diff --git "a/checkpoint-90000/trainer_state.json" "b/checkpoint-90000/trainer_state.json"
--- "a/checkpoint-90000/trainer_state.json"
+++ "b/checkpoint-90000/trainer_state.json"
@@ -1,5 +1,5 @@
 {
-  "best_metric": 3.302138566970825,
+  "best_metric": 3.304030179977417,
   "best_model_checkpoint": "/scratch/cl5625/exceptions/models/100M_low_500_6910/checkpoint-90000",
   "epoch": 9.703504043126685,
   "eval_steps": 1000,
@@ -10,13412 +10,13412 @@
   "log_history": [
     {
       "epoch": 0.005390835579514825,
-      "grad_norm": 2.9886317253112793,
+      "grad_norm": 4.202730655670166,
       "learning_rate": 0.000276,
-      "loss": 9.0224,
+      "loss": 9.0233,
       "step": 50
     },
     {
       "epoch": 0.01078167115902965,
-      "grad_norm": 0.8695021867752075,
+      "grad_norm": 3.1378865242004395,
       "learning_rate": 0.0005759999999999999,
-      "loss": 6.9609,
+      "loss": 6.9276,
       "step": 100
     },
     {
       "epoch": 0.016172506738544475,
-      "grad_norm": 1.7424172163009644,
+      "grad_norm": 1.2884689569473267,
       "learning_rate": 0.000599702104695089,
-      "loss": 6.5403,
+      "loss": 6.4839,
       "step": 150
     },
     {
       "epoch": 0.0215633423180593,
-      "grad_norm": 2.535945415496826,
+      "grad_norm": 1.6449090242385864,
       "learning_rate": 0.0005993783054506205,
-      "loss": 6.2529,
+      "loss": 6.2463,
       "step": 200
     },
     {
       "epoch": 0.026954177897574125,
-      "grad_norm": 1.765920877456665,
+      "grad_norm": 1.5109689235687256,
       "learning_rate": 0.0005990545062061521,
-      "loss": 6.0905,
+      "loss": 6.094,
       "step": 250
     },
     {
       "epoch": 0.03234501347708895,
-      "grad_norm": 1.0415717363357544,
+      "grad_norm": 1.170856237411499,
       "learning_rate": 0.0005987307069616836,
-      "loss": 5.9552,
+      "loss": 5.949,
       "step": 300
     },
     {
       "epoch": 0.03773584905660377,
-      "grad_norm": 1.3854601383209229,
+      "grad_norm": 2.105668783187866,
       "learning_rate": 0.0005984069077172153,
-      "loss": 5.8847,
+      "loss": 5.8791,
       "step": 350
     },
     {
       "epoch": 0.0431266846361186,
-      "grad_norm": 0.8729716539382935,
+      "grad_norm": 1.3761645555496216,
       "learning_rate": 0.0005980831084727469,
-      "loss": 5.8182,
+      "loss": 5.8165,
       "step": 400
     },
     {
       "epoch": 0.04851752021563342,
-      "grad_norm": 1.8327264785766602,
+      "grad_norm": 1.8226579427719116,
       "learning_rate": 0.0005977593092282784,
-      "loss": 5.7573,
+      "loss": 5.756,
       "step": 450
     },
     {
       "epoch": 0.05390835579514825,
-      "grad_norm": 1.8601025342941284,
+      "grad_norm": 0.8710860013961792,
       "learning_rate": 0.00059743550998381,
-      "loss": 5.6461,
+      "loss": 5.6467,
       "step": 500
     },
     {
       "epoch": 0.05929919137466307,
-      "grad_norm": 1.083327054977417,
+      "grad_norm": 0.9444846510887146,
       "learning_rate": 0.0005971117107393416,
-      "loss": 5.6014,
+      "loss": 5.6021,
       "step": 550
     },
     {
       "epoch": 0.0646900269541779,
-      "grad_norm": 0.8286775350570679,
+      "grad_norm": 1.4856325387954712,
       "learning_rate": 0.0005967879114948732,
-      "loss": 5.5275,
+      "loss": 5.5298,
       "step": 600
     },
     {
       "epoch": 0.07008086253369272,
-      "grad_norm": 1.0809779167175293,
+      "grad_norm": 1.6019251346588135,
       "learning_rate": 0.0005964641122504047,
-      "loss": 5.4639,
+      "loss": 5.4623,
       "step": 650
     },
     {
       "epoch": 0.07547169811320754,
-      "grad_norm": 1.215111255645752,
+      "grad_norm": 1.1106631755828857,
       "learning_rate": 0.0005961403130059363,
-      "loss": 5.3863,
+      "loss": 5.3906,
       "step": 700
     },
     {
       "epoch": 0.08086253369272237,
-      "grad_norm": 0.9550946950912476,
+      "grad_norm": 1.1751388311386108,
       "learning_rate": 0.0005958165137614678,
-      "loss": 5.3178,
+      "loss": 5.3255,
       "step": 750
     },
     {
       "epoch": 0.0862533692722372,
-      "grad_norm": 1.219592809677124,
+      "grad_norm": 1.028311848640442,
       "learning_rate": 0.0005954927145169995,
-      "loss": 5.2468,
+      "loss": 5.2479,
       "step": 800
     },
     {
       "epoch": 0.09164420485175202,
-      "grad_norm": 1.0037598609924316,
+      "grad_norm": 1.1774542331695557,
       "learning_rate": 0.0005951689152725309,
-      "loss": 5.2228,
+      "loss": 5.2316,
       "step": 850
     },
     {
       "epoch": 0.09703504043126684,
-      "grad_norm": 0.9939533472061157,
+      "grad_norm": 0.9043501019477844,
       "learning_rate": 0.0005948451160280626,
-      "loss": 5.167,
+      "loss": 5.1739,
       "step": 900
     },
     {
       "epoch": 0.10242587601078167,
-      "grad_norm": 0.8934879899024963,
+      "grad_norm": 0.9826597571372986,
       "learning_rate": 0.0005945213167835941,
-      "loss": 5.1051,
+      "loss": 5.1094,
       "step": 950
     },
     {
       "epoch": 0.1078167115902965,
-      "grad_norm": 1.4683802127838135,
+      "grad_norm": 1.1902110576629639,
       "learning_rate": 0.0005941975175391257,
-      "loss": 5.112,
+      "loss": 5.1176,
       "step": 1000
     },
     {
       "epoch": 0.1078167115902965,
-      "eval_accuracy": 0.22728771643247805,
-      "eval_loss": 5.021671295166016,
-      "eval_runtime": 185.7281,
-      "eval_samples_per_second": 96.975,
-      "eval_steps_per_second": 6.063,
+      "eval_accuracy": 0.22649987434293525,
+      "eval_loss": 5.031488418579102,
+      "eval_runtime": 185.546,
+      "eval_samples_per_second": 97.07,
+      "eval_steps_per_second": 6.069,
       "step": 1000
     },
     {
       "epoch": 0.11320754716981132,
-      "grad_norm": 0.9396394491195679,
+      "grad_norm": 1.1733556985855103,
       "learning_rate": 0.0005938737182946572,
-      "loss": 5.0399,
+      "loss": 5.0471,
       "step": 1050
     },
     {
       "epoch": 0.11859838274932614,
-      "grad_norm": 1.2031350135803223,
+      "grad_norm": 1.4435728788375854,
       "learning_rate": 0.0005935499190501888,
-      "loss": 5.0063,
+      "loss": 5.0089,
       "step": 1100
     },
     {
       "epoch": 0.12398921832884097,
-      "grad_norm": 1.067723274230957,
+      "grad_norm": 1.056373119354248,
       "learning_rate": 0.0005932261198057204,
-      "loss": 5.0062,
+      "loss": 5.01,
       "step": 1150
     },
     {
       "epoch": 0.1293800539083558,
-      "grad_norm": 1.0314335823059082,
+      "grad_norm": 0.8520533442497253,
       "learning_rate": 0.000592902320561252,
-      "loss": 4.9655,
+      "loss": 4.9711,
       "step": 1200
     },
     {
       "epoch": 0.1347708894878706,
-      "grad_norm": 0.7959302067756653,
+      "grad_norm": 0.9913970828056335,
       "learning_rate": 0.0005925785213167835,
-      "loss": 4.946,
+      "loss": 4.9486,
       "step": 1250
     },
     {
       "epoch": 0.14016172506738545,
-      "grad_norm": 1.1784368753433228,
+      "grad_norm": 1.615565299987793,
       "learning_rate": 0.0005922547220723151,
-      "loss": 4.9007,
+      "loss": 4.9054,
       "step": 1300
     },
     {
       "epoch": 0.14555256064690028,
-      "grad_norm": 0.9498503804206848,
+      "grad_norm": 1.1613094806671143,
       "learning_rate": 0.0005919309228278468,
-      "loss": 4.8504,
+      "loss": 4.8517,
       "step": 1350
     },
     {
       "epoch": 0.1509433962264151,
-      "grad_norm": 1.2894479036331177,
+      "grad_norm": 1.1608201265335083,
       "learning_rate": 0.0005916071235833783,
-      "loss": 4.8604,
+      "loss": 4.8605,
       "step": 1400
     },
     {
       "epoch": 0.15633423180592992,
-      "grad_norm": 0.9321544170379639,
+      "grad_norm": 1.0951576232910156,
       "learning_rate": 0.0005912833243389097,
-      "loss": 4.8149,
+      "loss": 4.8191,
       "step": 1450
     },
     {
       "epoch": 0.16172506738544473,
-      "grad_norm": 1.0761699676513672,
+      "grad_norm": 0.896648108959198,
       "learning_rate": 0.0005909595250944414,
-      "loss": 4.8135,
+      "loss": 4.8189,
       "step": 1500
     },
     {
       "epoch": 0.16711590296495957,
-      "grad_norm": 1.3190892934799194,
+      "grad_norm": 1.0958908796310425,
       "learning_rate": 0.000590635725849973,
-      "loss": 4.7957,
+      "loss": 4.8031,
       "step": 1550
     },
     {
       "epoch": 0.1725067385444744,
-      "grad_norm": 1.0538240671157837,
+      "grad_norm": 1.2471723556518555,
       "learning_rate": 0.0005903119266055045,
-      "loss": 4.7729,
+      "loss": 4.7804,
       "step": 1600
     },
     {
       "epoch": 0.1778975741239892,
-      "grad_norm": 0.9775657057762146,
+      "grad_norm": 1.2964743375778198,
       "learning_rate": 0.0005899881273610361,
-      "loss": 4.7419,
+      "loss": 4.7491,
       "step": 1650
     },
     {
       "epoch": 0.18328840970350405,
-      "grad_norm": 1.1415176391601562,
+      "grad_norm": 1.2858680486679077,
       "learning_rate": 0.0005896643281165677,
-      "loss": 4.7273,
+      "loss": 4.7306,
       "step": 1700
     },
     {
       "epoch": 0.18867924528301888,
-      "grad_norm": 1.054643154144287,
+      "grad_norm": 1.1489158868789673,
       "learning_rate": 0.0005893405288720993,
-      "loss": 4.676,
+      "loss": 4.6824,
       "step": 1750
     },
     {
       "epoch": 0.1940700808625337,
-      "grad_norm": 0.9605634212493896,
+      "grad_norm": 0.9289330244064331,
       "learning_rate": 0.0005890167296276308,
-      "loss": 4.6987,
+      "loss": 4.7103,
       "step": 1800
     },
     {
       "epoch": 0.19946091644204852,
-      "grad_norm": 1.2974011898040771,
+      "grad_norm": 1.142987608909607,
       "learning_rate": 0.0005886929303831624,
-      "loss": 4.6522,
+      "loss": 4.6613,
       "step": 1850
     },
     {
       "epoch": 0.20485175202156333,
-      "grad_norm": 0.8335526585578918,
+      "grad_norm": 0.80745929479599,
       "learning_rate": 0.0005883691311386939,
-      "loss": 4.6288,
+      "loss": 4.6385,
       "step": 1900
     },
     {
       "epoch": 0.21024258760107817,
-      "grad_norm": 0.8267828822135925,
+      "grad_norm": 0.6197893619537354,
       "learning_rate": 0.0005880453318942256,
-      "loss": 4.6095,
+      "loss": 4.6164,
       "step": 1950
     },
     {
       "epoch": 0.215633423180593,
-      "grad_norm": 0.8468182682991028,
+      "grad_norm": 0.6791684627532959,
       "learning_rate": 0.0005877215326497571,
-      "loss": 4.5942,
+      "loss": 4.6009,
       "step": 2000
     },
     {
       "epoch": 0.215633423180593,
-      "eval_accuracy": 0.2697699090390619,
-      "eval_loss": 4.5105061531066895,
-      "eval_runtime": 185.0374,
-      "eval_samples_per_second": 97.337,
-      "eval_steps_per_second": 6.085,
+      "eval_accuracy": 0.26871010877131946,
+      "eval_loss": 4.521060943603516,
+      "eval_runtime": 184.8086,
+      "eval_samples_per_second": 97.458,
+      "eval_steps_per_second": 6.093,
       "step": 2000
     },
     {
       "epoch": 0.2210242587601078,
-      "grad_norm": 0.8596065044403076,
+      "grad_norm": 0.865337073802948,
       "learning_rate": 0.0005873977334052887,
-      "loss": 4.577,
+      "loss": 4.5827,
       "step": 2050
     },
     {
       "epoch": 0.22641509433962265,
-      "grad_norm": 0.8865411877632141,
+      "grad_norm": 1.018101453781128,
       "learning_rate": 0.0005870739341608202,
-      "loss": 4.5573,
+      "loss": 4.5648,
       "step": 2100
     },
     {
       "epoch": 0.23180592991913745,
-      "grad_norm": 0.759678065776825,
+      "grad_norm": 0.7497203350067139,
       "learning_rate": 0.0005867501349163519,
-      "loss": 4.531,
+      "loss": 4.5389,
       "step": 2150
     },
     {
       "epoch": 0.2371967654986523,
-      "grad_norm": 0.9333673715591431,
+      "grad_norm": 0.9881597757339478,
       "learning_rate": 0.0005864263356718833,
-      "loss": 4.5288,
+      "loss": 4.5317,
       "step": 2200
     },
     {
       "epoch": 0.24258760107816713,
-      "grad_norm": 0.9579219818115234,
+      "grad_norm": 0.8000950813293457,
       "learning_rate": 0.000586102536427415,
-      "loss": 4.5046,
+      "loss": 4.5076,
       "step": 2250
     },
     {
       "epoch": 0.24797843665768193,
-      "grad_norm": 0.9088818430900574,
+      "grad_norm": 0.8468273878097534,
       "learning_rate": 0.0005857787371829465,
-      "loss": 4.498,
+      "loss": 4.4986,
       "step": 2300
     },
     {
       "epoch": 0.25336927223719674,
-      "grad_norm": 1.0662283897399902,
+      "grad_norm": 1.2672775983810425,
       "learning_rate": 0.0005854549379384781,
-      "loss": 4.476,
+      "loss": 4.4835,
       "step": 2350
     },
     {
       "epoch": 0.2587601078167116,
-      "grad_norm": 0.8263185024261475,
+      "grad_norm": 0.8388407826423645,
       "learning_rate": 0.0005851311386940096,
-      "loss": 4.4585,
+      "loss": 4.4637,
       "step": 2400
     },
     {
       "epoch": 0.2641509433962264,
-      "grad_norm": 0.7838813066482544,
+      "grad_norm": 0.8386316895484924,
       "learning_rate": 0.0005848073394495412,
-      "loss": 4.4501,
+      "loss": 4.4537,
       "step": 2450
     },
     {
       "epoch": 0.2695417789757412,
-      "grad_norm": 1.0609959363937378,
+      "grad_norm": 0.9034900665283203,
       "learning_rate": 0.0005844835402050728,
-      "loss": 4.4276,
+      "loss": 4.43,
       "step": 2500
     },
     {
       "epoch": 0.2749326145552561,
-      "grad_norm": 0.7959164381027222,
+      "grad_norm": 0.8214098215103149,
       "learning_rate": 0.0005841597409606044,
-      "loss": 4.3959,
+      "loss": 4.4001,
       "step": 2550
     },
     {
       "epoch": 0.2803234501347709,
-      "grad_norm": 0.7520493268966675,
+      "grad_norm": 0.803986132144928,
       "learning_rate": 0.000583835941716136,
-      "loss": 4.4101,
+      "loss": 4.4119,
       "step": 2600
     },
     {
       "epoch": 0.2857142857142857,
-      "grad_norm": 0.8316442370414734,
+      "grad_norm": 0.7659591436386108,
       "learning_rate": 0.0005835121424716675,
-      "loss": 4.3887,
+      "loss": 4.3984,
       "step": 2650
     },
     {
       "epoch": 0.29110512129380056,
-      "grad_norm": 0.8290981650352478,
+      "grad_norm": 0.6612359285354614,
       "learning_rate": 0.0005831883432271992,
-      "loss": 4.3746,
+      "loss": 4.3813,
       "step": 2700
     },
     {
       "epoch": 0.29649595687331537,
-      "grad_norm": 0.8543332815170288,
+      "grad_norm": 0.8132438659667969,
       "learning_rate": 0.0005828645439827307,
-      "loss": 4.3708,
+      "loss": 4.375,
       "step": 2750
     },
     {
       "epoch": 0.3018867924528302,
-      "grad_norm": 1.093960165977478,
+      "grad_norm": 0.734201192855835,
       "learning_rate": 0.0005825407447382622,
-      "loss": 4.329,
+      "loss": 4.3299,
       "step": 2800
     },
     {
       "epoch": 0.30727762803234504,
-      "grad_norm": 0.704647958278656,
+      "grad_norm": 0.6258540153503418,
       "learning_rate": 0.0005822169454937938,
-      "loss": 4.3542,
+      "loss": 4.3571,
       "step": 2850
     },
     {
       "epoch": 0.31266846361185985,
-      "grad_norm": 0.7982504963874817,
+      "grad_norm": 0.77208411693573,
       "learning_rate": 0.0005818931462493254,
-      "loss": 4.3236,
+      "loss": 4.3253,
       "step": 2900
     },
     {
       "epoch": 0.31805929919137466,
-      "grad_norm": 0.7402891516685486,
+      "grad_norm": 0.7789588570594788,
       "learning_rate": 0.0005815693470048569,
-      "loss": 4.3262,
+      "loss": 4.3301,
       "step": 2950
     },
     {
       "epoch": 0.32345013477088946,
-      "grad_norm": 0.7455512285232544,
+      "grad_norm": 0.6333003640174866,
       "learning_rate": 0.0005812455477603885,
-      "loss": 4.3169,
+      "loss": 4.3192,
       "step": 3000
     },
     {
       "epoch": 0.32345013477088946,
-      "eval_accuracy": 0.2977507439734861,
-      "eval_loss": 4.238898754119873,
-      "eval_runtime": 185.2161,
-      "eval_samples_per_second": 97.243,
-      "eval_steps_per_second": 6.079,
+      "eval_accuracy": 0.2970622106234057,
+      "eval_loss": 4.244666576385498,
+      "eval_runtime": 184.9563,
+      "eval_samples_per_second": 97.38,
+      "eval_steps_per_second": 6.088,
       "step": 3000
     },
     {
       "epoch": 0.3288409703504043,
-      "grad_norm": 0.7529988288879395,
+      "grad_norm": 0.8093159198760986,
       "learning_rate": 0.0005809217485159201,
-      "loss": 4.3159,
+      "loss": 4.3201,
       "step": 3050
     },
     {
       "epoch": 0.33423180592991913,
-      "grad_norm": 0.7738096117973328,
+      "grad_norm": 0.6877481341362,
       "learning_rate": 0.0005805979492714517,
-      "loss": 4.2958,
+      "loss": 4.2994,
       "step": 3100
     },
     {
       "epoch": 0.33962264150943394,
-      "grad_norm": 0.9844054579734802,
+      "grad_norm": 0.8910661339759827,
       "learning_rate": 0.0005802741500269832,
-      "loss": 4.2765,
+      "loss": 4.277,
       "step": 3150
     },
     {
       "epoch": 0.3450134770889488,
-      "grad_norm": 0.6849039793014526,
+      "grad_norm": 0.6189978122711182,
       "learning_rate": 0.0005799503507825148,
-      "loss": 4.2784,
+      "loss": 4.2794,
       "step": 3200
     },
     {
       "epoch": 0.3504043126684636,
-      "grad_norm": 0.8223270177841187,
+      "grad_norm": 0.7318760752677917,
       "learning_rate": 0.0005796265515380463,
-      "loss": 4.2661,
+      "loss": 4.2705,
       "step": 3250
     },
     {
       "epoch": 0.3557951482479784,
-      "grad_norm": 1.0759813785552979,
+      "grad_norm": 0.853126049041748,
       "learning_rate": 0.000579302752293578,
-      "loss": 4.2702,
+      "loss": 4.2752,
       "step": 3300
     },
     {
       "epoch": 0.3611859838274933,
-      "grad_norm": 0.8592544794082642,
+      "grad_norm": 0.7733789682388306,
       "learning_rate": 0.0005789789530491095,
-      "loss": 4.2441,
+      "loss": 4.245,
       "step": 3350
     },
     {
       "epoch": 0.3665768194070081,
-      "grad_norm": 0.7978523373603821,
+      "grad_norm": 0.7828088998794556,
       "learning_rate": 0.0005786551538046411,
