "grad_norm": 0.9249221682548523,
       "learning_rate": 8.688718888050856e-06,
-      "loss": 3.1392,
+      "loss": 3.1438,
       "step": 91600
     },
     {
       "epoch": 9.864384888601872,
-      "grad_norm": 0.9603930115699768,
+      "grad_norm": 0.9177095293998718,
       "learning_rate": 8.36547785798944e-06,
-      "loss": 3.1508,
+      "loss": 3.14,
       "step": 91650
     },
     {
       "epoch": 9.869766440641481,
-      "grad_norm": 0.8868028521537781,
+      "grad_norm": 0.8847270011901855,
       "learning_rate": 8.042236827928023e-06,
-      "loss": 3.1331,
+      "loss": 3.1215,
       "step": 91700
     },
     {
       "epoch": 9.87514799268109,
-      "grad_norm": 0.8630801439285278,
+      "grad_norm": 0.995018482208252,
       "learning_rate": 7.71899579786661e-06,
-      "loss": 3.1354,
+      "loss": 3.1238,
       "step": 91750
     },
     {
       "epoch": 9.880529544720698,
-      "grad_norm": 0.9191954135894775,
+      "grad_norm": 0.9142616391181946,
       "learning_rate": 7.395754767805193e-06,
-      "loss": 3.1435,
+      "loss": 3.1391,
       "step": 91800
     },
     {
       "epoch": 9.885911096760307,
-      "grad_norm": 0.8929259181022644,
+      "grad_norm": 0.9044401049613953,
       "learning_rate": 7.072513737743777e-06,
-      "loss": 3.1199,
+      "loss": 3.1295,
       "step": 91850
     },
     {
       "epoch": 9.891292648799913,
-      "grad_norm": 0.8772810101509094,
+      "grad_norm": 0.9191132187843323,
       "learning_rate": 6.749272707682361e-06,
-      "loss": 3.1623,
+      "loss": 3.1596,
       "step": 91900
     },
     {
       "epoch": 9.896674200839522,
-      "grad_norm": 0.8684717416763306,
-      "learning_rate": 6.432496498222174e-06,
-      "loss": 3.1247,
+      "grad_norm": 0.9123125076293945,
+      "learning_rate": 6.426031677620945e-06,
+      "loss": 3.1349,
       "step": 91950
     },
     {
       "epoch": 9.90205575287913,
-      "grad_norm": 0.8589189052581787,
-      "learning_rate": 6.1092554681607575e-06,
-      "loss": 3.1476,
+      "grad_norm": 0.9066173434257507,
+      "learning_rate": 6.10279064755953e-06,
+      "loss": 3.1362,
       "step": 92000
     },
     {
       "epoch": 9.90205575287913,
-      "eval_accuracy": 0.3944637659460336,
-      "eval_loss": 3.3020360469818115,
-      "eval_runtime": 180.13,
-      "eval_samples_per_second": 99.989,
-      "eval_steps_per_second": 6.251,
+      "eval_accuracy": 0.3948944659933302,
+      "eval_loss": 3.2965593338012695,
+      "eval_runtime": 181.9413,
+      "eval_samples_per_second": 98.993,
+      "eval_steps_per_second": 6.189,
       "step": 92000
     },
     {
       "epoch": 9.907437304918739,
-      "grad_norm": 0.8615749478340149,
-      "learning_rate": 5.792479258700571e-06,
-      "loss": 3.1224,
+      "grad_norm": 0.9226531982421875,
+      "learning_rate": 5.779549617498114e-06,
+      "loss": 3.1172,
       "step": 92050
     },
     {
       "epoch": 9.912818856958348,
-      "grad_norm": 0.8723222017288208,
-      "learning_rate": 5.469238228639154e-06,
-      "loss": 3.1229,
+      "grad_norm": 0.9109538197517395,
+      "learning_rate": 5.4563085874366985e-06,
+      "loss": 3.1083,
       "step": 92100
     },
     {
       "epoch": 9.918200408997954,
-      "grad_norm": 0.8633239269256592,
-      "learning_rate": 5.145997198577739e-06,
-      "loss": 3.1426,
+      "grad_norm": 0.9372076988220215,
+      "learning_rate": 5.133067557375282e-06,
+      "loss": 3.1291,
       "step": 92150
     },
     {
       "epoch": 9.923581961037563,
-      "grad_norm": 0.8642317652702332,
-      "learning_rate": 4.822756168516323e-06,
-      "loss": 3.1477,
+      "grad_norm": 0.9201867580413818,
+      "learning_rate": 4.809826527313866e-06,
+      "loss": 3.1307,
       "step": 92200
     },
     {
       "epoch": 9.928963513077171,
-      "grad_norm": 0.8761366605758667,
-      "learning_rate": 4.499515138454908e-06,
-      "loss": 3.1307,
+      "grad_norm": 0.9314913749694824,
+      "learning_rate": 4.48658549725245e-06,
+      "loss": 3.1247,
       "step": 92250
     },
     {
       "epoch": 9.93434506511678,