-      "loss": 4.2357,
+      "loss": 4.2378,
       "step": 3400
     },
     {
       "epoch": 0.3719676549865229,
-      "grad_norm": 0.671616792678833,
+      "grad_norm": 0.6192962527275085,
       "learning_rate": 0.0005783313545601726,
-      "loss": 4.2296,
+      "loss": 4.2325,
       "step": 3450
     },
     {
       "epoch": 0.37735849056603776,
-      "grad_norm": 0.9814181327819824,
+      "grad_norm": 0.9660407304763794,
       "learning_rate": 0.0005780075553157043,
-      "loss": 4.2266,
+      "loss": 4.2287,
       "step": 3500
     },
     {
       "epoch": 0.38274932614555257,
-      "grad_norm": 0.6863210797309875,
+      "grad_norm": 0.6514875888824463,
       "learning_rate": 0.0005776837560712357,
-      "loss": 4.2231,
+      "loss": 4.2273,
       "step": 3550
     },
     {
       "epoch": 0.3881401617250674,
-      "grad_norm": 0.8627532720565796,
+      "grad_norm": 0.7967301607131958,
       "learning_rate": 0.0005773599568267673,
-      "loss": 4.213,
+      "loss": 4.2175,
       "step": 3600
     },
     {
       "epoch": 0.3935309973045822,
-      "grad_norm": 0.7562503218650818,
+      "grad_norm": 0.7382152676582336,
       "learning_rate": 0.0005770361575822989,
-      "loss": 4.2039,
+      "loss": 4.2058,
       "step": 3650
     },
     {
       "epoch": 0.39892183288409705,
-      "grad_norm": 0.7050570249557495,
+      "grad_norm": 0.6826912760734558,
       "learning_rate": 0.0005767123583378305,
-      "loss": 4.1944,
+      "loss": 4.1984,
       "step": 3700
     },
     {
       "epoch": 0.40431266846361186,
-      "grad_norm": 0.7500070929527283,
+      "grad_norm": 0.7629984617233276,
       "learning_rate": 0.000576388559093362,
-      "loss": 4.1959,
+      "loss": 4.1999,
       "step": 3750
     },
     {
       "epoch": 0.40970350404312667,
-      "grad_norm": 0.7112507224082947,
+      "grad_norm": 0.7264732122421265,
       "learning_rate": 0.0005760647598488936,
-      "loss": 4.1652,
+      "loss": 4.1661,
       "step": 3800
     },
     {
       "epoch": 0.41509433962264153,
-      "grad_norm": 0.6652435660362244,
+      "grad_norm": 0.6775606870651245,
       "learning_rate": 0.0005757409606044253,
       "loss": 4.1705,
       "step": 3850
     },
     {
       "epoch": 0.42048517520215634,
-      "grad_norm": 0.7143222093582153,
+      "grad_norm": 0.6371147036552429,
       "learning_rate": 0.0005754171613599568,
-      "loss": 4.1853,
+      "loss": 4.1898,
       "step": 3900
     },
     {
       "epoch": 0.42587601078167114,
-      "grad_norm": 1.307312250137329,
+      "grad_norm": 0.9279155731201172,
       "learning_rate": 0.0005750933621154884,
-      "loss": 4.155,
+      "loss": 4.1564,
       "step": 3950
     },
     {
       "epoch": 0.431266846361186,
-      "grad_norm": 0.7014321088790894,
+      "grad_norm": 0.7013996243476868,
       "learning_rate": 0.0005747695628710199,
-      "loss": 4.1568,
+      "loss": 4.1584,
       "step": 4000
     },
     {
       "epoch": 0.431266846361186,
-      "eval_accuracy": 0.3122283066632144,
-      "eval_loss": 4.091763496398926,
-      "eval_runtime": 184.8885,
-      "eval_samples_per_second": 97.415,
-      "eval_steps_per_second": 6.09,
+      "eval_accuracy": 0.3116742855882484,
+      "eval_loss": 4.094541549682617,
+      "eval_runtime": 184.9211,
+      "eval_samples_per_second": 97.398,
+      "eval_steps_per_second": 6.089,
       "step": 4000
     },
     {
       "epoch": 0.4366576819407008,
-      "grad_norm": 0.6528450846672058,
+      "grad_norm": 0.6695048809051514,
       "learning_rate": 0.0005744457636265515,
-      "loss": 4.1642,
+      "loss": 4.1673,
       "step": 4050
     },
     {
       "epoch": 0.4420485175202156,
-      "grad_norm": 0.6965686082839966,
+      "grad_norm": 0.7323116660118103,
       "learning_rate": 0.0005741219643820831,
-      "loss": 4.1599,
+      "loss": 4.1606,
       "step": 4100
     },
     {
       "epoch": 0.4474393530997305,
-      "grad_norm": 0.672761857509613,
+      "grad_norm": 0.6928706169128418,
       "learning_rate": 0.0005737981651376146,
-      "loss": 4.1398,
+      "loss": 4.1431,
       "step": 4150
     },
     {
       "epoch": 0.4528301886792453,
-      "grad_norm": 0.6647113561630249,
+      "grad_norm": 0.7627272009849548,
       "learning_rate": 0.0005734743658931462,
-      "loss": 4.1447,
+      "loss": 4.1496,
       "step": 4200
     },
     {
       "epoch": 0.4582210242587601,
-      "grad_norm": 0.5870072245597839,
+      "grad_norm": 0.5734204053878784,
       "learning_rate": 0.0005731505666486778,
-      "loss": 4.1341,
+      "loss": 4.1349,
       "step": 4250
     },
     {
       "epoch": 0.4636118598382749,
-      "grad_norm": 0.7313339710235596,
+      "grad_norm": 0.7368983626365662,
       "learning_rate": 0.0005728267674042093,
-      "loss": 4.1366,
+      "loss": 4.135,
       "step": 4300
     },
     {
       "epoch": 0.46900269541778977,
-      "grad_norm": 0.8136706352233887,
+      "grad_norm": 0.7932204008102417,
       "learning_rate": 0.0005725029681597409,
-      "loss": 4.124,
+      "loss": 4.1263,
       "step": 4350
     },
     {
       "epoch": 0.4743935309973046,
-      "grad_norm": 0.6834421753883362,
+      "grad_norm": 0.69672030210495,
       "learning_rate": 0.0005721791689152725,
-      "loss": 4.1078,
+      "loss": 4.1093,
       "step": 4400
     },
     {
       "epoch": 0.4797843665768194,
-      "grad_norm": 0.6508508920669556,
+      "grad_norm": 0.743208110332489,
       "learning_rate": 0.0005718553696708041,
-      "loss": 4.1152,
+      "loss": 4.1153,
       "step": 4450
     },
     {
       "epoch": 0.48517520215633425,
-      "grad_norm": 0.6280442476272583,
+      "grad_norm": 0.6159020066261292,
       "learning_rate": 0.0005715315704263356,
-      "loss": 4.1121,
+      "loss": 4.1113,
       "step": 4500
     },
     {
       "epoch": 0.49056603773584906,
-      "grad_norm": 0.7698476910591125,
+      "grad_norm": 0.7856785655021667,
       "learning_rate": 0.0005712077711818672,
-      "loss": 4.1191,
+      "loss": 4.1188,
       "step": 4550
     },
     {
       "epoch": 0.49595687331536387,
-      "grad_norm": 0.7502248883247375,
+      "grad_norm": 0.8774890899658203,
       "learning_rate": 0.0005708839719373987,
-      "loss": 4.1148,
+      "loss": 4.1134,
       "step": 4600
     },
     {
       "epoch": 0.5013477088948787,
-      "grad_norm": 0.6504122018814087,
+      "grad_norm": 0.6919339299201965,
       "learning_rate": 0.0005705601726929304,
-      "loss": 4.0719,
+      "loss": 4.0736,
       "step": 4650
     },
     {
       "epoch": 0.5067385444743935,
-      "grad_norm": 0.663817286491394,
+      "grad_norm": 0.7403694987297058,
       "learning_rate": 0.0005702363734484619,
-      "loss": 4.0717,
+      "loss": 4.0742,
       "step": 4700
     },
     {
       "epoch": 0.5121293800539084,
-      "grad_norm": 0.6721544861793518,
+      "grad_norm": 0.6341198086738586,
       "learning_rate": 0.0005699125742039935,
-      "loss": 4.0791,
+      "loss": 4.0777,
       "step": 4750
     },
     {
       "epoch": 0.5175202156334232,
-      "grad_norm": 0.6749738454818726,
+      "grad_norm": 0.6231762170791626,
       "learning_rate": 0.000569588774959525,
       "loss": 4.074,
       "step": 4800
     },
     {
       "epoch": 0.522911051212938,
-      "grad_norm": 0.640407383441925,
+      "grad_norm": 0.6197823286056519,
       "learning_rate": 0.0005692649757150567,
-      "loss": 4.0707,
+      "loss": 4.0696,
       "step": 4850
     },
     {
       "epoch": 0.5283018867924528,
-      "grad_norm": 0.7699702382087708,
+      "grad_norm": 0.7616347670555115,
       "learning_rate": 0.0005689411764705881,
-      "loss": 4.071,
+      "loss": 4.072,
       "step": 4900
     },
     {
       "epoch": 0.5336927223719676,
-      "grad_norm": 0.7380169630050659,
+      "grad_norm": 0.6769264340400696,
       "learning_rate": 0.0005686173772261197,
-      "loss": 4.0611,
+      "loss": 4.0616,
       "step": 4950
     },
     {
       "epoch": 0.5390835579514824,
-      "grad_norm": 0.7213010191917419,
+      "grad_norm": 0.8056491017341614,
       "learning_rate": 0.0005682935779816514,
-      "loss": 4.0605,
+      "loss": 4.0604,
       "step": 5000
     },
     {
       "epoch": 0.5390835579514824,
-      "eval_accuracy": 0.32159711922079365,
-      "eval_loss": 3.9895858764648438,
-      "eval_runtime": 184.9121,
-      "eval_samples_per_second": 97.403,
-      "eval_steps_per_second": 6.089,
+      "eval_accuracy": 0.32076407752992003,
+      "eval_loss": 3.9996907711029053,
+      "eval_runtime": 185.0096,
+      "eval_samples_per_second": 97.352,
+      "eval_steps_per_second": 6.086,
       "step": 5000
     },
     {
       "epoch": 0.5444743935309974,
-      "grad_norm": 0.6962867379188538,
+      "grad_norm": 0.5992985367774963,
       "learning_rate": 0.0005679697787371829,
-      "loss": 4.0683,
+      "loss": 4.0685,
       "step": 5050
     },
     {
       "epoch": 0.5498652291105122,
-      "grad_norm": 0.6025412678718567,
+      "grad_norm": 0.623742938041687,
       "learning_rate": 0.0005676459794927145,
-      "loss": 4.0577,
+      "loss": 4.0541,
       "step": 5100
     },
     {
       "epoch": 0.555256064690027,
-      "grad_norm": 0.7060973048210144,
+      "grad_norm": 0.670574963092804,
       "learning_rate": 0.000567322180248246,
-      "loss": 4.0478,
+      "loss": 4.049,
       "step": 5150
     },
     {
       "epoch": 0.5606469002695418,
-      "grad_norm": 0.7477322816848755,
+      "grad_norm": 0.6862586140632629,
       "learning_rate": 0.0005669983810037777,
-      "loss": 4.0597,
+      "loss": 4.0581,
       "step": 5200
     },
     {
       "epoch": 0.5660377358490566,
-      "grad_norm": 0.6687774658203125,
+      "grad_norm": 0.7968271374702454,
       "learning_rate": 0.0005666745817593092,
-      "loss": 4.0342,
+      "loss": 4.0331,
       "step": 5250
     },
     {
       "epoch": 0.5714285714285714,
-      "grad_norm": 0.5782555937767029,
+      "grad_norm": 0.6383443474769592,
       "learning_rate": 0.0005663507825148408,
-      "loss": 4.0336,
+      "loss": 4.0329,
       "step": 5300
     },
     {
       "epoch": 0.5768194070080862,
-      "grad_norm": 0.7279411554336548,
+      "grad_norm": 0.6728504300117493,
       "learning_rate": 0.0005660269832703723,
-      "loss": 4.023,
+      "loss": 4.0234,
       "step": 5350
     },
     {
       "epoch": 0.5822102425876011,
-      "grad_norm": 0.6305038332939148,
+      "grad_norm": 0.6788572669029236,
       "learning_rate": 0.0005657031840259039,
-      "loss": 4.0282,
+      "loss": 4.0272,
       "step": 5400
     },
     {
       "epoch": 0.5876010781671159,
-      "grad_norm": 0.7332347631454468,
+      "grad_norm": 0.7220979928970337,
       "learning_rate": 0.0005653793847814355,
-      "loss": 4.0282,
+      "loss": 4.0266,
       "step": 5450
     },
     {
       "epoch": 0.5929919137466307,
-      "grad_norm": 0.7261084318161011,
+      "grad_norm": 0.7897782921791077,
       "learning_rate": 0.000565055585536967,
-      "loss": 4.0045,
+      "loss": 4.004,
       "step": 5500
     },
     {
       "epoch": 0.5983827493261455,
-      "grad_norm": 0.6593828797340393,
+      "grad_norm": 0.6859530210494995,
       "learning_rate": 0.0005647317862924986,
-      "loss": 4.0288,
+      "loss": 4.0272,
       "step": 5550
     },
     {
       "epoch": 0.6037735849056604,
-      "grad_norm": 0.6685374975204468,
+      "grad_norm": 0.6835060715675354,
       "learning_rate": 0.0005644079870480302,
-      "loss": 3.9767,
+      "loss": 3.9762,
       "step": 5600
     },
     {
       "epoch": 0.6091644204851752,
-      "grad_norm": 0.6912181973457336,
+      "grad_norm": 0.6331577301025391,
       "learning_rate": 0.0005640841878035617,
-      "loss": 4.013,
+      "loss": 4.0138,
       "step": 5650
     },
     {
       "epoch": 0.6145552560646901,
-      "grad_norm": 0.7449594736099243,
+      "grad_norm": 0.6407967209815979,
       "learning_rate": 0.0005637603885590933,
-      "loss": 4.0117,
+      "loss": 4.0133,
       "step": 5700
     },
     {
       "epoch": 0.6199460916442049,
-      "grad_norm": 0.6433312892913818,
+      "grad_norm": 0.5845387578010559,
       "learning_rate": 0.0005634365893146248,
-      "loss": 4.012,
+      "loss": 4.0109,
       "step": 5750
     },
     {
       "epoch": 0.6253369272237197,
-      "grad_norm": 0.6818335652351379,
+      "grad_norm": 0.660351037979126,
       "learning_rate": 0.0005631127900701565,
-      "loss": 4.002,
+      "loss": 4.0022,
       "step": 5800
     },
     {
       "epoch": 0.6307277628032345,
-      "grad_norm": 0.6618849039077759,
+      "grad_norm": 0.607296347618103,
       "learning_rate": 0.000562788990825688,
-      "loss": 3.9769,
+      "loss": 3.9779,
       "step": 5850
     },
     {
       "epoch": 0.6361185983827493,
-      "grad_norm": 0.5509379506111145,
+      "grad_norm": 0.5927079916000366,
       "learning_rate": 0.0005624651915812196,
-      "loss": 3.989,
+      "loss": 3.9895,
       "step": 5900
     },
     {
       "epoch": 0.6415094339622641,
-      "grad_norm": 0.794357419013977,
+      "grad_norm": 0.6702271103858948,
       "learning_rate": 0.0005621413923367511,
-      "loss": 3.995,
+      "loss": 3.9958,
       "step": 5950
     },
     {
       "epoch": 0.6469002695417789,
-      "grad_norm": 0.5914753675460815,
+      "grad_norm": 0.6015697121620178,
       "learning_rate": 0.0005618175930922828,
-      "loss": 4.0027,
+      "loss": 3.9995,
       "step": 6000
     },
     {
       "epoch": 0.6469002695417789,
-      "eval_accuracy": 0.32826688539217336,
-      "eval_loss": 3.91684627532959,
-      "eval_runtime": 184.8458,
-      "eval_samples_per_second": 97.438,
-      "eval_steps_per_second": 6.092,
+      "eval_accuracy": 0.32856003088349683,
+      "eval_loss": 3.9154956340789795,
+      "eval_runtime": 184.9163,
+      "eval_samples_per_second": 97.401,
+      "eval_steps_per_second": 6.089,
       "step": 6000
     },
     {
       "epoch": 0.6522911051212938,
-      "grad_norm": 0.6078530550003052,
+      "grad_norm": 0.6032575368881226,
       "learning_rate": 0.0005614937938478143,
-      "loss": 3.9924,
+      "loss": 3.9936,
       "step": 6050
     },
     {
       "epoch": 0.6576819407008087,
-      "grad_norm": 0.6487135887145996,
+      "grad_norm": 0.67863529920578,
       "learning_rate": 0.0005611699946033459,
-      "loss": 3.9847,
+      "loss": 3.9843,
       "step": 6100
     },
     {
       "epoch": 0.6630727762803235,
-      "grad_norm": 0.5724067687988281,
+      "grad_norm": 0.5835185647010803,
       "learning_rate": 0.0005608461953588774,
-      "loss": 3.9778,
+      "loss": 3.9774,
       "step": 6150
     },
     {
       "epoch": 0.6684636118598383,
-      "grad_norm": 0.5982683300971985,
+      "grad_norm": 0.6360609531402588,
       "learning_rate": 0.000560522396114409,
-      "loss": 3.9716,
+      "loss": 3.9718,
       "step": 6200
     },
     {
       "epoch": 0.6738544474393531,
-      "grad_norm": 0.6258571743965149,
+      "grad_norm": 0.5755443572998047,
       "learning_rate": 0.0005601985968699405,
-      "loss": 3.9527,
+      "loss": 3.9553,
       "step": 6250
     },
     {
       "epoch": 0.6792452830188679,
-      "grad_norm": 0.6944162249565125,
+      "grad_norm": 0.6723095178604126,
       "learning_rate": 0.0005598747976254721,
-      "loss": 3.9692,
+      "loss": 3.97,
       "step": 6300
     },
     {
       "epoch": 0.6846361185983828,
-      "grad_norm": 0.5871492624282837,
+      "grad_norm": 0.6210877299308777,
       "learning_rate": 0.0005595509983810038,
-      "loss": 3.94,
+      "loss": 3.9446,
       "step": 6350
     },
     {
       "epoch": 0.6900269541778976,
-      "grad_norm": 0.6267895102500916,
+      "grad_norm": 0.6090090870857239,
       "learning_rate": 0.0005592271991365353,
-      "loss": 3.9551,
+      "loss": 3.9554,
       "step": 6400
     },
     {
       "epoch": 0.6954177897574124,
-      "grad_norm": 0.6211588978767395,
+      "grad_norm": 0.6561355590820312,
       "learning_rate": 0.0005589033998920669,
-      "loss": 3.9716,
+      "loss": 3.9706,
       "step": 6450
     },
     {
       "epoch": 0.7008086253369272,
-      "grad_norm": 0.6593815088272095,
+      "grad_norm": 0.6227344870567322,
       "learning_rate": 0.0005585796006475984,
-      "loss": 3.9397,
+      "loss": 3.9421,
       "step": 6500
     },
     {
       "epoch": 0.706199460916442,
-      "grad_norm": 0.7002833485603333,
+      "grad_norm": 0.679628312587738,
       "learning_rate": 0.0005582558014031301,
-      "loss": 3.9657,
+      "loss": 3.9669,
       "step": 6550
     },
     {
       "epoch": 0.7115902964959568,
-      "grad_norm": 0.6196280121803284,
+      "grad_norm": 0.5673393607139587,
       "learning_rate": 0.0005579320021586616,
-      "loss": 3.9491,
+      "loss": 3.9506,
       "step": 6600
     },
     {
       "epoch": 0.7169811320754716,
-      "grad_norm": 0.7528320550918579,
+      "grad_norm": 0.603226900100708,
       "learning_rate": 0.0005576082029141932,
-      "loss": 3.9453,
+      "loss": 3.9447,
       "step": 6650
     },
     {
       "epoch": 0.7223719676549866,
-      "grad_norm": 0.5507506728172302,
+      "grad_norm": 0.5796992778778076,
       "learning_rate": 0.0005572844036697247,
-      "loss": 3.9307,
+      "loss": 3.9305,
       "step": 6700
     },
     {
       "epoch": 0.7277628032345014,
-      "grad_norm": 0.5690281987190247,
+      "grad_norm": 0.5560516119003296,
       "learning_rate": 0.0005569606044252563,
-      "loss": 3.9365,
+      "loss": 3.9388,
       "step": 6750
     },
     {
       "epoch": 0.7331536388140162,
-      "grad_norm": 0.8382434844970703,
+      "grad_norm": 0.711595356464386,
       "learning_rate": 0.0005566368051807879,
-      "loss": 3.9536,
+      "loss": 3.9534,
       "step": 6800
     },
     {
       "epoch": 0.738544474393531,
-      "grad_norm": 0.7441468238830566,
+      "grad_norm": 0.6935557126998901,
       "learning_rate": 0.0005563130059363194,
-      "loss": 3.9389,
+      "loss": 3.9397,
       "step": 6850
     },
     {
       "epoch": 0.7439353099730458,
-      "grad_norm": 0.5325085520744324,
+      "grad_norm": 0.5889500379562378,
       "learning_rate": 0.000555989206691851,
-      "loss": 3.9368,
+      "loss": 3.9376,
       "step": 6900
     },
     {
       "epoch": 0.7493261455525606,
-      "grad_norm": 0.6673761606216431,
+      "grad_norm": 0.6055012941360474,
       "learning_rate": 0.0005556654074473826,
-      "loss": 3.942,
+      "loss": 3.9432,
       "step": 6950
     },
     {
       "epoch": 0.7547169811320755,
-      "grad_norm": 0.6331908702850342,
+      "grad_norm": 0.7356480956077576,
       "learning_rate": 0.0005553416082029141,
-      "loss": 3.9256,
+      "loss": 3.9243,
       "step": 7000
     },
     {
       "epoch": 0.7547169811320755,
-      "eval_accuracy": 0.33349167680850833,
-      "eval_loss": 3.861177682876587,
-      "eval_runtime": 184.9411,
-      "eval_samples_per_second": 97.388,
-      "eval_steps_per_second": 6.088,
+      "eval_accuracy": 0.3333837844909085,
+      "eval_loss": 3.861570358276367,
+      "eval_runtime": 185.0498,
+      "eval_samples_per_second": 97.331,
+      "eval_steps_per_second": 6.085,
       "step": 7000
     },
     {
       "epoch": 0.7601078167115903,
-      "grad_norm": 0.5495949387550354,
+      "grad_norm": 0.6299076676368713,
       "learning_rate": 0.0005550178089584457,
       "loss": 3.9179,
       "step": 7050
     },
     {
       "epoch": 0.7654986522911051,
-      "grad_norm": 0.5872019529342651,
+      "grad_norm": 0.6104505062103271,
       "learning_rate": 0.0005546940097139772,
-      "loss": 3.9178,
+      "loss": 3.9182,
       "step": 7100
     },
     {
       "epoch": 0.77088948787062,
-      "grad_norm": 0.6886711120605469,
+      "grad_norm": 0.7075724005699158,
       "learning_rate": 0.0005543702104695089,
-      "loss": 3.9203,
+      "loss": 3.921,
       "step": 7150
     },
     {
       "epoch": 0.7762803234501348,
-      "grad_norm": 0.5374932885169983,
+      "grad_norm": 0.5794752240180969,
       "learning_rate": 0.0005540464112250404,
-      "loss": 3.918,
+      "loss": 3.9183,
       "step": 7200
     },
     {
       "epoch": 0.7816711590296496,
-      "grad_norm": 0.6181663274765015,
+      "grad_norm": 0.6488854289054871,
       "learning_rate": 0.000553722611980572,
-      "loss": 3.9298,
+      "loss": 3.9308,
       "step": 7250
     },
     {
       "epoch": 0.7870619946091644,
-      "grad_norm": 0.5765889883041382,
+      "grad_norm": 0.632280707359314,
       "learning_rate": 0.0005533988127361035,
-      "loss": 3.9127,
+      "loss": 3.9145,
       "step": 7300
     },
     {
       "epoch": 0.7924528301886793,
-      "grad_norm": 0.6276585459709167,
+      "grad_norm": 0.5979216694831848,
       "learning_rate": 0.0005530750134916352,
-      "loss": 3.9291,
+      "loss": 3.9286,
       "step": 7350
     },
     {
       "epoch": 0.7978436657681941,
-      "grad_norm": 0.6201632618904114,
+      "grad_norm": 0.6133264899253845,
       "learning_rate": 0.0005527512142471668,
-      "loss": 3.9022,
+      "loss": 3.9042,
       "step": 7400
     },
     {
       "epoch": 0.8032345013477089,
-      "grad_norm": 0.5955806970596313,
+      "grad_norm": 0.6683976650238037,
       "learning_rate": 0.0005524274150026982,
-      "loss": 3.9034,
+      "loss": 3.9044,
       "step": 7450
     },
     {
       "epoch": 0.8086253369272237,
-      "grad_norm": 0.6247159242630005,
+      "grad_norm": 0.5934157967567444,
       "learning_rate": 0.0005521036157582299,
-      "loss": 3.8948,
+      "loss": 3.8953,
       "step": 7500
     },
     {
       "epoch": 0.8140161725067385,
-      "grad_norm": 0.6614282131195068,
+      "grad_norm": 0.6800050735473633,
       "learning_rate": 0.0005517798165137614,
-      "loss": 3.8995,
+      "loss": 3.9003,
       "step": 7550
     },
     {
       "epoch": 0.8194070080862533,
-      "grad_norm": 0.599558413028717,
+      "grad_norm": 0.6356220841407776,
       "learning_rate": 0.000551456017269293,
-      "loss": 3.8749,
+      "loss": 3.8741,
       "step": 7600
     },
     {
       "epoch": 0.8247978436657682,
-      "grad_norm": 0.5789263844490051,
+      "grad_norm": 0.6044333577156067,
       "learning_rate": 0.0005511322180248245,
-      "loss": 3.8905,
+      "loss": 3.8914,
       "step": 7650
     },
     {
       "epoch": 0.8301886792452831,
-      "grad_norm": 0.616805911064148,
+      "grad_norm": 0.6075876951217651,
       "learning_rate": 0.0005508084187803562,
-      "loss": 3.9028,
+      "loss": 3.9014,
       "step": 7700
     },
     {
       "epoch": 0.8355795148247979,
-      "grad_norm": 0.5778168439865112,
+      "grad_norm": 0.5683922171592712,
       "learning_rate": 0.0005504846195358877,
-      "loss": 3.8969,
+      "loss": 3.8975,
       "step": 7750
     },
     {
       "epoch": 0.8409703504043127,
-      "grad_norm": 0.5527185201644897,
+      "grad_norm": 0.6191318035125732,
       "learning_rate": 0.0005501608202914193,
-      "loss": 3.8859,
+      "loss": 3.8865,
       "step": 7800
     },
     {
       "epoch": 0.8463611859838275,
-      "grad_norm": 0.6867278218269348,
+      "grad_norm": 0.6902214288711548,
       "learning_rate": 0.0005498370210469508,
-      "loss": 3.8885,
+      "loss": 3.8898,
       "step": 7850
     },
     {
       "epoch": 0.8517520215633423,
-      "grad_norm": 0.6439316272735596,
+      "grad_norm": 0.6202863454818726,
       "learning_rate": 0.0005495132218024824,
-      "loss": 3.8741,
+      "loss": 3.875,
       "step": 7900
     },
     {
       "epoch": 0.8571428571428571,
-      "grad_norm": 0.5910453796386719,
+      "grad_norm": 0.594792902469635,
       "learning_rate": 0.000549189422558014,
-      "loss": 3.8856,
+      "loss": 3.887,
       "step": 7950
     },
     {
       "epoch": 0.862533692722372,
-      "grad_norm": 0.5654035210609436,
+      "grad_norm": 0.5793900489807129,
       "learning_rate": 0.0005488656233135456,
-      "loss": 3.8821,
+      "loss": 3.8798,
       "step": 8000
     },
     {
       "epoch": 0.862533692722372,
-      "eval_accuracy": 0.33826936159148663,
-      "eval_loss": 3.815802574157715,
-      "eval_runtime": 184.9876,
-      "eval_samples_per_second": 97.363,
-      "eval_steps_per_second": 6.087,
+      "eval_accuracy": 0.33811703024277073,
+      "eval_loss": 3.8144943714141846,
+      "eval_runtime": 184.7794,
+      "eval_samples_per_second": 97.473,
+      "eval_steps_per_second": 6.094,
       "step": 8000
     },
     {
       "epoch": 0.8679245283018868,
-      "grad_norm": 0.6305513381958008,
+      "grad_norm": 0.5921681523323059,
       "learning_rate": 0.0005485418240690771,
-      "loss": 3.8629,
+      "loss": 3.8632,
       "step": 8050
     },
     {
       "epoch": 0.8733153638814016,
-      "grad_norm": 0.6361500024795532,
+      "grad_norm": 0.6462425589561462,
       "learning_rate": 0.0005482180248246087,
-      "loss": 3.8813,
+      "loss": 3.8786,
       "step": 8100
     },
     {
       "epoch": 0.8787061994609164,
-      "grad_norm": 0.650390625,
+      "grad_norm": 0.5800376534461975,
       "learning_rate": 0.0005478942255801403,
-      "loss": 3.8776,
+      "loss": 3.8749,
       "step": 8150
     },
     {
       "epoch": 0.8840970350404312,
-      "grad_norm": 0.5771349668502808,
+      "grad_norm": 0.5411289930343628,
       "learning_rate": 0.0005475704263356718,
-      "loss": 3.863,
+      "loss": 3.8658,
       "step": 8200
     },
     {
       "epoch": 0.889487870619946,
-      "grad_norm": 0.6004456877708435,
+      "grad_norm": 0.5984181761741638,
       "learning_rate": 0.0005472466270912034,
-      "loss": 3.874,
+      "loss": 3.8742,
       "step": 8250
     },
     {
       "epoch": 0.894878706199461,
-      "grad_norm": 0.646091878414154,
+      "grad_norm": 0.5967395901679993,
       "learning_rate": 0.000546922827846735,
-      "loss": 3.862,
+      "loss": 3.8625,
       "step": 8300
     },
     {
       "epoch": 0.9002695417789758,
-      "grad_norm": 0.5574592351913452,
+      "grad_norm": 0.6251577138900757,
       "learning_rate": 0.0005465990286022665,
       "loss": 3.8631,
       "step": 8350
     },
     {
       "epoch": 0.9056603773584906,
-      "grad_norm": 0.6087723970413208,
+      "grad_norm": 0.5644067525863647,
       "learning_rate": 0.0005462752293577981,
       "loss": 3.8792,
       "step": 8400
     },
     {
       "epoch": 0.9110512129380054,
-      "grad_norm": 0.5992222428321838,
+      "grad_norm": 0.5981811285018921,
       "learning_rate": 0.0005459514301133296,
-      "loss": 3.8725,
+      "loss": 3.8723,
       "step": 8450
     },
     {
       "epoch": 0.9164420485175202,
-      "grad_norm": 0.5356247425079346,
+      "grad_norm": 0.5163571834564209,
       "learning_rate": 0.0005456276308688613,
-      "loss": 3.8668,
+      "loss": 3.8679,
       "step": 8500
     },
     {
       "epoch": 0.921832884097035,
-      "grad_norm": 0.5599382519721985,
+      "grad_norm": 0.5923091173171997,
       "learning_rate": 0.0005453038316243929,
-      "loss": 3.861,
+      "loss": 3.8621,
       "step": 8550
     },
     {
       "epoch": 0.9272237196765498,
-      "grad_norm": 0.5570565462112427,
+      "grad_norm": 0.5390773415565491,
       "learning_rate": 0.0005449800323799244,
-      "loss": 3.874,
+      "loss": 3.8713,
       "step": 8600
     },
     {
       "epoch": 0.9326145552560647,
-      "grad_norm": 0.5484678149223328,
+      "grad_norm": 0.5587196350097656,
       "learning_rate": 0.000544656233135456,
-      "loss": 3.8519,
+      "loss": 3.851,
       "step": 8650
     },
     {
       "epoch": 0.9380053908355795,
-      "grad_norm": 0.6040818095207214,
+      "grad_norm": 0.6344064474105835,
       "learning_rate": 0.0005443324338909875,
-      "loss": 3.8544,
+      "loss": 3.8547,
       "step": 8700
     },
     {
       "epoch": 0.9433962264150944,
-      "grad_norm": 0.6044812202453613,
+      "grad_norm": 0.6433781981468201,
       "learning_rate": 0.0005440086346465192,
-      "loss": 3.8669,
+      "loss": 3.8682,
       "step": 8750
     },
     {
       "epoch": 0.9487870619946092,
-      "grad_norm": 0.5890029072761536,
+      "grad_norm": 0.5929216146469116,
       "learning_rate": 0.0005436848354020506,
-      "loss": 3.8472,
+      "loss": 3.8499,
       "step": 8800
     },
     {
       "epoch": 0.954177897574124,
-      "grad_norm": 0.5707723498344421,
+      "grad_norm": 0.5464727282524109,
       "learning_rate": 0.0005433610361575823,
-      "loss": 3.8532,
+      "loss": 3.8531,
       "step": 8850
     },
     {
       "epoch": 0.9595687331536388,
-      "grad_norm": 0.5903648734092712,
+      "grad_norm": 0.5909833908081055,
       "learning_rate": 0.0005430372369131138,
-      "loss": 3.8367,
+      "loss": 3.8389,
       "step": 8900
     },
     {
       "epoch": 0.9649595687331537,
-      "grad_norm": 0.5946788191795349,
+      "grad_norm": 0.6431576013565063,
       "learning_rate": 0.0005427134376686454,
-      "loss": 3.85,
+      "loss": 3.8517,
       "step": 8950
     },
     {
       "epoch": 0.9703504043126685,
-      "grad_norm": 0.6123566031455994,
+      "grad_norm": 0.6032969951629639,
       "learning_rate": 0.0005423896384241769,
-      "loss": 3.8366,
+      "loss": 3.8389,
       "step": 9000
     },
     {
       "epoch": 0.9703504043126685,
-      "eval_accuracy": 0.3415915322893022,
-      "eval_loss": 3.775902509689331,
-      "eval_runtime": 184.7441,
-      "eval_samples_per_second": 97.492,
-      "eval_steps_per_second": 6.095,
+      "eval_accuracy": 0.341406605074242,
+      "eval_loss": 3.777738332748413,
+      "eval_runtime": 185.0795,
+      "eval_samples_per_second": 97.315,
+      "eval_steps_per_second": 6.084,
       "step": 9000
     },
     {
       "epoch": 0.9757412398921833,
-      "grad_norm": 0.604682445526123,
+      "grad_norm": 0.5933600664138794,
       "learning_rate": 0.0005420658391797086,
-      "loss": 3.8309,
+      "loss": 3.8341,
       "step": 9050
     },
     {
       "epoch": 0.9811320754716981,
-      "grad_norm": 0.5386356115341187,
+      "grad_norm": 0.5503290891647339,
       "learning_rate": 0.0005417420399352401,
-      "loss": 3.843,
+      "loss": 3.8429,
       "step": 9100
     },
     {
       "epoch": 0.9865229110512129,
-      "grad_norm": 0.6552569270133972,
+      "grad_norm": 0.667158305644989,
       "learning_rate": 0.0005414182406907717,
-      "loss": 3.8365,
+      "loss": 3.8374,
       "step": 9150
     },
     {
       "epoch": 0.9919137466307277,
-      "grad_norm": 0.6831485629081726,
+      "grad_norm": 0.6329292058944702,
       "learning_rate": 0.0005410944414463032,
-      "loss": 3.8525,
+      "loss": 3.8518,
       "step": 9200
     },
     {
       "epoch": 0.9973045822102425,
-      "grad_norm": 0.5169652700424194,
+      "grad_norm": 0.602115273475647,
       "learning_rate": 0.0005407706422018348,
-      "loss": 3.8489,
+      "loss": 3.8477,
       "step": 9250
     },
     {
       "epoch": 1.0026954177897573,
-      "grad_norm": 0.556510329246521,
+      "grad_norm": 0.5481619834899902,
       "learning_rate": 0.0005404468429573664,
-      "loss": 3.7966,
+      "loss": 3.7957,
       "step": 9300
     },
     {
       "epoch": 1.0080862533692723,
-      "grad_norm": 0.6710174083709717,
+      "grad_norm": 0.7251120805740356,
       "learning_rate": 0.000540123043712898,
-      "loss": 3.7679,
+      "loss": 3.7684,
       "step": 9350
     },
     {
       "epoch": 1.013477088948787,
-      "grad_norm": 0.6061473488807678,
+      "grad_norm": 0.6227707862854004,
       "learning_rate": 0.0005397992444684295,
-      "loss": 3.7491,
+      "loss": 3.7501,
       "step": 9400
     },
     {
       "epoch": 1.0188679245283019,
-      "grad_norm": 0.6734007000923157,
+      "grad_norm": 0.6860117316246033,
       "learning_rate": 0.0005394754452239611,
-      "loss": 3.7614,
+      "loss": 3.7643,
       "step": 9450
     },
     {
       "epoch": 1.0242587601078168,
-      "grad_norm": 0.5629554986953735,
+      "grad_norm": 0.5848320126533508,
       "learning_rate": 0.0005391516459794927,
-      "loss": 3.7738,
+      "loss": 3.773,
       "step": 9500
     },
     {
       "epoch": 1.0296495956873315,
-      "grad_norm": 0.5931057929992676,
+      "grad_norm": 0.5614584684371948,
       "learning_rate": 0.0005388278467350242,
-      "loss": 3.7805,
+      "loss": 3.7825,
       "step": 9550
     },
     {
       "epoch": 1.0350404312668464,
-      "grad_norm": 0.5350716710090637,
+      "grad_norm": 0.5323917865753174,
       "learning_rate": 0.0005385040474905557,
-      "loss": 3.7747,
+      "loss": 3.7755,
       "step": 9600
     },
     {
       "epoch": 1.0404312668463611,
-      "grad_norm": 0.570182204246521,
+      "grad_norm": 0.5698872804641724,
       "learning_rate": 0.0005381802482460874,
-      "loss": 3.785,
+      "loss": 3.7865,
       "step": 9650
     },
     {
       "epoch": 1.045822102425876,
-      "grad_norm": 0.5359431505203247,
+      "grad_norm": 0.5502426028251648,
       "learning_rate": 0.000537856449001619,
-      "loss": 3.7839,
+      "loss": 3.7843,
       "step": 9700
     },
     {
       "epoch": 1.0512129380053907,
-      "grad_norm": 0.590753436088562,
+      "grad_norm": 0.6106315851211548,
       "learning_rate": 0.0005375326497571505,
-      "loss": 3.7542,
+      "loss": 3.7552,
       "step": 9750
     },
     {
       "epoch": 1.0566037735849056,
-      "grad_norm": 0.6063977479934692,
+      "grad_norm": 0.6247063279151917,
       "learning_rate": 0.000537208850512682,
-      "loss": 3.7645,
+      "loss": 3.7613,
       "step": 9800
     },
     {
       "epoch": 1.0619946091644206,
-      "grad_norm": 0.6077662706375122,
+      "grad_norm": 0.5800876617431641,
       "learning_rate": 0.0005368850512682137,
-      "loss": 3.7599,
+      "loss": 3.7608,
       "step": 9850
     },
     {
       "epoch": 1.0673854447439353,
-      "grad_norm": 0.52806156873703,
+      "grad_norm": 0.5341230630874634,
       "learning_rate": 0.0005365612520237453,
-      "loss": 3.7725,
+      "loss": 3.7733,
       "step": 9900
     },
     {
       "epoch": 1.0727762803234502,
-      "grad_norm": 0.5774953365325928,
+      "grad_norm": 0.5820408463478088,
       "learning_rate": 0.0005362374527792768,
-      "loss": 3.7737,
+      "loss": 3.7746,
       "step": 9950
     },
     {
       "epoch": 1.0781671159029649,
-      "grad_norm": 0.5876293182373047,
+      "grad_norm": 0.5988202691078186,
       "learning_rate": 0.0005359136535348084,
-      "loss": 3.7721,
+      "loss": 3.7747,
       "step": 10000
     },
     {
       "epoch": 1.0781671159029649,
-      "eval_accuracy": 0.3451080830034179,
-      "eval_loss": 3.7522928714752197,
-      "eval_runtime": 184.8196,
-      "eval_samples_per_second": 97.452,
-      "eval_steps_per_second": 6.092,
+      "eval_accuracy": 0.34501203385058987,
+      "eval_loss": 3.7526464462280273,
+      "eval_runtime": 184.85,
+      "eval_samples_per_second": 97.436,
+      "eval_steps_per_second": 6.091,
       "step": 10000
     },
     {
       "epoch": 1.0835579514824798,
-      "grad_norm": 0.5495566129684448,
+      "grad_norm": 0.5564333200454712,
       "learning_rate": 0.0005355898542903399,
-      "loss": 3.7495,
+      "loss": 3.7497,
       "step": 10050
     },
     {
       "epoch": 1.0889487870619945,
-      "grad_norm": 0.5702345967292786,
+      "grad_norm": 0.6517497301101685,
       "learning_rate": 0.0005352660550458716,
-      "loss": 3.7524,
+      "loss": 3.7541,
       "step": 10100
     },
     {
       "epoch": 1.0943396226415094,
-      "grad_norm": 0.6201688051223755,
+      "grad_norm": 0.6653234958648682,
       "learning_rate": 0.000534942255801403,
       "loss": 3.7444,
       "step": 10150
     },
     {
       "epoch": 1.0997304582210243,
-      "grad_norm": 0.6020630598068237,
+      "grad_norm": 0.6291157007217407,
       "learning_rate": 0.0005346184565569347,
-      "loss": 3.7593,
+      "loss": 3.7606,
       "step": 10200
     },
     {
       "epoch": 1.105121293800539,
-      "grad_norm": 0.5759705901145935,
+      "grad_norm": 0.5519572496414185,
       "learning_rate": 0.0005342946573124662,
-      "loss": 3.7785,
+      "loss": 3.7783,
       "step": 10250
     },
     {
       "epoch": 1.110512129380054,
-      "grad_norm": 0.6400898098945618,
+      "grad_norm": 0.6604307889938354,