-      "grad_norm": 0.880110502243042,
-      "learning_rate": 4.1762741083934915e-06,
-      "loss": 3.1246,
+      "grad_norm": 0.8634207248687744,
+      "learning_rate": 4.163344467191035e-06,
+      "loss": 3.129,
       "step": 92300
     },
     {
       "epoch": 9.939726617156388,
-      "grad_norm": 0.89411860704422,
-      "learning_rate": 3.853033078332076e-06,
+      "grad_norm": 0.9112765192985535,
+      "learning_rate": 3.840103437129619e-06,
       "loss": 3.1239,
       "step": 92350
     },
     {
       "epoch": 9.945108169195997,
-      "grad_norm": 0.8907015323638916,
-      "learning_rate": 3.52979204827066e-06,
-      "loss": 3.1234,
+      "grad_norm": 0.9108915328979492,
+      "learning_rate": 3.5168624070682038e-06,
+      "loss": 3.1153,
       "step": 92400
     },
     {
       "epoch": 9.950489721235604,
-      "grad_norm": 0.8394994735717773,
-      "learning_rate": 3.206551018209244e-06,
-      "loss": 3.1339,
+      "grad_norm": 0.8764602541923523,
+      "learning_rate": 3.193621377006788e-06,
+      "loss": 3.119,
       "step": 92450
     },
     {
       "epoch": 9.955871273275212,
-      "grad_norm": 0.8820265531539917,
-      "learning_rate": 2.8833099881478286e-06,
-      "loss": 3.1329,
+      "grad_norm": 0.8718408942222595,
+      "learning_rate": 2.870380346945372e-06,
+      "loss": 3.1305,
       "step": 92500
     },
     {
       "epoch": 9.961252825314821,
-      "grad_norm": 0.8925215005874634,
-      "learning_rate": 2.5600689580864127e-06,
-      "loss": 3.1347,
+      "grad_norm": 0.9285328984260559,
+      "learning_rate": 2.5471393168839564e-06,
+      "loss": 3.1259,
       "step": 92550
     },
     {
       "epoch": 9.96663437735443,
-      "grad_norm": 0.9452439546585083,
-      "learning_rate": 2.2368279280249972e-06,
-      "loss": 3.1504,
+      "grad_norm": 0.8861116170883179,
+      "learning_rate": 2.230363107423769e-06,
+      "loss": 3.1422,
       "step": 92600
     },
     {
       "epoch": 9.972015929394038,
-      "grad_norm": 0.873985767364502,
-      "learning_rate": 1.9135868979635813e-06,
-      "loss": 3.1342,
+      "grad_norm": 0.9022042751312256,
+      "learning_rate": 1.9071220773623531e-06,
+      "loss": 3.1388,
       "step": 92650
     },
     {
       "epoch": 9.977397481433645,
-      "grad_norm": 0.8428253531455994,
-      "learning_rate": 1.5903458679021656e-06,
-      "loss": 3.1225,
+      "grad_norm": 0.8694204092025757,
+      "learning_rate": 1.5838810473009372e-06,
+      "loss": 3.1256,
       "step": 92700
     },
     {
       "epoch": 9.982779033473253,
-      "grad_norm": 0.9114113450050354,
-      "learning_rate": 1.2671048378407496e-06,
-      "loss": 3.1229,
+      "grad_norm": 0.9325965642929077,
+      "learning_rate": 1.2606400172395215e-06,
+      "loss": 3.1182,
       "step": 92750
     },
     {
       "epoch": 9.988160585512862,
-      "grad_norm": 0.8490803837776184,
-      "learning_rate": 9.438638077793341e-07,
-      "loss": 3.1276,
+      "grad_norm": 0.8977979421615601,
+      "learning_rate": 9.373989871781058e-07,
+      "loss": 3.1406,
       "step": 92800
     },
     {
       "epoch": 9.99354213755247,
-      "grad_norm": 0.8661682605743408,
-      "learning_rate": 6.206227777179182e-07,
-      "loss": 3.1459,
+      "grad_norm": 0.9170072674751282,
+      "learning_rate": 6.1415795711669e-07,
+      "loss": 3.1151,
       "step": 92850
     },
     {
       "epoch": 9.998923689592079,
-      "grad_norm": 0.8826032876968384,
-      "learning_rate": 2.9738174765650254e-07,
-      "loss": 3.1464,
+      "grad_norm": 0.8770419955253601,
+      "learning_rate": 2.909169270552742e-07,
+      "loss": 3.1432,
       "step": 92900
     },
     {
       "epoch": 10.0,
       "step": 92910,
       "total_flos": 7.7681598529536e+17,
-      "train_loss": 3.46011078748004,
-      "train_runtime": 80911.2202,
-      "train_samples_per_second": 36.744,
-      "train_steps_per_second": 1.148
+      "train_loss": 3.4552165603473464,
+      "train_runtime": 80467.4262,
+      "train_samples_per_second": 36.946,
+      "train_steps_per_second": 1.155
     }
   ],
   "logging_steps": 50,