       "learning_rate": 0.0005339708580679978,
-      "loss": 3.7777,
+      "loss": 3.7808,
       "step": 10300
     },
     {
       "epoch": 1.1159029649595686,
-      "grad_norm": 0.587372362613678,
+      "grad_norm": 0.5644993185997009,
       "learning_rate": 0.0005336470588235293,
-      "loss": 3.7561,
+      "loss": 3.755,
       "step": 10350
     },
     {
       "epoch": 1.1212938005390836,
-      "grad_norm": 0.5996315479278564,
+      "grad_norm": 0.6334912776947021,
       "learning_rate": 0.000533323259579061,
-      "loss": 3.7586,
+      "loss": 3.7597,
       "step": 10400
     },
     {
       "epoch": 1.1266846361185983,
-      "grad_norm": 0.5710271000862122,
+      "grad_norm": 0.5180814266204834,
       "learning_rate": 0.0005329994603345925,
-      "loss": 3.7552,
+      "loss": 3.7527,
       "step": 10450
     },
     {
       "epoch": 1.1320754716981132,
-      "grad_norm": 0.5642911195755005,
+      "grad_norm": 0.6314664483070374,
       "learning_rate": 0.0005326756610901241,
-      "loss": 3.7533,
+      "loss": 3.7526,
       "step": 10500
     },
     {
       "epoch": 1.137466307277628,
-      "grad_norm": 0.5622043609619141,
+      "grad_norm": 0.5654125213623047,
       "learning_rate": 0.0005323518618456556,
-      "loss": 3.7532,
+      "loss": 3.7534,
       "step": 10550
     },
     {
       "epoch": 1.1428571428571428,
-      "grad_norm": 0.5839791893959045,
+      "grad_norm": 0.6075195670127869,
       "learning_rate": 0.0005320280626011872,
-      "loss": 3.7666,
+      "loss": 3.7658,
       "step": 10600
     },
     {
       "epoch": 1.1482479784366577,
-      "grad_norm": 0.5701001882553101,
+      "grad_norm": 0.6319811940193176,
       "learning_rate": 0.0005317042633567188,
-      "loss": 3.7541,
+      "loss": 3.7513,
       "step": 10650
     },
     {
       "epoch": 1.1536388140161726,
-      "grad_norm": 0.6332500576972961,
+      "grad_norm": 0.6518101096153259,
       "learning_rate": 0.0005313804641122504,
-      "loss": 3.7563,
+      "loss": 3.7577,
       "step": 10700
     },
     {
       "epoch": 1.1590296495956873,
-      "grad_norm": 0.5746287703514099,
+      "grad_norm": 0.5641539692878723,
       "learning_rate": 0.0005310566648677819,
-      "loss": 3.752,
+      "loss": 3.7533,
       "step": 10750
     },
     {
       "epoch": 1.1644204851752022,
-      "grad_norm": 0.5689650774002075,
+      "grad_norm": 0.6202017068862915,
       "learning_rate": 0.0005307328656233135,
-      "loss": 3.7422,
+      "loss": 3.742,
       "step": 10800
     },
     {
       "epoch": 1.169811320754717,
-      "grad_norm": 0.5871029496192932,
+      "grad_norm": 0.5958346724510193,
       "learning_rate": 0.000530409066378845,
-      "loss": 3.7375,
+      "loss": 3.7384,
       "step": 10850
     },
     {
       "epoch": 1.1752021563342319,
-      "grad_norm": 0.7632201313972473,
+      "grad_norm": 1.0744181871414185,
       "learning_rate": 0.0005300852671343766,
-      "loss": 3.7499,
+      "loss": 3.7539,
       "step": 10900
     },
     {
       "epoch": 1.1805929919137466,
-      "grad_norm": 0.7118698358535767,
+      "grad_norm": 0.6128881573677063,
       "learning_rate": 0.0005297614678899081,
-      "loss": 3.7436,
+      "loss": 3.7442,
       "step": 10950
     },
     {
       "epoch": 1.1859838274932615,
-      "grad_norm": 0.5203385353088379,
+      "grad_norm": 0.5362077951431274,
       "learning_rate": 0.0005294376686454398,
-      "loss": 3.733,
+      "loss": 3.7347,
       "step": 11000
     },
     {
       "epoch": 1.1859838274932615,
-      "eval_accuracy": 0.3484138471151735,
-      "eval_loss": 3.71924090385437,
-      "eval_runtime": 185.1007,
-      "eval_samples_per_second": 97.304,
-      "eval_steps_per_second": 6.083,
+      "eval_accuracy": 0.34794022917719014,
+      "eval_loss": 3.718404769897461,
+      "eval_runtime": 184.5725,
+      "eval_samples_per_second": 97.582,
+      "eval_steps_per_second": 6.101,
       "step": 11000
     },
     {
       "epoch": 1.1913746630727764,
-      "grad_norm": 0.5789610743522644,
+      "grad_norm": 0.5539353489875793,
       "learning_rate": 0.0005291138694009714,
-      "loss": 3.7447,
+      "loss": 3.7456,
       "step": 11050
     },
     {
       "epoch": 1.196765498652291,
-      "grad_norm": 0.6329853534698486,
+      "grad_norm": 0.6573549509048462,
       "learning_rate": 0.0005287900701565029,
-      "loss": 3.7378,
+      "loss": 3.7375,
       "step": 11100
     },
     {
       "epoch": 1.202156334231806,
-      "grad_norm": 0.6139509081840515,
+      "grad_norm": 0.5665242075920105,
       "learning_rate": 0.0005284662709120345,
-      "loss": 3.7366,
+      "loss": 3.7363,
       "step": 11150
     },
     {
       "epoch": 1.2075471698113207,
-      "grad_norm": 0.7028432488441467,
+      "grad_norm": 0.6505994200706482,
       "learning_rate": 0.0005281424716675661,
-      "loss": 3.7382,
+      "loss": 3.7368,
       "step": 11200
     },
     {
       "epoch": 1.2129380053908356,
-      "grad_norm": 0.6056612730026245,
+      "grad_norm": 0.586138904094696,
       "learning_rate": 0.0005278186724230977,
-      "loss": 3.7298,
+      "loss": 3.7303,
       "step": 11250
     },
     {
       "epoch": 1.2183288409703503,
-      "grad_norm": 0.5431009531021118,
+      "grad_norm": 0.5795910954475403,
       "learning_rate": 0.0005274948731786292,
-      "loss": 3.7267,
+      "loss": 3.7274,
       "step": 11300
     },
     {
       "epoch": 1.2237196765498652,
-      "grad_norm": 0.6328643560409546,
+      "grad_norm": 0.6727373003959656,
       "learning_rate": 0.0005271710739341608,
-      "loss": 3.7457,
+      "loss": 3.7443,
       "step": 11350
     },
     {
       "epoch": 1.2291105121293802,
-      "grad_norm": 0.5836541056632996,
+      "grad_norm": 0.5898151397705078,
       "learning_rate": 0.0005268472746896923,
-      "loss": 3.7387,
+      "loss": 3.7391,
       "step": 11400
     },
     {
       "epoch": 1.2345013477088949,
-      "grad_norm": 0.6509481072425842,
+      "grad_norm": 0.7265191674232483,
       "learning_rate": 0.000526523475445224,
-      "loss": 3.7302,
+      "loss": 3.7297,
       "step": 11450
     },
     {
       "epoch": 1.2398921832884098,
-      "grad_norm": 0.5459529757499695,
+      "grad_norm": 0.5766255259513855,
       "learning_rate": 0.0005261996762007554,
       "loss": 3.7167,
       "step": 11500
     },
     {
       "epoch": 1.2452830188679245,
-      "grad_norm": 0.575712263584137,
+      "grad_norm": 0.5683987140655518,
       "learning_rate": 0.0005258758769562871,
-      "loss": 3.738,
+      "loss": 3.7372,
       "step": 11550
     },
     {
       "epoch": 1.2506738544474394,
-      "grad_norm": 0.6626757383346558,
+      "grad_norm": 0.611061692237854,
       "learning_rate": 0.0005255520777118186,
-      "loss": 3.7379,
+      "loss": 3.7395,
       "step": 11600
     },
     {
       "epoch": 1.256064690026954,
-      "grad_norm": 0.6254045963287354,
+      "grad_norm": 0.5668408274650574,
       "learning_rate": 0.0005252282784673502,
-      "loss": 3.7293,
+      "loss": 3.7299,
       "step": 11650
     },
     {
       "epoch": 1.261455525606469,
-      "grad_norm": 0.5604497194290161,
+      "grad_norm": 0.5914852023124695,
       "learning_rate": 0.0005249044792228817,
-      "loss": 3.7315,
+      "loss": 3.7326,
       "step": 11700
     },
     {
       "epoch": 1.266846361185984,
-      "grad_norm": 0.5226693153381348,
+      "grad_norm": 0.5146292448043823,
       "learning_rate": 0.0005245806799784133,
-      "loss": 3.7267,
+      "loss": 3.7275,
       "step": 11750
     },
     {
       "epoch": 1.2722371967654986,
-      "grad_norm": 0.5691946744918823,
+      "grad_norm": 0.5495054721832275,
       "learning_rate": 0.0005242568807339449,
-      "loss": 3.7223,
+      "loss": 3.7229,
       "step": 11800
     },
     {
       "epoch": 1.2776280323450135,
-      "grad_norm": 0.5967990159988403,
+      "grad_norm": 0.6662834286689758,
       "learning_rate": 0.0005239330814894765,
-      "loss": 3.718,
+      "loss": 3.719,
       "step": 11850
     },
     {
       "epoch": 1.2830188679245282,
-      "grad_norm": 0.5392739772796631,
+      "grad_norm": 0.5426035523414612,
       "learning_rate": 0.000523609282245008,
-      "loss": 3.7396,
+      "loss": 3.738,
       "step": 11900
     },
     {
       "epoch": 1.2884097035040432,
-      "grad_norm": 0.6016465425491333,
+      "grad_norm": 0.638138473033905,
       "learning_rate": 0.0005232854830005396,
-      "loss": 3.7361,
+      "loss": 3.7367,
       "step": 11950
     },
     {
       "epoch": 1.2938005390835579,
-      "grad_norm": 0.5855374336242676,
+      "grad_norm": 0.5627890229225159,
       "learning_rate": 0.0005229616837560712,
-      "loss": 3.7312,
+      "loss": 3.7311,
       "step": 12000
     },
     {
       "epoch": 1.2938005390835579,
-      "eval_accuracy": 0.3501790219306075,
-      "eval_loss": 3.6982507705688477,
-      "eval_runtime": 185.0644,
-      "eval_samples_per_second": 97.323,
-      "eval_steps_per_second": 6.084,
+      "eval_accuracy": 0.3499498729902068,
+      "eval_loss": 3.698021411895752,
+      "eval_runtime": 185.1248,
+      "eval_samples_per_second": 97.291,
+      "eval_steps_per_second": 6.082,
       "step": 12000
     },
     {
       "epoch": 1.2991913746630728,
-      "grad_norm": 0.5653311610221863,
+      "grad_norm": 0.5502892136573792,
       "learning_rate": 0.0005226378845116028,
-      "loss": 3.7125,
+      "loss": 3.7115,
       "step": 12050
     },
     {
       "epoch": 1.3045822102425877,
-      "grad_norm": 0.5788018703460693,
+      "grad_norm": 0.5573764443397522,
       "learning_rate": 0.0005223140852671344,
-      "loss": 3.7259,
+      "loss": 3.7253,
       "step": 12100
     },
     {
       "epoch": 1.3099730458221024,
-      "grad_norm": 0.6671530604362488,
+      "grad_norm": 0.6834853887557983,
       "learning_rate": 0.0005219902860226659,
-      "loss": 3.7371,
+      "loss": 3.7359,
       "step": 12150
     },
     {
       "epoch": 1.3153638814016173,
-      "grad_norm": 0.5864307880401611,
+      "grad_norm": 0.5750380158424377,
       "learning_rate": 0.0005216664867781975,
-      "loss": 3.7351,
+      "loss": 3.7342,
       "step": 12200
     },
     {
       "epoch": 1.320754716981132,
-      "grad_norm": 0.655349850654602,
+      "grad_norm": 0.6798782348632812,
       "learning_rate": 0.000521342687533729,
-      "loss": 3.7162,
+      "loss": 3.715,
       "step": 12250
     },
     {
       "epoch": 1.326145552560647,
-      "grad_norm": 0.5737414956092834,
+      "grad_norm": 0.6055248379707336,
       "learning_rate": 0.0005210188882892606,
-      "loss": 3.7245,
+      "loss": 3.727,
       "step": 12300
     },
     {
       "epoch": 1.3315363881401616,
-      "grad_norm": 0.5403628945350647,
+      "grad_norm": 0.5729120373725891,
       "learning_rate": 0.0005206950890447922,
-      "loss": 3.7137,
+      "loss": 3.7154,
       "step": 12350
     },
     {
       "epoch": 1.3369272237196765,
-      "grad_norm": 0.5257432460784912,
+      "grad_norm": 0.505923867225647,
       "learning_rate": 0.0005203712898003238,
       "loss": 3.73,
       "step": 12400
     },
     {
       "epoch": 1.3423180592991915,
-      "grad_norm": 0.6442195177078247,
+      "grad_norm": 0.6457091569900513,
       "learning_rate": 0.0005200474905558553,
-      "loss": 3.7193,
+      "loss": 3.7195,
       "step": 12450
     },
     {
       "epoch": 1.3477088948787062,
-      "grad_norm": 0.6109151244163513,
+      "grad_norm": 0.6898245215415955,
       "learning_rate": 0.0005197236913113869,
-      "loss": 3.722,
+      "loss": 3.7196,
       "step": 12500
     },
     {
       "epoch": 1.353099730458221,
-      "grad_norm": 0.5608176589012146,
+      "grad_norm": 0.5175744891166687,
       "learning_rate": 0.0005193998920669184,
-      "loss": 3.7286,
+      "loss": 3.728,
       "step": 12550
     },
     {
       "epoch": 1.3584905660377358,
-      "grad_norm": 0.5631012916564941,
+      "grad_norm": 0.6096599102020264,
       "learning_rate": 0.0005190760928224501,
-      "loss": 3.7126,
+      "loss": 3.7121,
       "step": 12600
     },
     {
       "epoch": 1.3638814016172507,
-      "grad_norm": 0.5685343146324158,
+      "grad_norm": 0.5449326634407043,
       "learning_rate": 0.0005187522935779816,
-      "loss": 3.7195,
+      "loss": 3.7192,
       "step": 12650
     },
     {
       "epoch": 1.3692722371967654,
-      "grad_norm": 0.5835712552070618,
+      "grad_norm": 0.6141244173049927,
       "learning_rate": 0.0005184284943335132,
-      "loss": 3.7342,
+      "loss": 3.7337,
       "step": 12700
     },
     {
       "epoch": 1.3746630727762803,
-      "grad_norm": 0.5368698239326477,
+      "grad_norm": 0.5343438386917114,
       "learning_rate": 0.0005181046950890447,
-      "loss": 3.7183,
+      "loss": 3.7162,
       "step": 12750
     },
     {
       "epoch": 1.3800539083557952,
-      "grad_norm": 0.5741342306137085,
+      "grad_norm": 0.5652272701263428,
       "learning_rate": 0.0005177808958445764,
-      "loss": 3.7084,
+      "loss": 3.7075,
       "step": 12800
     },
     {
       "epoch": 1.38544474393531,
-      "grad_norm": 0.6180281043052673,
+      "grad_norm": 0.5998131036758423,
       "learning_rate": 0.0005174570966001078,
-      "loss": 3.7058,
+      "loss": 3.7052,
       "step": 12850
     },
     {
       "epoch": 1.3908355795148248,
-      "grad_norm": 0.5740448832511902,
+      "grad_norm": 0.5848682522773743,
       "learning_rate": 0.0005171332973556395,
-      "loss": 3.7109,
+      "loss": 3.7117,
       "step": 12900
     },
     {
       "epoch": 1.3962264150943398,
-      "grad_norm": 0.5574175119400024,
+      "grad_norm": 0.5479270219802856,
       "learning_rate": 0.000516809498111171,
-      "loss": 3.7104,
+      "loss": 3.7096,
       "step": 12950
     },
     {
       "epoch": 1.4016172506738545,
-      "grad_norm": 0.6154153347015381,
+      "grad_norm": 0.582422137260437,
       "learning_rate": 0.0005164856988667026,
-      "loss": 3.7055,
+      "loss": 3.705,
       "step": 13000
     },
     {
       "epoch": 1.4016172506738545,
-      "eval_accuracy": 0.35201014904894495,
-      "eval_loss": 3.6755025386810303,
-      "eval_runtime": 184.7138,
-      "eval_samples_per_second": 97.508,
-      "eval_steps_per_second": 6.096,
+      "eval_accuracy": 0.35188910973191967,
+      "eval_loss": 3.6742594242095947,
+      "eval_runtime": 184.8492,
+      "eval_samples_per_second": 97.436,
+      "eval_steps_per_second": 6.091,
       "step": 13000
     },
     {
       "epoch": 1.4070080862533692,
-      "grad_norm": 0.6463391780853271,
+      "grad_norm": 0.5786948800086975,
       "learning_rate": 0.0005161618996222341,
-      "loss": 3.7168,
+      "loss": 3.7174,
       "step": 13050
     },
     {
       "epoch": 1.412398921832884,
-      "grad_norm": 0.5860043168067932,
+      "grad_norm": 0.5718762278556824,
       "learning_rate": 0.0005158381003777657,
-      "loss": 3.7283,
+      "loss": 3.7291,
       "step": 13100
     },
     {
       "epoch": 1.417789757412399,
-      "grad_norm": 0.6286531686782837,
+      "grad_norm": 0.5814722776412964,
       "learning_rate": 0.0005155143011332973,
-      "loss": 3.7071,
+      "loss": 3.7076,
       "step": 13150
     },
     {
       "epoch": 1.4231805929919137,
-      "grad_norm": 0.545039713382721,
+      "grad_norm": 0.5653195977210999,
       "learning_rate": 0.0005151905018888289,
-      "loss": 3.7097,
+      "loss": 3.7109,
       "step": 13200
     },
     {
       "epoch": 1.4285714285714286,
-      "grad_norm": 0.5539329648017883,
-      "learning_rate": 0.0005148667026443604,
-      "loss": 3.7058,
+      "grad_norm": 0.5738582015037537,
+      "learning_rate": 0.0005148731786292498,
+      "loss": 3.7059,
       "step": 13250
     },
     {
       "epoch": 1.4339622641509435,
-      "grad_norm": 0.6011884212493896,
-      "learning_rate": 0.000514542903399892,
-      "loss": 3.7048,
+      "grad_norm": 0.5981550812721252,
+      "learning_rate": 0.0005145493793847814,
+      "loss": 3.7036,
       "step": 13300
     },
     {
       "epoch": 1.4393530997304582,
-      "grad_norm": 0.5569776892662048,
-      "learning_rate": 0.0005142191041554237,
-      "loss": 3.6906,
+      "grad_norm": 0.5781781077384949,
+      "learning_rate": 0.000514225580140313,
+      "loss": 3.6913,
       "step": 13350
     },
     {
       "epoch": 1.444743935309973,
-      "grad_norm": 0.6196861863136292,
-      "learning_rate": 0.0005138953049109552,
-      "loss": 3.7005,
+      "grad_norm": 0.6027509570121765,
+      "learning_rate": 0.0005139017808958445,
+      "loss": 3.7001,
       "step": 13400
     },
     {
       "epoch": 1.4501347708894878,
-      "grad_norm": 0.6113570332527161,
-      "learning_rate": 0.0005135715056664868,
-      "loss": 3.7094,
+      "grad_norm": 0.6018847227096558,
+      "learning_rate": 0.0005135779816513762,
+      "loss": 3.7117,
       "step": 13450
     },
     {
       "epoch": 1.4555256064690028,
-      "grad_norm": 0.528312087059021,
-      "learning_rate": 0.0005132477064220183,
-      "loss": 3.6953,
+      "grad_norm": 0.5335890650749207,
+      "learning_rate": 0.0005132541824069076,
+      "loss": 3.6938,
       "step": 13500
     },
     {
       "epoch": 1.4609164420485174,
-      "grad_norm": 0.6123299598693848,
-      "learning_rate": 0.0005129239071775499,
-      "loss": 3.7059,
+      "grad_norm": 0.5770127177238464,
+      "learning_rate": 0.0005129303831624393,
+      "loss": 3.7058,
       "step": 13550
     },
     {
       "epoch": 1.4663072776280324,
-      "grad_norm": 0.5983057618141174,
-      "learning_rate": 0.0005126001079330814,
-      "loss": 3.7097,
+      "grad_norm": 0.5771490335464478,
+      "learning_rate": 0.0005126065839179708,
+      "loss": 3.7089,
       "step": 13600
     },
     {
       "epoch": 1.4716981132075473,
-      "grad_norm": 0.5747043490409851,
-      "learning_rate": 0.000512276308688613,
-      "loss": 3.6826,
+      "grad_norm": 0.620576024055481,
+      "learning_rate": 0.0005122827846735024,
+      "loss": 3.6832,
       "step": 13650
     },
     {
       "epoch": 1.477088948787062,
-      "grad_norm": 0.5804892182350159,
-      "learning_rate": 0.0005119525094441446,
-      "loss": 3.6802,
+      "grad_norm": 0.5802292227745056,
+      "learning_rate": 0.0005119589854290339,
+      "loss": 3.6798,
       "step": 13700
     },
     {
       "epoch": 1.482479784366577,
-      "grad_norm": 0.5395514369010925,
-      "learning_rate": 0.0005116287101996762,
-      "loss": 3.6981,
+      "grad_norm": 0.603459358215332,
+      "learning_rate": 0.0005116351861845655,
+      "loss": 3.6995,
       "step": 13750
     },
     {
       "epoch": 1.4878706199460916,
-      "grad_norm": 0.635943591594696,
-      "learning_rate": 0.0005113049109552077,
-      "loss": 3.7122,
+      "grad_norm": 0.5906022787094116,
+      "learning_rate": 0.0005113113869400971,
+      "loss": 3.7129,
       "step": 13800
     },
     {
       "epoch": 1.4932614555256065,
-      "grad_norm": 0.6453544497489929,
-      "learning_rate": 0.0005109811117107393,
-      "loss": 3.7088,
+      "grad_norm": 0.580629289150238,
+      "learning_rate": 0.0005109875876956287,
+      "loss": 3.707,
       "step": 13850
     },
     {
       "epoch": 1.4986522911051212,
-      "grad_norm": 0.5905641913414001,
-      "learning_rate": 0.0005106573124662708,
-      "loss": 3.7223,
+      "grad_norm": 0.6544400453567505,
+      "learning_rate": 0.0005106637884511602,
+      "loss": 3.7212,
       "step": 13900
     },
     {
       "epoch": 1.5040431266846361,
-      "grad_norm": 0.5349699854850769,
-      "learning_rate": 0.0005103335132218025,
-      "loss": 3.6964,
+      "grad_norm": 0.5429078936576843,
+      "learning_rate": 0.0005103399892066918,
+      "loss": 3.6957,
       "step": 13950
     },
     {
       "epoch": 1.509433962264151,
-      "grad_norm": 0.6119936108589172,
-      "learning_rate": 0.000510009713977334,
-      "loss": 3.697,
+      "grad_norm": 0.5718664526939392,
+      "learning_rate": 0.0005100161899622234,
+      "loss": 3.6961,
       "step": 14000
     },
     {
       "epoch": 1.509433962264151,
-      "eval_accuracy": 0.35438225889571073,
-      "eval_loss": 3.6525440216064453,
-      "eval_runtime": 184.8462,
-      "eval_samples_per_second": 97.438,
-      "eval_steps_per_second": 6.092,
+      "eval_accuracy": 0.35413159668352273,
+      "eval_loss": 3.6542694568634033,
+      "eval_runtime": 185.2098,
+      "eval_samples_per_second": 97.246,
+      "eval_steps_per_second": 6.08,
       "step": 14000
     },
     {
       "epoch": 1.5148247978436657,
-      "grad_norm": 0.549912691116333,
-      "learning_rate": 0.0005096859147328656,
-      "loss": 3.6853,
+      "grad_norm": 0.530343770980835,
+      "learning_rate": 0.000509692390717755,
+      "loss": 3.6869,
       "step": 14050
     },
     {
       "epoch": 1.5202156334231804,
-      "grad_norm": 0.6812787055969238,
+      "grad_norm": 0.6733381152153015,
       "learning_rate": 0.0005093685914732865,
-      "loss": 3.6843,
+      "loss": 3.6809,
       "step": 14100
     },
     {
       "epoch": 1.5256064690026954,
-      "grad_norm": 0.5409011840820312,
+      "grad_norm": 0.5676352977752686,
       "learning_rate": 0.0005090447922288181,
-      "loss": 3.6802,
+      "loss": 3.6794,
       "step": 14150
     },
     {
       "epoch": 1.5309973045822103,
-      "grad_norm": 0.5610489845275879,
+      "grad_norm": 0.5399664640426636,
       "learning_rate": 0.0005087209929843496,
-      "loss": 3.7142,
+      "loss": 3.7143,
       "step": 14200
     },
     {
       "epoch": 1.536388140161725,
-      "grad_norm": 0.5780990123748779,
+      "grad_norm": 0.6497036218643188,
       "learning_rate": 0.0005083971937398812,
-      "loss": 3.6901,
+      "loss": 3.6866,
       "step": 14250
     },
     {
       "epoch": 1.54177897574124,
-      "grad_norm": 0.5896368026733398,
+      "grad_norm": 0.6060945391654968,
       "learning_rate": 0.0005080733944954127,
-      "loss": 3.6923,
+      "loss": 3.6915,
       "step": 14300
     },
     {
       "epoch": 1.5471698113207548,
-      "grad_norm": 0.5635419487953186,
+      "grad_norm": 0.6297807693481445,
       "learning_rate": 0.0005077495952509444,
-      "loss": 3.6735,
+      "loss": 3.6741,
       "step": 14350
     },
     {
       "epoch": 1.5525606469002695,
-      "grad_norm": 0.5706263184547424,
+      "grad_norm": 0.5488743782043457,
       "learning_rate": 0.0005074257960064759,
-      "loss": 3.6872,
+      "loss": 3.6868,
       "step": 14400
     },
     {
       "epoch": 1.5579514824797842,
-      "grad_norm": 0.5970407724380493,
+      "grad_norm": 0.5748883485794067,
       "learning_rate": 0.0005071019967620075,
-      "loss": 3.6924,
+      "loss": 3.6918,
       "step": 14450
     },
     {
       "epoch": 1.5633423180592994,
-      "grad_norm": 0.5763431787490845,
+      "grad_norm": 0.5836804509162903,
       "learning_rate": 0.000506778197517539,
-      "loss": 3.6854,
+      "loss": 3.6834,
       "step": 14500
     },
     {
       "epoch": 1.568733153638814,
-      "grad_norm": 0.6334835290908813,
+      "grad_norm": 0.612476110458374,
       "learning_rate": 0.0005064543982730707,
-      "loss": 3.6955,
+      "loss": 3.6924,
       "step": 14550
     },
     {
       "epoch": 1.5741239892183287,
-      "grad_norm": 0.5830768942832947,
+      "grad_norm": 0.6098917126655579,
       "learning_rate": 0.0005061305990286023,
-      "loss": 3.6875,
+      "loss": 3.6865,
       "step": 14600
     },
     {
       "epoch": 1.5795148247978437,
-      "grad_norm": 0.6113629341125488,
+      "grad_norm": 0.6164373755455017,
       "learning_rate": 0.0005058067997841338,
-      "loss": 3.6957,
+      "loss": 3.6947,
       "step": 14650
     },
     {
       "epoch": 1.5849056603773586,
-      "grad_norm": 0.5635071992874146,
+      "grad_norm": 0.5995565056800842,
       "learning_rate": 0.0005054830005396654,
-      "loss": 3.6826,
+      "loss": 3.6821,
       "step": 14700
     },
     {
       "epoch": 1.5902964959568733,
-      "grad_norm": 0.6076672673225403,
+      "grad_norm": 0.5815383791923523,
       "learning_rate": 0.0005051592012951969,
-      "loss": 3.6742,
+      "loss": 3.6734,
       "step": 14750
     },
     {
       "epoch": 1.595687331536388,
-      "grad_norm": 0.6091296076774597,
+      "grad_norm": 0.5790833830833435,
       "learning_rate": 0.0005048354020507286,
-      "loss": 3.6943,
+      "loss": 3.6948,
       "step": 14800
     },
     {
       "epoch": 1.6010781671159031,
-      "grad_norm": 0.6177613139152527,
+      "grad_norm": 0.6062554717063904,
       "learning_rate": 0.00050451160280626,
-      "loss": 3.6943,
+      "loss": 3.6944,
       "step": 14850
     },
     {
       "epoch": 1.6064690026954178,
-      "grad_norm": 0.5547499060630798,
+      "grad_norm": 0.5852131247520447,
       "learning_rate": 0.0005041878035617917,
-      "loss": 3.6845,
+      "loss": 3.6853,
       "step": 14900
     },
     {
       "epoch": 1.6118598382749325,
-      "grad_norm": 0.6204666495323181,
+      "grad_norm": 0.5513139367103577,
       "learning_rate": 0.0005038640043173232,
-      "loss": 3.6796,
+      "loss": 3.6793,
       "step": 14950
     },
     {
       "epoch": 1.6172506738544474,
-      "grad_norm": 0.7229707837104797,
+      "grad_norm": 0.6422461867332458,
       "learning_rate": 0.0005035402050728548,
-      "loss": 3.6759,
+      "loss": 3.6739,
       "step": 15000
     },
     {
       "epoch": 1.6172506738544474,
-      "eval_accuracy": 0.35614971542178886,
-      "eval_loss": 3.6349523067474365,
-      "eval_runtime": 184.7752,
-      "eval_samples_per_second": 97.475,
-      "eval_steps_per_second": 6.094,
+      "eval_accuracy": 0.35602030983240185,
+      "eval_loss": 3.6368651390075684,
+      "eval_runtime": 184.8953,
+      "eval_samples_per_second": 97.412,
+      "eval_steps_per_second": 6.09,
       "step": 15000
     },
     {
       "epoch": 1.6226415094339623,
-      "grad_norm": 0.5129961967468262,
+      "grad_norm": 0.5301791429519653,
       "learning_rate": 0.0005032164058283863,
-      "loss": 3.6765,
+      "loss": 3.6771,
       "step": 15050
     },
     {
       "epoch": 1.628032345013477,
-      "grad_norm": 0.5776543617248535,
+      "grad_norm": 0.5744919180870056,
       "learning_rate": 0.0005028926065839179,
-      "loss": 3.6737,
+      "loss": 3.6726,
       "step": 15100
     },
     {
       "epoch": 1.633423180592992,
-      "grad_norm": 0.5395156145095825,
+      "grad_norm": 0.6005228161811829,
       "learning_rate": 0.0005025688073394495,
-      "loss": 3.6743,
+      "loss": 3.6734,
       "step": 15150
     },
     {
       "epoch": 1.6388140161725069,
-      "grad_norm": 0.566416323184967,
+      "grad_norm": 0.5289723873138428,
       "learning_rate": 0.0005022450080949811,
-      "loss": 3.6799,
+      "loss": 3.681,
       "step": 15200
     },
     {
       "epoch": 1.6442048517520216,
-      "grad_norm": 0.6514060497283936,
+      "grad_norm": 0.6079738140106201,
       "learning_rate": 0.0005019212088505126,
-      "loss": 3.6803,
+      "loss": 3.6818,
       "step": 15250
     },
     {
       "epoch": 1.6495956873315363,
-      "grad_norm": 0.5592887997627258,
+      "grad_norm": 0.5393943190574646,
       "learning_rate": 0.0005015974096060442,
-      "loss": 3.6829,
+      "loss": 3.6825,
       "step": 15300
     },
     {
       "epoch": 1.6549865229110512,
-      "grad_norm": 0.5593628287315369,
+      "grad_norm": 0.5830407738685608,
       "learning_rate": 0.0005012736103615758,
-      "loss": 3.6681,
+      "loss": 3.6678,
       "step": 15350
     },
     {
       "epoch": 1.6603773584905661,
-      "grad_norm": 0.5848199129104614,
+      "grad_norm": 0.6033615469932556,
       "learning_rate": 0.0005009498111171074,
-      "loss": 3.6728,
+      "loss": 3.6727,
       "step": 15400
     },
     {
       "epoch": 1.6657681940700808,
-      "grad_norm": 0.5796011090278625,
+      "grad_norm": 0.5500646233558655,
       "learning_rate": 0.0005006260118726389,
-      "loss": 3.6723,
+      "loss": 3.67,
       "step": 15450
     },
     {
       "epoch": 1.6711590296495957,
-      "grad_norm": 0.5999243855476379,
+      "grad_norm": 0.5812546610832214,
       "learning_rate": 0.0005003022126281705,
-      "loss": 3.6724,
+      "loss": 3.6717,
       "step": 15500
     },
     {
       "epoch": 1.6765498652291106,
-      "grad_norm": 0.6494645476341248,
+      "grad_norm": 0.6952954530715942,
       "learning_rate": 0.000499978413383702,
-      "loss": 3.667,
+      "loss": 3.666,
       "step": 15550
     },
     {
       "epoch": 1.6819407008086253,
-      "grad_norm": 0.592124879360199,
+      "grad_norm": 0.591118335723877,
       "learning_rate": 0.0004996546141392336,
-      "loss": 3.663,
+      "loss": 3.6629,
       "step": 15600
     },
     {
       "epoch": 1.68733153638814,
-      "grad_norm": 0.5345093607902527,
+      "grad_norm": 0.5555945038795471,
       "learning_rate": 0.0004993308148947651,
-      "loss": 3.6713,
+      "loss": 3.6708,
       "step": 15650
     },
     {
       "epoch": 1.692722371967655,
-      "grad_norm": 0.5050610303878784,
+      "grad_norm": 0.5302060842514038,
       "learning_rate": 0.0004990070156502968,
-      "loss": 3.6604,
+      "loss": 3.6621,
       "step": 15700
     },
     {
       "epoch": 1.6981132075471699,
-      "grad_norm": 0.559531033039093,
+      "grad_norm": 0.5419127941131592,
       "learning_rate": 0.0004986832164058284,
-      "loss": 3.6585,
+      "loss": 3.6563,
       "step": 15750
     },
     {
       "epoch": 1.7035040431266846,
-      "grad_norm": 0.5738038420677185,
+      "grad_norm": 0.6328555941581726,
       "learning_rate": 0.0004983594171613599,
-      "loss": 3.6718,
+      "loss": 3.6704,
       "step": 15800
     },
     {
       "epoch": 1.7088948787061995,
-      "grad_norm": 0.573388934135437,
+      "grad_norm": 0.5873494744300842,
       "learning_rate": 0.0004980356179168915,
-      "loss": 3.6761,
+      "loss": 3.6774,
       "step": 15850
     },
     {
       "epoch": 1.7142857142857144,
-      "grad_norm": 0.5734029412269592,
+      "grad_norm": 0.5820769667625427,
       "learning_rate": 0.000497711818672423,
-      "loss": 3.6985,
+      "loss": 3.6993,
       "step": 15900
     },
     {
       "epoch": 1.719676549865229,
-      "grad_norm": 0.5339508652687073,
+      "grad_norm": 0.5341653227806091,
       "learning_rate": 0.0004973880194279547,
-      "loss": 3.6647,
+      "loss": 3.6659,
       "step": 15950
     },
     {
       "epoch": 1.7250673854447438,
-      "grad_norm": 0.58454430103302,
+      "grad_norm": 0.5919328927993774,
       "learning_rate": 0.0004970642201834862,
-      "loss": 3.6742,
+      "loss": 3.6728,
       "step": 16000
     },
     {
       "epoch": 1.7250673854447438,
-      "eval_accuracy": 0.3578629541568476,
-      "eval_loss": 3.6172311305999756,
-      "eval_runtime": 185.271,
-      "eval_samples_per_second": 97.214,
-      "eval_steps_per_second": 6.078,
+      "eval_accuracy": 0.357589800796882,
+      "eval_loss": 3.6153509616851807,
+      "eval_runtime": 184.7079,
+      "eval_samples_per_second": 97.511,
+      "eval_steps_per_second": 6.096,
       "step": 16000
     },
     {
       "epoch": 1.7304582210242587,
-      "grad_norm": 0.5619959831237793,
+      "grad_norm": 0.5269375443458557,
       "learning_rate": 0.0004967404209390178,
-      "loss": 3.6425,
+      "loss": 3.641,
       "step": 16050
     },
     {
       "epoch": 1.7358490566037736,
-      "grad_norm": 0.551400899887085,
-      "learning_rate": 0.0004964230976794387,
-      "loss": 3.6503,
+      "grad_norm": 0.5364742279052734,
+      "learning_rate": 0.0004964166216945493,
+      "loss": 3.6507,
       "step": 16100
     },
     {
       "epoch": 1.7412398921832883,
-      "grad_norm": 0.5677595734596252,
-      "learning_rate": 0.0004960992984349703,
-      "loss": 3.6675,
+      "grad_norm": 0.5592376589775085,
+      "learning_rate": 0.000496092822450081,
+      "loss": 3.6648,
       "step": 16150
     },
     {
       "epoch": 1.7466307277628033,
-      "grad_norm": 0.587281346321106,
-      "learning_rate": 0.0004957754991905018,
-      "loss": 3.6526,
+      "grad_norm": 0.6180751323699951,
+      "learning_rate": 0.0004957690232056125,
+      "loss": 3.6529,
       "step": 16200
     },
     {
       "epoch": 1.7520215633423182,
-      "grad_norm": 0.536168098449707,
-      "learning_rate": 0.0004954516999460334,
-      "loss": 3.6623,
+      "grad_norm": 0.5617481470108032,
+      "learning_rate": 0.0004954452239611441,
+      "loss": 3.6616,
       "step": 16250
     },
     {
       "epoch": 1.7574123989218329,
-      "grad_norm": 0.5161067843437195,
-      "learning_rate": 0.0004951279007015649,
-      "loss": 3.6526,
+      "grad_norm": 0.5292233824729919,
+      "learning_rate": 0.0004951214247166756,
+      "loss": 3.652,
       "step": 16300
     },
     {
       "epoch": 1.7628032345013476,
-      "grad_norm": 0.5492506623268127,
-      "learning_rate": 0.0004948041014570966,
-      "loss": 3.6495,
+      "grad_norm": 0.5463518500328064,
+      "learning_rate": 0.0004947976254722072,
+      "loss": 3.6471,
       "step": 16350
     },
     {
       "epoch": 1.7681940700808625,
-      "grad_norm": 0.5799858570098877,
-      "learning_rate": 0.0004944803022126281,
-      "loss": 3.6613,
+      "grad_norm": 0.5895617008209229,
+      "learning_rate": 0.0004944738262277387,
+      "loss": 3.6603,
       "step": 16400
     },
     {
       "epoch": 1.7735849056603774,
-      "grad_norm": 0.538158655166626,
-      "learning_rate": 0.0004941565029681597,
-      "loss": 3.6595,
+      "grad_norm": 0.541398286819458,
+      "learning_rate": 0.0004941500269832703,
+      "loss": 3.6592,
       "step": 16450
     },
     {
       "epoch": 1.778975741239892,
-      "grad_norm": 0.5593761205673218,
-      "learning_rate": 0.0004938327037236912,
+      "grad_norm": 0.5419638752937317,
+      "learning_rate": 0.0004938262277388019,
       "loss": 3.6481,
       "step": 16500
     },
     {
       "epoch": 1.784366576819407,
-      "grad_norm": 0.6031091809272766,
-      "learning_rate": 0.0004935089044792229,
-      "loss": 3.6445,
+      "grad_norm": 0.5745832324028015,
+      "learning_rate": 0.0004935024284943335,
+      "loss": 3.6433,
       "step": 16550
     },
     {
       "epoch": 1.789757412398922,
-      "grad_norm": 0.5604456067085266,
-      "learning_rate": 0.0004931851052347544,
-      "loss": 3.6512,
+      "grad_norm": 0.5785013437271118,
+      "learning_rate": 0.000493178629249865,
+      "loss": 3.6502,
       "step": 16600
     },
     {
       "epoch": 1.7951482479784366,
-      "grad_norm": 0.5797213315963745,
-      "learning_rate": 0.000492861305990286,
-      "loss": 3.6587,
+      "grad_norm": 0.5288052558898926,
+      "learning_rate": 0.0004928548300053966,
+      "loss": 3.6573,
       "step": 16650
     },
     {
       "epoch": 1.8005390835579513,
-      "grad_norm": 0.5687010884284973,
-      "learning_rate": 0.0004925375067458175,
-      "loss": 3.6505,
+      "grad_norm": 0.5558264255523682,
+      "learning_rate": 0.0004925310307609282,
+      "loss": 3.6499,
       "step": 16700
     },
     {
       "epoch": 1.8059299191374663,
-      "grad_norm": 0.5328428745269775,
-      "learning_rate": 0.0004922137075013491,
-      "loss": 3.6433,
+      "grad_norm": 0.5273551344871521,
+      "learning_rate": 0.0004922072315164598,
+      "loss": 3.6437,
       "step": 16750
     },
     {
       "epoch": 1.8113207547169812,
-      "grad_norm": 0.5607346892356873,
-      "learning_rate": 0.0004918899082568807,
-      "loss": 3.6503,
+      "grad_norm": 0.5901119709014893,
+      "learning_rate": 0.0004918834322719913,
+      "loss": 3.6486,
       "step": 16800
     },
     {
       "epoch": 1.8167115902964959,
-      "grad_norm": 0.5864443182945251,
-      "learning_rate": 0.0004915661090124123,
-      "loss": 3.6385,
+      "grad_norm": 0.5500622987747192,
+      "learning_rate": 0.0004915596330275229,
+      "loss": 3.6377,
       "step": 16850
     },
     {
       "epoch": 1.8221024258760108,
-      "grad_norm": 0.6052229404449463,
-      "learning_rate": 0.0004912423097679439,
-      "loss": 3.6449,
+      "grad_norm": 0.5930582284927368,
+      "learning_rate": 0.0004912358337830544,
+      "loss": 3.6429,
       "step": 16900
     },
     {
       "epoch": 1.8274932614555257,
-      "grad_norm": 0.542421281337738,
-      "learning_rate": 0.0004909185105234754,
-      "loss": 3.6512,
+      "grad_norm": 0.5709859728813171,
+      "learning_rate": 0.000490912034538586,
+      "loss": 3.6515,
       "step": 16950
     },
     {
       "epoch": 1.8328840970350404,
-      "grad_norm": 0.5311638712882996,
-      "learning_rate": 0.000490594711279007,
-      "loss": 3.6581,
+      "grad_norm": 0.5371534824371338,
+      "learning_rate": 0.0004905882352941175,
+      "loss": 3.6584,
       "step": 17000
     },
     {
       "epoch": 1.8328840970350404,
-      "eval_accuracy": 0.359163529223987,
-      "eval_loss": 3.602339506149292,
-      "eval_runtime": 184.8776,
-      "eval_samples_per_second": 97.421,
-      "eval_steps_per_second": 6.091,
+      "eval_accuracy": 0.35917635026474903,
+      "eval_loss": 3.601609230041504,
+      "eval_runtime": 185.0779,
+      "eval_samples_per_second": 97.316,
+      "eval_steps_per_second": 6.084,
       "step": 17000
     },
     {
       "epoch": 1.838274932614555,
-      "grad_norm": 0.5611234307289124,
-      "learning_rate": 0.0004902709120345385,
-      "loss": 3.6485,
+      "grad_norm": 0.580851137638092,
+      "learning_rate": 0.0004902644360496492,
+      "loss": 3.651,
       "step": 17050
     },
     {
       "epoch": 1.8436657681940702,
-      "grad_norm": 0.5618613362312317,
-      "learning_rate": 0.00048994711279007,
-      "loss": 3.6394,
+      "grad_norm": 0.5615604519844055,
+      "learning_rate": 0.0004899406368051808,
+      "loss": 3.6398,
       "step": 17100
     },
     {
       "epoch": 1.849056603773585,
-      "grad_norm": 0.6035332679748535,
-      "learning_rate": 0.0004896233135456017,
-      "loss": 3.6409,
+      "grad_norm": 0.6272096037864685,
+      "learning_rate": 0.0004896168375607123,
+      "loss": 3.6379,
       "step": 17150
     },
     {
       "epoch": 1.8544474393530996,
-      "grad_norm": 0.6086268424987793,
-      "learning_rate": 0.0004892995143011333,
-      "loss": 3.6381,
+      "grad_norm": 0.5532569289207458,
+      "learning_rate": 0.0004892930383162439,
+      "loss": 3.6385,
       "step": 17200
     },
     {
       "epoch": 1.8598382749326146,
-      "grad_norm": 0.5861483812332153,
+      "grad_norm": 0.5877029299736023,
       "learning_rate": 0.0004889757150566648,
-      "loss": 3.6353,
+      "loss": 3.6333,
       "step": 17250
     },
     {
       "epoch": 1.8652291105121295,
-      "grad_norm": 0.5193630456924438,
+      "grad_norm": 0.5302947163581848,
       "learning_rate": 0.0004886519158121964,
-      "loss": 3.6603,
+      "loss": 3.6591,
       "step": 17300
     },
     {
       "epoch": 1.8706199460916442,
-      "grad_norm": 0.5715697407722473,
+      "grad_norm": 0.5727546811103821,
       "learning_rate": 0.000488328116567728,
-      "loss": 3.6605,
+      "loss": 3.6621,
       "step": 17350
     },
     {
       "epoch": 1.8760107816711589,
-      "grad_norm": 0.5773372054100037,
+      "grad_norm": 0.6038647890090942,
       "learning_rate": 0.0004880043173232595,
-      "loss": 3.6424,
+      "loss": 3.6402,
       "step": 17400
     },
     {
       "epoch": 1.881401617250674,
-      "grad_norm": 0.583741307258606,
+      "grad_norm": 0.554803192615509,
       "learning_rate": 0.0004876805180787911,
-      "loss": 3.6527,
+      "loss": 3.6514,
       "step": 17450
     },
     {
       "epoch": 1.8867924528301887,
-      "grad_norm": 0.5900930762290955,
+      "grad_norm": 0.5694112181663513,
       "learning_rate": 0.0004873567188343227,
-      "loss": 3.6337,
+      "loss": 3.6342,
       "step": 17500
     },
     {
       "epoch": 1.8921832884097034,
-      "grad_norm": 0.593124508857727,
+      "grad_norm": 0.5965739488601685,
       "learning_rate": 0.0004870329195898542,
-      "loss": 3.6494,
+      "loss": 3.6502,
       "step": 17550
     },
     {
       "epoch": 1.8975741239892183,
-      "grad_norm": 0.5326477885246277,
+      "grad_norm": 0.5624055862426758,
       "learning_rate": 0.00048670912034538583,
-      "loss": 3.6383,
+      "loss": 3.6382,
       "step": 17600
     },
     {
       "epoch": 1.9029649595687332,
-      "grad_norm": 0.5563052296638489,
+      "grad_norm": 0.5538950562477112,
       "learning_rate": 0.0004863853211009174,
-      "loss": 3.6343,
+      "loss": 3.6351,
       "step": 17650
     },
     {
       "epoch": 1.908355795148248,
-      "grad_norm": 0.5851816534996033,
+      "grad_norm": 0.5740907788276672,
       "learning_rate": 0.000486061521856449,
-      "loss": 3.6484,
+      "loss": 3.6466,
       "step": 17700
     },
     {
       "epoch": 1.9137466307277629,
-      "grad_norm": 0.538767397403717,
+      "grad_norm": 0.5794373154640198,
       "learning_rate": 0.00048573772261198054,
-      "loss": 3.6369,
+      "loss": 3.6373,
       "step": 17750
     },
     {
       "epoch": 1.9191374663072778,
-      "grad_norm": 0.5974882245063782,
+      "grad_norm": 0.5746911764144897,
       "learning_rate": 0.00048541392336751214,
-      "loss": 3.6523,
+      "loss": 3.652,
       "step": 17800
     },
     {
       "epoch": 1.9245283018867925,
-      "grad_norm": 0.631157398223877,
+      "grad_norm": 0.5943441390991211,
       "learning_rate": 0.0004850901241230437,
-      "loss": 3.6591,
+      "loss": 3.6579,
       "step": 17850
     },
     {
       "epoch": 1.9299191374663072,
-      "grad_norm": 0.6010056734085083,
-      "learning_rate": 0.0004847663248785753,
-      "loss": 3.6333,
+      "grad_norm": 0.5650938153266907,
+      "learning_rate": 0.00048477280086346464,
+      "loss": 3.6337,
       "step": 17900
     },
     {
       "epoch": 1.935309973045822,
-      "grad_norm": 0.5925098061561584,
-      "learning_rate": 0.0004844425256341068,
-      "loss": 3.6426,
+      "grad_norm": 0.5545641779899597,
+      "learning_rate": 0.00048444900161899614,
+      "loss": 3.6433,
       "step": 17950
     },
     {
       "epoch": 1.940700808625337,
-      "grad_norm": 0.6020907163619995,
-      "learning_rate": 0.00048411872638963834,
-      "loss": 3.6398,
+      "grad_norm": 0.6059868931770325,
+      "learning_rate": 0.00048412520237452774,
+      "loss": 3.6383,
       "step": 18000
     },
     {
       "epoch": 1.940700808625337,
-      "eval_accuracy": 0.36062664901129676,
-      "eval_loss": 3.586404323577881,
-      "eval_runtime": 184.7776,
-      "eval_samples_per_second": 97.474,
-      "eval_steps_per_second": 6.094,
+      "eval_accuracy": 0.360640556580937,
+      "eval_loss": 3.586578607559204,
+      "eval_runtime": 184.65,
+      "eval_samples_per_second": 97.541,
+      "eval_steps_per_second": 6.098,
       "step": 18000
     },
     {
       "epoch": 1.9460916442048517,
-      "grad_norm": 0.5491907596588135,
-      "learning_rate": 0.00048379492714516995,
+      "grad_norm": 0.5713542103767395,
+      "learning_rate": 0.0004838014031300593,
       "loss": 3.6239,
       "step": 18050
     },
     {
       "epoch": 1.9514824797843666,
-      "grad_norm": 0.5421905517578125,
-      "learning_rate": 0.0004834711279007015,
-      "loss": 3.6544,
+      "grad_norm": 0.5506201386451721,
+      "learning_rate": 0.0004834776038855909,
+      "loss": 3.6535,
       "step": 18100
     },
     {
       "epoch": 1.9568733153638815,
-      "grad_norm": 0.5486034154891968,
+      "grad_norm": 0.5446757674217224,
       "learning_rate": 0.00048315380464112245,
-      "loss": 3.6525,
+      "loss": 3.6543,
       "step": 18150
     },
     {
       "epoch": 1.9622641509433962,
-      "grad_norm": 0.6060916185379028,
+      "grad_norm": 0.5964956283569336,
       "learning_rate": 0.00048283000539665405,
-      "loss": 3.6409,
+      "loss": 3.641,
       "step": 18200
     },
     {
       "epoch": 1.967654986522911,
-      "grad_norm": 0.5878822207450867,
+      "grad_norm": 0.5960961580276489,
       "learning_rate": 0.0004825062061521856,
-      "loss": 3.6446,
+      "loss": 3.6438,
       "step": 18250
     },
     {
       "epoch": 1.9730458221024259,
-      "grad_norm": 0.5905686020851135,
+      "grad_norm": 0.5544432401657104,
       "learning_rate": 0.00048218240690771716,
-      "loss": 3.6221,
+      "loss": 3.6236,
       "step": 18300
     },
     {
       "epoch": 1.9784366576819408,
-      "grad_norm": 0.8368579149246216,
+      "grad_norm": 0.766299843788147,
       "learning_rate": 0.00048185860766324876,
-      "loss": 3.6497,
+      "loss": 3.6485,
       "step": 18350
     },
     {
       "epoch": 1.9838274932614555,
-      "grad_norm": 0.5327203869819641,
+      "grad_norm": 0.5031926035881042,
       "learning_rate": 0.0004815348084187803,
-      "loss": 3.6241,
+      "loss": 3.6232,
       "step": 18400
     },
     {
       "epoch": 1.9892183288409704,
-      "grad_norm": 0.5534670948982239,
+      "grad_norm": 0.5916299223899841,
       "learning_rate": 0.0004812110091743119,
-      "loss": 3.6272,
+      "loss": 3.6269,
       "step": 18450
     },
     {
       "epoch": 1.9946091644204853,
-      "grad_norm": 0.5606032013893127,
+      "grad_norm": 0.5806682705879211,
       "learning_rate": 0.00048088720992984347,
-      "loss": 3.6188,
+      "loss": 3.618,
       "step": 18500
     },
     {
       "epoch": 2.0,
-      "grad_norm": 1.0810041427612305,
+      "grad_norm": 1.0938396453857422,
       "learning_rate": 0.00048056341068537507,
-      "loss": 3.6353,
+      "loss": 3.6348,
       "step": 18550
     },
     {
       "epoch": 2.0053908355795147,
-      "grad_norm": 0.6048265099525452,
+      "grad_norm": 0.6288138628005981,
       "learning_rate": 0.00048023961144090657,
-      "loss": 3.5478,
+      "loss": 3.5499,
       "step": 18600
     },
     {
       "epoch": 2.01078167115903,
-      "grad_norm": 0.6086897253990173,
+      "grad_norm": 0.5435187816619873,
       "learning_rate": 0.00047991581219643817,
-      "loss": 3.5393,
+      "loss": 3.539,
       "step": 18650
     },
     {
       "epoch": 2.0161725067385445,
-      "grad_norm": 0.5395742654800415,
+      "grad_norm": 0.5978911519050598,
       "learning_rate": 0.0004795920129519697,
-      "loss": 3.5481,
+      "loss": 3.5485,
       "step": 18700
     },
     {
       "epoch": 2.0215633423180592,
-      "grad_norm": 0.5895045399665833,
+      "grad_norm": 0.5855212211608887,
       "learning_rate": 0.0004792682137075013,
-      "loss": 3.5393,
+      "loss": 3.5387,
       "step": 18750
     },
     {
       "epoch": 2.026954177897574,
-      "grad_norm": 0.5812539458274841,
+      "grad_norm": 0.5721486210823059,
       "learning_rate": 0.0004789444144630329,
-      "loss": 3.5482,
+      "loss": 3.5468,
       "step": 18800
     },
     {
       "epoch": 2.032345013477089,
-      "grad_norm": 0.5699259638786316,
+      "grad_norm": 0.5630236864089966,
       "learning_rate": 0.00047862061521856443,
-      "loss": 3.5458,
+      "loss": 3.5456,
       "step": 18850
     },
     {
       "epoch": 2.0377358490566038,
-      "grad_norm": 0.59361332654953,
+      "grad_norm": 0.5774985551834106,
       "learning_rate": 0.00047829681597409603,
-      "loss": 3.5598,
+      "loss": 3.5596,
       "step": 18900
     },
     {
       "epoch": 2.0431266846361185,
-      "grad_norm": 0.5848776698112488,
+      "grad_norm": 0.6049696207046509,
       "learning_rate": 0.0004779730167296276,
-      "loss": 3.5539,
+      "loss": 3.5537,
       "step": 18950
     },
     {
       "epoch": 2.0485175202156336,
-      "grad_norm": 0.6138538718223572,
+      "grad_norm": 0.6631167531013489,
       "learning_rate": 0.0004776492174851592,
-      "loss": 3.5537,
+      "loss": 3.5558,
       "step": 19000
     },
     {
       "epoch": 2.0485175202156336,
-      "eval_accuracy": 0.3620915072448116,
-      "eval_loss": 3.577709197998047,
-      "eval_runtime": 184.7136,
-      "eval_samples_per_second": 97.508,
-      "eval_steps_per_second": 6.096,
+      "eval_accuracy": 0.36188202447710527,
+      "eval_loss": 3.580307960510254,
+      "eval_runtime": 185.1459,
+      "eval_samples_per_second": 97.28,
+      "eval_steps_per_second": 6.082,
       "step": 19000
     },
     {
       "epoch": 2.0539083557951483,
-      "grad_norm": 0.5327613949775696,
+      "grad_norm": 0.5472202301025391,
       "learning_rate": 0.00047732541824069074,
-      "loss": 3.5637,
+      "loss": 3.5642,
       "step": 19050
     },
     {
       "epoch": 2.059299191374663,
-      "grad_norm": 0.5497714281082153,
+      "grad_norm": 0.5591016411781311,
       "learning_rate": 0.0004770016189962223,
-      "loss": 3.5572,
+      "loss": 3.5543,
       "step": 19100
     },
     {
       "epoch": 2.0646900269541777,
-      "grad_norm": 0.5397418737411499,
+      "grad_norm": 0.545282781124115,
       "learning_rate": 0.0004766778197517539,
-      "loss": 3.5523,
+      "loss": 3.5512,
       "step": 19150
     },
     {
       "epoch": 2.070080862533693,
-      "grad_norm": 0.6005552411079407,
+      "grad_norm": 0.5512953400611877,
       "learning_rate": 0.0004763540205072854,
-      "loss": 3.5718,
+      "loss": 3.5699,
       "step": 19200
     },
     {
       "epoch": 2.0754716981132075,
-      "grad_norm": 0.5779317617416382,
+      "grad_norm": 0.5678845047950745,
       "learning_rate": 0.00047603022126281705,
-      "loss": 3.5608,
+      "loss": 3.5604,
       "step": 19250
     },
     {
       "epoch": 2.0808625336927222,
-      "grad_norm": 0.5900464057922363,
+      "grad_norm": 0.5643090605735779,
       "learning_rate": 0.00047570642201834855,
-      "loss": 3.5525,
+      "loss": 3.5518,
       "step": 19300
     },
     {
       "epoch": 2.0862533692722374,
-      "grad_norm": 0.5555198788642883,
+      "grad_norm": 0.6203526258468628,
       "learning_rate": 0.00047538262277388015,
-      "loss": 3.5668,
+      "loss": 3.5674,
       "step": 19350
     },
     {
       "epoch": 2.091644204851752,
-      "grad_norm": 0.5912312269210815,
+      "grad_norm": 0.5465493202209473,
       "learning_rate": 0.0004750588235294117,
-      "loss": 3.5583,
+      "loss": 3.5609,
       "step": 19400
     },
     {
       "epoch": 2.0970350404312668,
-      "grad_norm": 0.62127685546875,
+      "grad_norm": 0.7021563053131104,
       "learning_rate": 0.0004747350242849433,
-      "loss": 3.5424,
+      "loss": 3.5422,
       "step": 19450
     },
     {
       "epoch": 2.1024258760107815,
-      "grad_norm": 0.5501288175582886,
+      "grad_norm": 0.5742030739784241,
       "learning_rate": 0.00047441122504047486,
-      "loss": 3.5675,
+      "loss": 3.565,
       "step": 19500
     },
     {
       "epoch": 2.1078167115902966,
-      "grad_norm": 0.5837997198104858,
+      "grad_norm": 0.567764163017273,
       "learning_rate": 0.0004740874257960064,
-      "loss": 3.5581,
+      "loss": 3.5572,
       "step": 19550
     },
     {
       "epoch": 2.1132075471698113,
-      "grad_norm": 0.6498492360115051,
+      "grad_norm": 0.5916157364845276,
       "learning_rate": 0.000473763626551538,
-      "loss": 3.5545,
+      "loss": 3.5543,
       "step": 19600
     },
     {
       "epoch": 2.118598382749326,
-      "grad_norm": 0.5806640982627869,
+      "grad_norm": 0.557542622089386,
       "learning_rate": 0.00047343982730706956,
-      "loss": 3.558,
+      "loss": 3.5573,
       "step": 19650
     },
     {
       "epoch": 2.123989218328841,
-      "grad_norm": 0.5803947448730469,
+      "grad_norm": 0.5682848691940308,
       "learning_rate": 0.00047311602806260117,
-      "loss": 3.5497,
+      "loss": 3.5494,
       "step": 19700
     },
     {
       "epoch": 2.129380053908356,
-      "grad_norm": 0.583768904209137,
+      "grad_norm": 0.6219768524169922,
       "learning_rate": 0.0004727922288181327,
-      "loss": 3.5417,
+      "loss": 3.5406,
       "step": 19750
     },
     {
       "epoch": 2.1347708894878705,
-      "grad_norm": 0.5913686156272888,
+      "grad_norm": 0.6204505562782288,
       "learning_rate": 0.0004724684295736643,
-      "loss": 3.5624,
+      "loss": 3.5623,
       "step": 19800
     },
     {
       "epoch": 2.1401617250673857,
-      "grad_norm": 0.5883073210716248,
+      "grad_norm": 0.5561828017234802,
       "learning_rate": 0.0004721446303291959,
       "loss": 3.5538,
       "step": 19850
     },
     {
       "epoch": 2.1455525606469004,
-      "grad_norm": 0.5684877634048462,
+      "grad_norm": 0.5439440011978149,
       "learning_rate": 0.0004718208310847275,
-      "loss": 3.5582,
+      "loss": 3.5568,
       "step": 19900
     },
     {
       "epoch": 2.150943396226415,
-      "grad_norm": 0.5947640538215637,
+      "grad_norm": 0.556121289730072,
       "learning_rate": 0.000471497031840259,
-      "loss": 3.5498,
+      "loss": 3.5496,
       "step": 19950
     },
     {
       "epoch": 2.1563342318059298,
-      "grad_norm": 0.5927711725234985,
+      "grad_norm": 0.6231219172477722,
       "learning_rate": 0.0004711732325957905,
-      "loss": 3.5547,
+      "loss": 3.5512,
       "step": 20000
     },
     {
       "epoch": 2.1563342318059298,
-      "eval_accuracy": 0.3632744112344478,
-      "eval_loss": 3.5697343349456787,
-      "eval_runtime": 185.0798,
-      "eval_samples_per_second": 97.315,
-      "eval_steps_per_second": 6.084,
+      "eval_accuracy": 0.3632916870436103,
+      "eval_loss": 3.5682075023651123,
+      "eval_runtime": 185.4001,
+      "eval_samples_per_second": 97.147,
+      "eval_steps_per_second": 6.073,
       "step": 20000
     },
     {
       "epoch": 2.161725067385445,
-      "grad_norm": 0.6061388254165649,
+      "grad_norm": 0.5872836709022522,
       "learning_rate": 0.00047084943335132213,
-      "loss": 3.555,
+      "loss": 3.5553,
       "step": 20050
     },
     {
       "epoch": 2.1671159029649596,
-      "grad_norm": 0.6365026235580444,
+      "grad_norm": 0.6557894945144653,
       "learning_rate": 0.0004705256341068537,
-      "loss": 3.5514,
+      "loss": 3.5517,
       "step": 20100
     },
     {
       "epoch": 2.1725067385444743,
-      "grad_norm": 0.5569615960121155,
-      "learning_rate": 0.00047020831084727463,
-      "loss": 3.5625,
+      "grad_norm": 0.5851044654846191,
+      "learning_rate": 0.0004702018348623853,
+      "loss": 3.5631,
       "step": 20150
     },
     {
       "epoch": 2.177897574123989,
-      "grad_norm": 0.5334418416023254,
-      "learning_rate": 0.00046988451160280624,
-      "loss": 3.566,
+      "grad_norm": 0.5255047678947449,
+      "learning_rate": 0.00046987803561791684,
+      "loss": 3.5632,
       "step": 20200
     },
     {
       "epoch": 2.183288409703504,
-      "grad_norm": 0.5309886336326599,
-      "learning_rate": 0.0004695607123583378,
-      "loss": 3.5691,
+      "grad_norm": 0.541897177696228,
+      "learning_rate": 0.00046955423637344844,
+      "loss": 3.5704,
       "step": 20250
     },
     {
       "epoch": 2.188679245283019,
-      "grad_norm": 0.6989066004753113,
-      "learning_rate": 0.00046923691311386934,
-      "loss": 3.5521,
+      "grad_norm": 0.6552609205245972,
+      "learning_rate": 0.00046923043712898,
+      "loss": 3.5505,
       "step": 20300
     },
     {
       "epoch": 2.1940700808625335,
-      "grad_norm": 0.5249040722846985,
-      "learning_rate": 0.00046891311386940094,
-      "loss": 3.5447,
+      "grad_norm": 0.54781574010849,
+      "learning_rate": 0.0004689066378845116,
+      "loss": 3.5429,
       "step": 20350
     },
     {
       "epoch": 2.1994609164420487,
-      "grad_norm": 0.5840235352516174,
-      "learning_rate": 0.0004685893146249325,
-      "loss": 3.5579,
+      "grad_norm": 0.587614119052887,
+      "learning_rate": 0.00046858283864004315,
+      "loss": 3.558,
       "step": 20400
     },
     {
       "epoch": 2.2048517520215634,
-      "grad_norm": 0.8253178000450134,
-      "learning_rate": 0.0004682655153804641,
-      "loss": 3.5529,
+      "grad_norm": 0.7089626789093018,
+      "learning_rate": 0.0004682590393955747,
+      "loss": 3.5531,
       "step": 20450
     },
     {
       "epoch": 2.210242587601078,
-      "grad_norm": 0.5192914009094238,
-      "learning_rate": 0.00046794171613599565,
-      "loss": 3.5688,
+      "grad_norm": 0.571935772895813,
+      "learning_rate": 0.0004679352401511063,
+      "loss": 3.5682,
       "step": 20500
     },
     {
       "epoch": 2.215633423180593,
-      "grad_norm": 0.6052975058555603,
-      "learning_rate": 0.00046761791689152725,
-      "loss": 3.5645,
+      "grad_norm": 0.5847911834716797,
+      "learning_rate": 0.0004676114409066378,
+      "loss": 3.563,
       "step": 20550
     },
     {
       "epoch": 2.221024258760108,
-      "grad_norm": 0.6332587599754333,
-      "learning_rate": 0.00046729411764705875,
-      "loss": 3.5602,
+      "grad_norm": 0.6640413999557495,
+      "learning_rate": 0.00046728764166216946,
+      "loss": 3.5612,
       "step": 20600
     },
     {
       "epoch": 2.2264150943396226,
-      "grad_norm": 0.5721376538276672,
-      "learning_rate": 0.0004669703184025904,
-      "loss": 3.5692,
+      "grad_norm": 0.5688964128494263,
+      "learning_rate": 0.00046696384241770095,
+      "loss": 3.5682,
       "step": 20650
     },
     {
       "epoch": 2.2318059299191373,
-      "grad_norm": 0.5925312042236328,
-      "learning_rate": 0.0004666465191581219,
-      "loss": 3.5511,
+      "grad_norm": 0.6168888807296753,
+      "learning_rate": 0.00046664004317323256,
+      "loss": 3.5507,
       "step": 20700
     },
     {
       "epoch": 2.2371967654986524,
-      "grad_norm": 0.5780835151672363,
-      "learning_rate": 0.00046632271991365346,
-      "loss": 3.5705,
+      "grad_norm": 0.5358821749687195,
+      "learning_rate": 0.0004663162439287641,
+      "loss": 3.5699,
       "step": 20750
     },
     {
       "epoch": 2.242587601078167,
-      "grad_norm": 0.5989285111427307,
-      "learning_rate": 0.00046599892066918506,
-      "loss": 3.564,
+      "grad_norm": 0.6521897315979004,
+      "learning_rate": 0.00046599244468429566,
+      "loss": 3.5645,
       "step": 20800
     },
     {
       "epoch": 2.247978436657682,
-      "grad_norm": 0.5859040021896362,
-      "learning_rate": 0.0004656751214247166,
-      "loss": 3.5667,
+      "grad_norm": 0.580860435962677,
+      "learning_rate": 0.00046566864543982726,
+      "loss": 3.5668,
       "step": 20850
     },
     {
       "epoch": 2.2533692722371965,
-      "grad_norm": 0.5282529592514038,
-      "learning_rate": 0.0004653513221802482,
-      "loss": 3.5562,
+      "grad_norm": 0.5867982506752014,
+      "learning_rate": 0.0004653448461953588,
+      "loss": 3.5576,
       "step": 20900
     },
     {
       "epoch": 2.2587601078167117,
-      "grad_norm": 0.5917240381240845,
-      "learning_rate": 0.00046502752293577977,
-      "loss": 3.5576,
+      "grad_norm": 0.5754712224006653,
+      "learning_rate": 0.0004650210469508904,
+      "loss": 3.5556,
       "step": 20950
     },
     {
       "epoch": 2.2641509433962264,
-      "grad_norm": 0.5365268588066101,
-      "learning_rate": 0.00046470372369131137,
-      "loss": 3.5552,
+      "grad_norm": 0.5653140544891357,
+      "learning_rate": 0.00046469724770642197,
+      "loss": 3.5558,
       "step": 21000
     },
     {
       "epoch": 2.2641509433962264,
-      "eval_accuracy": 0.36426271790200837,
-      "eval_loss": 3.5592100620269775,
-      "eval_runtime": 184.8544,
-      "eval_samples_per_second": 97.433,
-      "eval_steps_per_second": 6.091,
+      "eval_accuracy": 0.3638638531508414,
+      "eval_loss": 3.5593972206115723,
+      "eval_runtime": 184.9946,
+      "eval_samples_per_second": 97.36,
+      "eval_steps_per_second": 6.087,
       "step": 21000
     },
     {
       "epoch": 2.269541778975741,
-      "grad_norm": 0.6421787142753601,
-      "learning_rate": 0.0004643799244468429,
-      "loss": 3.5506,
+      "grad_norm": 0.6279696226119995,
+      "learning_rate": 0.0004643734484619536,
+      "loss": 3.5504,
       "step": 21050
     },
     {
       "epoch": 2.274932614555256,
-      "grad_norm": 0.6219934821128845,
-      "learning_rate": 0.0004640561252023745,
-      "loss": 3.5415,
+      "grad_norm": 0.6201480627059937,
+      "learning_rate": 0.0004640496492174851,
+      "loss": 3.5406,
       "step": 21100
     },
     {
       "epoch": 2.280323450134771,
-      "grad_norm": 0.5714690089225769,
-      "learning_rate": 0.0004637323259579061,
-      "loss": 3.5694,
+      "grad_norm": 0.6064199805259705,
+      "learning_rate": 0.00046372584997301673,
+      "loss": 3.5679,
       "step": 21150
     },
     {
       "epoch": 2.2857142857142856,
-      "grad_norm": 0.5706064701080322,
-      "learning_rate": 0.0004634085267134376,
-      "loss": 3.548,
+      "grad_norm": 0.5709848999977112,
+      "learning_rate": 0.0004634020507285483,
+      "loss": 3.5475,
       "step": 21200
     },
     {
       "epoch": 2.2911051212938007,
-      "grad_norm": 0.5707243084907532,
-      "learning_rate": 0.00046308472746896923,
+      "grad_norm": 0.5898020267486572,
+      "learning_rate": 0.0004630782514840798,
       "loss": 3.5456,
       "step": 21250
     },
     {
       "epoch": 2.2964959568733154,
-      "grad_norm": 0.5919787883758545,
-      "learning_rate": 0.00046276092822450073,
-      "loss": 3.5567,
+      "grad_norm": 0.6298962235450745,
+      "learning_rate": 0.0004627544522396114,
+      "loss": 3.5555,
       "step": 21300
     },
     {
       "epoch": 2.30188679245283,
-      "grad_norm": 0.581691563129425,
-      "learning_rate": 0.00046243712898003233,
-      "loss": 3.5333,
+      "grad_norm": 0.5597742795944214,
+      "learning_rate": 0.00046243065299514293,
+      "loss": 3.5329,
       "step": 21350
     },
     {
       "epoch": 2.3072776280323453,
-      "grad_norm": 0.5467907786369324,
-      "learning_rate": 0.0004621133297355639,
-      "loss": 3.5726,
+      "grad_norm": 0.5654948949813843,
+      "learning_rate": 0.00046210685375067454,
+      "loss": 3.5718,
       "step": 21400
     },
     {
       "epoch": 2.31266846361186,
-      "grad_norm": 0.5869949460029602,
-      "learning_rate": 0.0004617895304910955,
-      "loss": 3.5297,
+      "grad_norm": 0.6111935377120972,
+      "learning_rate": 0.0004617830545062061,
+      "loss": 3.53,
       "step": 21450
     },
     {
       "epoch": 2.3180592991913747,
-      "grad_norm": 0.546911895275116,
-      "learning_rate": 0.00046146573124662704,
-      "loss": 3.5624,
+      "grad_norm": 0.5851426720619202,
+      "learning_rate": 0.0004614592552617377,
+      "loss": 3.5606,
       "step": 21500
     },
     {
       "epoch": 2.3234501347708894,
-      "grad_norm": 0.5730652213096619,
-      "learning_rate": 0.00046114193200215864,
-      "loss": 3.5607,
+      "grad_norm": 0.5756331086158752,
+      "learning_rate": 0.00046113545601726924,
+      "loss": 3.5628,
       "step": 21550
     },
     {
       "epoch": 2.3288409703504045,
-      "grad_norm": 0.5786536335945129,
-      "learning_rate": 0.0004608181327576902,
-      "loss": 3.5352,
+      "grad_norm": 0.5970633625984192,
+      "learning_rate": 0.00046081165677280085,
+      "loss": 3.5357,
       "step": 21600
     },
     {
       "epoch": 2.334231805929919,
-      "grad_norm": 0.5828068852424622,
-      "learning_rate": 0.00046049433351322175,
-      "loss": 3.5558,
+      "grad_norm": 0.6157798171043396,
+      "learning_rate": 0.0004604878575283324,
+      "loss": 3.5545,
       "step": 21650
     },
     {
       "epoch": 2.339622641509434,
-      "grad_norm": 0.57169508934021,
-      "learning_rate": 0.00046017053426875335,
-      "loss": 3.5672,
+      "grad_norm": 0.5729793906211853,
+      "learning_rate": 0.00046016405828386395,
+      "loss": 3.567,
       "step": 21700
     },
     {
       "epoch": 2.3450134770889486,
-      "grad_norm": 0.6271734237670898,
-      "learning_rate": 0.0004598467350242849,
-      "loss": 3.55,
+      "grad_norm": 0.6059099435806274,
+      "learning_rate": 0.00045984025903939555,
+      "loss": 3.5498,
       "step": 21750
     },
     {
       "epoch": 2.3504043126684637,
-      "grad_norm": 0.589015543460846,
-      "learning_rate": 0.0004595229357798165,
-      "loss": 3.5597,
+      "grad_norm": 0.5564419031143188,
+      "learning_rate": 0.0004595164597949271,
+      "loss": 3.561,
       "step": 21800
     },
     {
       "epoch": 2.3557951482479784,
-      "grad_norm": 0.5915622115135193,
-      "learning_rate": 0.00045919913653534806,
-      "loss": 3.561,
+      "grad_norm": 0.6240026950836182,
+      "learning_rate": 0.0004591926605504587,
+      "loss": 3.5613,
       "step": 21850
     },
     {
       "epoch": 2.361185983827493,
-      "grad_norm": 0.5321695804595947,
-      "learning_rate": 0.00045887533729087966,
-      "loss": 3.5536,
+      "grad_norm": 0.609406590461731,
+      "learning_rate": 0.0004588688613059902,
+      "loss": 3.555,
       "step": 21900
     },
     {
       "epoch": 2.3665768194070083,
-      "grad_norm": 0.5577532649040222,
-      "learning_rate": 0.00045855153804641116,
-      "loss": 3.5468,
+      "grad_norm": 0.5573409795761108,
+      "learning_rate": 0.00045854506206152186,
+      "loss": 3.5453,
       "step": 21950
     },
     {
       "epoch": 2.371967654986523,
-      "grad_norm": 0.5654785633087158,
+      "grad_norm": 0.5612466335296631,
       "learning_rate": 0.0004582277388019427,
-      "loss": 3.5389,
+      "loss": 3.5396,
       "step": 22000
     },
     {
       "epoch": 2.371967654986523,
-      "eval_accuracy": 0.36546517941145124,
-      "eval_loss": 3.5472066402435303,
-      "eval_runtime": 184.8497,
-      "eval_samples_per_second": 97.436,
-      "eval_steps_per_second": 6.091,
+      "eval_accuracy": 0.36513335349206577,
+      "eval_loss": 3.549255132675171,
+      "eval_runtime": 185.1446,
+      "eval_samples_per_second": 97.281,
+      "eval_steps_per_second": 6.082,
       "step": 22000
     },
     {
       "epoch": 2.3773584905660377,
-      "grad_norm": 0.5921724438667297,
+      "grad_norm": 0.6105454564094543,
       "learning_rate": 0.0004579039395574743,
-      "loss": 3.5423,
+      "loss": 3.5424,
       "step": 22050
     },
     {
       "epoch": 2.382749326145553,
-      "grad_norm": 0.5459533929824829,
+      "grad_norm": 0.5819680690765381,
       "learning_rate": 0.00045758014031300586,
-      "loss": 3.5607,
+      "loss": 3.5612,
       "step": 22100
     },
     {
       "epoch": 2.3881401617250675,
-      "grad_norm": 0.5743412971496582,
+      "grad_norm": 0.5976172089576721,
       "learning_rate": 0.00045725634106853747,
-      "loss": 3.5526,
+      "loss": 3.552,
       "step": 22150
     },
     {
       "epoch": 2.393530997304582,
-      "grad_norm": 0.5897930264472961,
-      "learning_rate": 0.0004569390178089584,
-      "loss": 3.5395,
+      "grad_norm": 0.5951831936836243,
+      "learning_rate": 0.000456932541824069,
+      "loss": 3.5414,
       "step": 22200
     },
     {
       "epoch": 2.398921832884097,
-      "grad_norm": 0.7762314677238464,
-      "learning_rate": 0.00045661521856448997,
-      "loss": 3.5346,
+      "grad_norm": 0.785123348236084,
+      "learning_rate": 0.0004566087425796006,
+      "loss": 3.534,
       "step": 22250
     },
     {
       "epoch": 2.404312668463612,
-      "grad_norm": 0.5821401476860046,
-      "learning_rate": 0.0004562914193200216,
-      "loss": 3.5631,
+      "grad_norm": 0.5873304009437561,
+      "learning_rate": 0.0004562849433351322,
+      "loss": 3.5623,
       "step": 22300
     },
     {
       "epoch": 2.4097035040431267,
-      "grad_norm": 0.6131104826927185,
-      "learning_rate": 0.0004559676200755531,
-      "loss": 3.5702,
+      "grad_norm": 0.5963989496231079,
+      "learning_rate": 0.0004559611440906638,
+      "loss": 3.5714,
       "step": 22350
     },
     {
       "epoch": 2.4150943396226414,
-      "grad_norm": 0.64360111951828,
-      "learning_rate": 0.0004556438208310847,
-      "loss": 3.5555,
+      "grad_norm": 0.6528258323669434,
+      "learning_rate": 0.00045563734484619533,
+      "loss": 3.5552,
       "step": 22400
     },
     {
       "epoch": 2.420485175202156,
-      "grad_norm": 0.5733264684677124,
-      "learning_rate": 0.0004553200215866163,
-      "loss": 3.5617,
+      "grad_norm": 0.5636788606643677,
+      "learning_rate": 0.0004553135456017269,
+      "loss": 3.5598,
       "step": 22450
     },
     {
       "epoch": 2.4258760107816713,
-      "grad_norm": 0.6092203855514526,
-      "learning_rate": 0.00045499622234214783,
-      "loss": 3.5574,
+      "grad_norm": 0.6383706331253052,
+      "learning_rate": 0.0004549897463572585,
+      "loss": 3.5585,
       "step": 22500
     },
     {
       "epoch": 2.431266846361186,
-      "grad_norm": 0.5596003532409668,
-      "learning_rate": 0.00045467242309767944,
-      "loss": 3.5368,
+      "grad_norm": 0.5658510327339172,
+      "learning_rate": 0.00045466594711279,
+      "loss": 3.5387,
       "step": 22550
     },
     {
       "epoch": 2.4366576819407006,
-      "grad_norm": 0.5687330961227417,
-      "learning_rate": 0.00045434862385321093,
-      "loss": 3.5544,
+      "grad_norm": 0.6189313530921936,
+      "learning_rate": 0.00045434214786832164,
+      "loss": 3.5571,
       "step": 22600
     },
     {
       "epoch": 2.442048517520216,
-      "grad_norm": 0.5486346483230591,
-      "learning_rate": 0.0004540248246087426,
-      "loss": 3.5514,
+      "grad_norm": 0.5614275336265564,
+      "learning_rate": 0.00045401834862385314,
+      "loss": 3.5504,
       "step": 22650
     },
     {
       "epoch": 2.4474393530997305,
-      "grad_norm": 0.5911766290664673,
-      "learning_rate": 0.0004537010253642741,
-      "loss": 3.558,
+      "grad_norm": 0.6230819225311279,
+      "learning_rate": 0.00045369454937938474,
+      "loss": 3.5575,
       "step": 22700
     },
     {
       "epoch": 2.452830188679245,
-      "grad_norm": 0.6131563186645508,
-      "learning_rate": 0.00045337722611980564,
-      "loss": 3.5475,
+      "grad_norm": 0.6143561005592346,
+      "learning_rate": 0.0004533707501349163,
+      "loss": 3.5467,
       "step": 22750
     },
     {
       "epoch": 2.4582210242587603,
-      "grad_norm": 0.5406696796417236,
-      "learning_rate": 0.00045305342687533724,
-      "loss": 3.5442,
+      "grad_norm": 0.5280505418777466,
+      "learning_rate": 0.0004530469508904479,
+      "loss": 3.5427,
       "step": 22800
     },
     {
       "epoch": 2.463611859838275,
-      "grad_norm": 0.6020655632019043,
-      "learning_rate": 0.0004527296276308688,
-      "loss": 3.5327,
+      "grad_norm": 0.5940893888473511,
+      "learning_rate": 0.00045272315164597945,
+      "loss": 3.535,
       "step": 22850
     },
     {
       "epoch": 2.4690026954177897,
-      "grad_norm": 0.6064804792404175,
-      "learning_rate": 0.0004524058283864004,
-      "loss": 3.5471,
+      "grad_norm": 0.5543529391288757,
+      "learning_rate": 0.000452399352401511,
+      "loss": 3.5475,
       "step": 22900
     },
     {
       "epoch": 2.4743935309973044,
-      "grad_norm": 0.6027748584747314,
-      "learning_rate": 0.00045208202914193195,
-      "loss": 3.5323,
+      "grad_norm": 0.630915641784668,
+      "learning_rate": 0.0004520755531570426,
+      "loss": 3.5332,
       "step": 22950
     },
     {
       "epoch": 2.4797843665768196,
-      "grad_norm": 0.6128622889518738,
-      "learning_rate": 0.00045175822989746355,
-      "loss": 3.5484,
+      "grad_norm": 0.5720409750938416,
+      "learning_rate": 0.00045175175391257415,
+      "loss": 3.55,
       "step": 23000
     },
     {
       "epoch": 2.4797843665768196,
-      "eval_accuracy": 0.3659796508352528,
-      "eval_loss": 3.539062261581421,
-      "eval_runtime": 185.3247,
-      "eval_samples_per_second": 97.186,
-      "eval_steps_per_second": 6.076,
+      "eval_accuracy": 0.36602463313080796,
+      "eval_loss": 3.5372166633605957,
+      "eval_runtime": 185.2165,
+      "eval_samples_per_second": 97.243,
+      "eval_steps_per_second": 6.079,
       "step": 23000
     },
     {
       "epoch": 2.4851752021563343,
-      "grad_norm": 0.6107762455940247,
-      "learning_rate": 0.0004514344306529951,
-      "loss": 3.5545,
+      "grad_norm": 0.6525805592536926,
+      "learning_rate": 0.00045142795466810576,
+      "loss": 3.5552,
       "step": 23050
     },
     {
       "epoch": 2.490566037735849,
-      "grad_norm": 0.5848401784896851,
-      "learning_rate": 0.0004511106314085267,
-      "loss": 3.5517,
+      "grad_norm": 0.5877313613891602,
+      "learning_rate": 0.0004511041554236373,
+      "loss": 3.5516,
       "step": 23100
     },
     {
       "epoch": 2.4959568733153636,
-      "grad_norm": 0.5667966604232788,
-      "learning_rate": 0.00045078683216405826,
-      "loss": 3.5641,
+      "grad_norm": 0.6073674559593201,
+      "learning_rate": 0.0004507803561791689,
+      "loss": 3.5651,
       "step": 23150
     },
     {
       "epoch": 2.501347708894879,
-      "grad_norm": 0.6321859359741211,
-      "learning_rate": 0.00045046303291958976,
-      "loss": 3.556,
+      "grad_norm": 0.6147558689117432,
+      "learning_rate": 0.00045045655693470046,
+      "loss": 3.5572,
       "step": 23200
     },
     {
       "epoch": 2.5067385444743935,
-      "grad_norm": 0.5637291669845581,
-      "learning_rate": 0.0004501392336751214,
-      "loss": 3.5556,
+      "grad_norm": 0.5562535524368286,
+      "learning_rate": 0.00045013275769023207,
+      "loss": 3.5547,
       "step": 23250
     },
     {
       "epoch": 2.512129380053908,
-      "grad_norm": 0.5089269280433655,
-      "learning_rate": 0.0004498154344306529,
-      "loss": 3.5528,
+      "grad_norm": 0.5438495874404907,
+      "learning_rate": 0.00044980895844576356,
+      "loss": 3.5517,
       "step": 23300
     },
     {
       "epoch": 2.5175202156334233,
-      "grad_norm": 0.5717602968215942,
-      "learning_rate": 0.0004494916351861845,
-      "loss": 3.5522,
+      "grad_norm": 0.5914117693901062,
+      "learning_rate": 0.0004494851592012951,
+      "loss": 3.5508,
       "step": 23350
     },
     {
       "epoch": 2.522911051212938,
-      "grad_norm": 0.5922253727912903,
-      "learning_rate": 0.00044916783594171607,
-      "loss": 3.5279,
+      "grad_norm": 0.593285322189331,
+      "learning_rate": 0.0004491613599568267,
+      "loss": 3.529,
       "step": 23400
     },
     {
       "epoch": 2.5283018867924527,
-      "grad_norm": 0.6174511909484863,
-      "learning_rate": 0.00044884403669724767,
-      "loss": 3.5454,
+      "grad_norm": 0.5920347571372986,
+      "learning_rate": 0.00044883756071235827,
+      "loss": 3.546,
       "step": 23450
     },
     {
       "epoch": 2.533692722371968,
-      "grad_norm": 0.5651429891586304,
-      "learning_rate": 0.0004485202374527792,
-      "loss": 3.541,
+      "grad_norm": 0.5992326736450195,
+      "learning_rate": 0.0004485137614678899,
+      "loss": 3.5419,
       "step": 23500
     },
     {
       "epoch": 2.5390835579514826,
-      "grad_norm": 0.5340074896812439,
-      "learning_rate": 0.00044819643820831083,
-      "loss": 3.5434,
+      "grad_norm": 0.5436864495277405,
+      "learning_rate": 0.0004481899622234214,
+      "loss": 3.5438,
       "step": 23550
     },
     {
       "epoch": 2.5444743935309972,
-      "grad_norm": 0.5876573920249939,
-      "learning_rate": 0.0004478726389638424,
-      "loss": 3.5491,
+      "grad_norm": 0.6082063913345337,
+      "learning_rate": 0.00044786616297895303,
+      "loss": 3.5472,
       "step": 23600
     },
     {
       "epoch": 2.5498652291105124,
-      "grad_norm": 0.544531524181366,
-      "learning_rate": 0.00044754883971937393,
-      "loss": 3.5358,
+      "grad_norm": 0.5552800893783569,
+      "learning_rate": 0.0004475423637344846,
+      "loss": 3.5362,
       "step": 23650
     },
     {
       "epoch": 2.555256064690027,
-      "grad_norm": 0.5737462043762207,
-      "learning_rate": 0.00044722504047490553,
-      "loss": 3.5247,
+      "grad_norm": 0.5726051926612854,
+      "learning_rate": 0.00044721856449001613,
+      "loss": 3.5254,
       "step": 23700
     },
     {
       "epoch": 2.560646900269542,
-      "grad_norm": 0.5849447846412659,
-      "learning_rate": 0.0004469012412304371,
-      "loss": 3.5321,
+      "grad_norm": 0.6111428737640381,
+      "learning_rate": 0.00044689476524554774,
+      "loss": 3.5348,
       "step": 23750
     },
     {
       "epoch": 2.5660377358490565,
-      "grad_norm": 0.548485517501831,
-      "learning_rate": 0.0004465774419859687,
-      "loss": 3.5518,
+      "grad_norm": 0.5674059987068176,
+      "learning_rate": 0.0004465709660010793,
+      "loss": 3.5519,
       "step": 23800
     },
     {
       "epoch": 2.571428571428571,
-      "grad_norm": 0.5926434993743896,
-      "learning_rate": 0.00044625364274150024,
-      "loss": 3.5384,
+      "grad_norm": 0.5440058708190918,
+      "learning_rate": 0.0004462471667566109,
+      "loss": 3.5381,
       "step": 23850
     },
     {
       "epoch": 2.5768194070080863,
-      "grad_norm": 0.5925818681716919,
-      "learning_rate": 0.00044592984349703184,
-      "loss": 3.5406,
+      "grad_norm": 0.57765132188797,
+      "learning_rate": 0.00044592336751214244,
+      "loss": 3.5399,
       "step": 23900
     },
     {
       "epoch": 2.582210242587601,
-      "grad_norm": 0.5864969491958618,
-      "learning_rate": 0.00044560604425256334,
-      "loss": 3.5313,
+      "grad_norm": 0.6293656826019287,
+      "learning_rate": 0.00044559956826767405,
+      "loss": 3.5326,
       "step": 23950
     },
     {
       "epoch": 2.5876010781671157,
-      "grad_norm": 0.5714327096939087,
+      "grad_norm": 0.5684266686439514,
       "learning_rate": 0.000445282245008095,
-      "loss": 3.5354,
+      "loss": 3.5369,
       "step": 24000
     },
     {
       "epoch": 2.5876010781671157,
-      "eval_accuracy": 0.36750046530599206,
-      "eval_loss": 3.5272812843322754,
-      "eval_runtime": 184.8982,
-      "eval_samples_per_second": 97.41,
-      "eval_steps_per_second": 6.09,
+      "eval_accuracy": 0.3672214446900845,
+      "eval_loss": 3.5291759967803955,
+      "eval_runtime": 184.7979,
+      "eval_samples_per_second": 97.463,
+      "eval_steps_per_second": 6.093,
       "step": 24000
     },
     {
       "epoch": 2.592991913746631,
-      "grad_norm": 0.5608143210411072,
+      "grad_norm": 0.5553839802742004,
       "learning_rate": 0.0004449584457636265,
-      "loss": 3.5616,
+      "loss": 3.5619,
       "step": 24050
     },
     {
       "epoch": 2.5983827493261455,
-      "grad_norm": 0.6214185357093811,
+      "grad_norm": 0.6244831681251526,
       "learning_rate": 0.00044463464651915805,
-      "loss": 3.5359,
+      "loss": 3.536,
       "step": 24100
     },
     {
       "epoch": 2.6037735849056602,
-      "grad_norm": 0.561044454574585,
+      "grad_norm": 0.5737437009811401,
       "learning_rate": 0.00044431084727468965,
-      "loss": 3.5475,
+      "loss": 3.5483,
       "step": 24150
     },
     {
       "epoch": 2.6091644204851754,
-      "grad_norm": 0.5782501101493835,
-      "learning_rate": 0.0004439935240151106,
-      "loss": 3.5338,
+      "grad_norm": 0.6138004660606384,
+      "learning_rate": 0.0004439870480302212,
+      "loss": 3.5341,
       "step": 24200
     },
     {
       "epoch": 2.61455525606469,
-      "grad_norm": 0.5893598198890686,
-      "learning_rate": 0.00044366972477064215,
-      "loss": 3.5526,
+      "grad_norm": 0.5860102772712708,
+      "learning_rate": 0.0004436632487857528,
+      "loss": 3.5505,
       "step": 24250
     },
     {
       "epoch": 2.6199460916442048,
-      "grad_norm": 0.6279098987579346,
-      "learning_rate": 0.00044334592552617376,
-      "loss": 3.5458,
+      "grad_norm": 0.6097062230110168,
+      "learning_rate": 0.00044333944954128436,
+      "loss": 3.5457,
       "step": 24300
     },
     {
       "epoch": 2.62533692722372,
-      "grad_norm": 0.5761608481407166,
-      "learning_rate": 0.0004430221262817053,
-      "loss": 3.5383,
+      "grad_norm": 0.6122974157333374,
+      "learning_rate": 0.00044301565029681596,
+      "loss": 3.538,
       "step": 24350
     },
     {
       "epoch": 2.6307277628032346,
-      "grad_norm": 0.5751183032989502,
-      "learning_rate": 0.00044269832703723686,
-      "loss": 3.5401,
+      "grad_norm": 0.5906981825828552,
+      "learning_rate": 0.0004426918510523475,
+      "loss": 3.5389,
       "step": 24400
     },
     {
       "epoch": 2.6361185983827493,
-      "grad_norm": 0.5772199034690857,
-      "learning_rate": 0.00044237452779276846,
-      "loss": 3.5392,
+      "grad_norm": 0.5892931222915649,
+      "learning_rate": 0.0004423680518078791,
+      "loss": 3.5378,
       "step": 24450
     },
     {
       "epoch": 2.641509433962264,
-      "grad_norm": 0.592406690120697,
-      "learning_rate": 0.0004420507285483,
-      "loss": 3.5474,
+      "grad_norm": 0.5747993588447571,
+      "learning_rate": 0.00044204425256341067,
+      "loss": 3.5483,
       "step": 24500
     },
     {
       "epoch": 2.6469002695417787,
-      "grad_norm": 0.6260506510734558,
-      "learning_rate": 0.0004417269293038316,
-      "loss": 3.5316,
+      "grad_norm": 0.5863109827041626,
+      "learning_rate": 0.0004417204533189422,
+      "loss": 3.5317,
       "step": 24550
     },
     {
       "epoch": 2.652291105121294,
-      "grad_norm": 0.603329062461853,
-      "learning_rate": 0.0004414031300593631,
-      "loss": 3.5403,
+      "grad_norm": 0.6000504493713379,
+      "learning_rate": 0.0004413966540744738,
+      "loss": 3.5408,
       "step": 24600
     },
     {
       "epoch": 2.6576819407008085,
-      "grad_norm": 0.5417729020118713,
-      "learning_rate": 0.0004410793308148948,
-      "loss": 3.5352,
+      "grad_norm": 0.5356788635253906,
+      "learning_rate": 0.0004410728548300053,
+      "loss": 3.536,
       "step": 24650
     },
     {
       "epoch": 2.6630727762803232,
-      "grad_norm": 0.5827444791793823,
-      "learning_rate": 0.00044075553157042627,
-      "loss": 3.5417,
+      "grad_norm": 0.5767205953598022,
+      "learning_rate": 0.0004407490555855369,
+      "loss": 3.5419,
       "step": 24700
     },
     {
       "epoch": 2.6684636118598384,
-      "grad_norm": 0.5908746719360352,
-      "learning_rate": 0.0004404317323259579,
-      "loss": 3.5581,
+      "grad_norm": 0.5705932974815369,
+      "learning_rate": 0.0004404252563410685,
+      "loss": 3.5596,
       "step": 24750
     },
     {
       "epoch": 2.673854447439353,
-      "grad_norm": 0.5514307618141174,
-      "learning_rate": 0.0004401079330814894,
-      "loss": 3.5529,
+      "grad_norm": 0.5411615967750549,
+      "learning_rate": 0.0004401014570966001,
+      "loss": 3.5523,
       "step": 24800
     },
     {
       "epoch": 2.6792452830188678,
-      "grad_norm": 0.5702772736549377,
-      "learning_rate": 0.000439784133837021,
-      "loss": 3.5533,
+      "grad_norm": 0.5571548342704773,
+      "learning_rate": 0.00043977765785213163,
+      "loss": 3.5519,
       "step": 24850
     },
     {
       "epoch": 2.684636118598383,
-      "grad_norm": 0.6261325478553772,
-      "learning_rate": 0.0004394603345925526,
-      "loss": 3.5245,
+      "grad_norm": 0.6669823527336121,
+      "learning_rate": 0.0004394538586076632,
+      "loss": 3.5243,
       "step": 24900
     },
     {
       "epoch": 2.6900269541778976,
-      "grad_norm": 0.602677047252655,
-      "learning_rate": 0.00043913653534808413,
-      "loss": 3.5444,
+      "grad_norm": 0.5549421906471252,
+      "learning_rate": 0.0004391300593631948,
+      "loss": 3.5437,
       "step": 24950
     },
     {
       "epoch": 2.6954177897574123,
-      "grad_norm": 0.5706930160522461,
-      "learning_rate": 0.00043881273610361574,
-      "loss": 3.5216,
+      "grad_norm": 0.6019677519798279,
+      "learning_rate": 0.00043880626011872634,
+      "loss": 3.5235,
       "step": 25000
     },
     {
       "epoch": 2.6954177897574123,
-      "eval_accuracy": 0.3683038447584923,
-      "eval_loss": 3.5193662643432617,
-      "eval_runtime": 185.1095,
-      "eval_samples_per_second": 97.299,
-      "eval_steps_per_second": 6.083,
+      "eval_accuracy": 0.3678405488448514,
+      "eval_loss": 3.520496129989624,
+      "eval_runtime": 184.864,
+      "eval_samples_per_second": 97.428,
+      "eval_steps_per_second": 6.091,
       "step": 25000
     },
     {
       "epoch": 2.7008086253369274,
-      "grad_norm": 0.5820638537406921,
-      "learning_rate": 0.0004384889368591473,
-      "loss": 3.537,
+      "grad_norm": 0.5758979916572571,
+      "learning_rate": 0.00043848246087425794,
+      "loss": 3.5348,
       "step": 25050
     },
     {
       "epoch": 2.706199460916442,
-      "grad_norm": 0.5792041420936584,
-      "learning_rate": 0.0004381651376146789,
-      "loss": 3.5356,
+      "grad_norm": 0.5502886176109314,
+      "learning_rate": 0.0004381586616297895,
+      "loss": 3.538,
       "step": 25100
     },
     {
       "epoch": 2.711590296495957,
-      "grad_norm": 0.5905164480209351,
-      "learning_rate": 0.00043784133837021044,
-      "loss": 3.5398,
+      "grad_norm": 0.5907125473022461,
+      "learning_rate": 0.0004378348623853211,
+      "loss": 3.5389,
       "step": 25150
     },
     {
       "epoch": 2.7169811320754715,
-      "grad_norm": 0.5989269018173218,
-      "learning_rate": 0.00043751753912574205,
-      "loss": 3.5395,
+      "grad_norm": 0.5703557729721069,
+      "learning_rate": 0.00043751106314085265,
+      "loss": 3.5389,
       "step": 25200
     },
     {
       "epoch": 2.7223719676549867,
-      "grad_norm": 0.5796598196029663,
-      "learning_rate": 0.0004371937398812736,
-      "loss": 3.5316,
+      "grad_norm": 0.605928361415863,
+      "learning_rate": 0.00043718726389638425,
+      "loss": 3.5305,
       "step": 25250
     },
     {
       "epoch": 2.7277628032345014,
-      "grad_norm": 0.5831441283226013,
-      "learning_rate": 0.0004368699406368051,
-      "loss": 3.54,
+      "grad_norm": 0.5767539739608765,
+      "learning_rate": 0.00043686346465191575,
+      "loss": 3.5402,
       "step": 25300
     },
     {
       "epoch": 2.733153638814016,
-      "grad_norm": 0.662450909614563,
-      "learning_rate": 0.0004365461413923367,
-      "loss": 3.5337,
+      "grad_norm": 0.6421152353286743,
+      "learning_rate": 0.0004365396654074473,
+      "loss": 3.5344,
       "step": 25350
     },
     {
       "epoch": 2.7385444743935308,
-      "grad_norm": 0.5832592844963074,
-      "learning_rate": 0.00043622234214786825,
-      "loss": 3.5294,
+      "grad_norm": 0.5596376061439514,
+      "learning_rate": 0.0004362158661629789,
+      "loss": 3.5272,
       "step": 25400
     },
     {
       "epoch": 2.743935309973046,
-      "grad_norm": 0.6044817566871643,
-      "learning_rate": 0.00043589854290339985,
-      "loss": 3.5455,
+      "grad_norm": 0.5788835883140564,
+      "learning_rate": 0.00043589206691851045,
+      "loss": 3.545,
       "step": 25450
     },
     {
       "epoch": 2.7493261455525606,
-      "grad_norm": 0.5936079025268555,
-      "learning_rate": 0.0004355747436589314,
-      "loss": 3.5239,
+      "grad_norm": 0.6065754294395447,
+      "learning_rate": 0.00043556826767404206,
+      "loss": 3.5221,
       "step": 25500
     },
     {
       "epoch": 2.7547169811320753,
-      "grad_norm": 0.6509273052215576,
-      "learning_rate": 0.000435250944414463,
-      "loss": 3.5151,
+      "grad_norm": 0.6557634472846985,
+      "learning_rate": 0.0004352444684295736,
+      "loss": 3.5164,
       "step": 25550
     },
     {
       "epoch": 2.7601078167115904,
-      "grad_norm": 0.6094829440116882,
-      "learning_rate": 0.00043492714516999456,
-      "loss": 3.5212,
+      "grad_norm": 0.5632269382476807,
+      "learning_rate": 0.0004349206691851052,
+      "loss": 3.5205,
       "step": 25600
     },
     {
       "epoch": 2.765498652291105,
-      "grad_norm": 0.6017758846282959,
-      "learning_rate": 0.0004346033459255261,
-      "loss": 3.5276,
+      "grad_norm": 0.7714040875434875,
+      "learning_rate": 0.00043459686994063676,
+      "loss": 3.5271,
       "step": 25650
     },
     {
       "epoch": 2.77088948787062,
-      "grad_norm": 0.5787220597267151,
-      "learning_rate": 0.0004342795466810577,
-      "loss": 3.5148,
+      "grad_norm": 0.572063684463501,
+      "learning_rate": 0.00043427307069616837,
+      "loss": 3.5135,
       "step": 25700
     },
     {
       "epoch": 2.776280323450135,
-      "grad_norm": 0.6034836769104004,
-      "learning_rate": 0.00043395574743658927,
-      "loss": 3.5171,
+      "grad_norm": 0.5755751132965088,
+      "learning_rate": 0.0004339492714516999,
+      "loss": 3.5176,
       "step": 25750
     },
     {
       "epoch": 2.7816711590296497,
-      "grad_norm": 0.6130497455596924,
-      "learning_rate": 0.00043363194819212087,
-      "loss": 3.5329,
+      "grad_norm": 0.5770022869110107,
+      "learning_rate": 0.00043362547220723147,
+      "loss": 3.5318,
       "step": 25800
     },
     {
       "epoch": 2.7870619946091644,
-      "grad_norm": 0.5641265511512756,
-      "learning_rate": 0.0004333081489476524,
-      "loss": 3.5179,
+      "grad_norm": 0.5711796283721924,
+      "learning_rate": 0.0004333016729627631,
+      "loss": 3.5175,
       "step": 25850
     },
     {
       "epoch": 2.7924528301886795,
-      "grad_norm": 0.5724943280220032,
-      "learning_rate": 0.000432984349703184,
-      "loss": 3.5203,
+      "grad_norm": 0.561193585395813,
+      "learning_rate": 0.0004329778737182946,
+      "loss": 3.5207,
       "step": 25900
     },
     {
       "epoch": 2.797843665768194,
-      "grad_norm": 0.5406250357627869,
-      "learning_rate": 0.0004326605504587155,
-      "loss": 3.5313,
+      "grad_norm": 0.5335695743560791,
+      "learning_rate": 0.00043265407447382623,
+      "loss": 3.5301,
       "step": 25950
     },
     {
       "epoch": 2.803234501347709,
-      "grad_norm": 0.6399490833282471,
-      "learning_rate": 0.0004323367512142472,
-      "loss": 3.5355,
+      "grad_norm": 0.6352400779724121,
+      "learning_rate": 0.0004323302752293577,
+      "loss": 3.5368,
       "step": 26000
     },
     {
       "epoch": 2.803234501347709,
-      "eval_accuracy": 0.3691859975546581,
-      "eval_loss": 3.5100622177124023,
-      "eval_runtime": 184.655,
-      "eval_samples_per_second": 97.539,
-      "eval_steps_per_second": 6.098,
+      "eval_accuracy": 0.3689961809596462,
+      "eval_loss": 3.5120186805725098,
+      "eval_runtime": 185.5238,
+      "eval_samples_per_second": 97.082,
+      "eval_steps_per_second": 6.069,
       "step": 26000
     },
     {
       "epoch": 2.8086253369272236,
-      "grad_norm": 0.6148707866668701,
+      "grad_norm": 0.6344072222709656,
       "learning_rate": 0.0004320129519697787,
-      "loss": 3.5351,
+      "loss": 3.5355,
       "step": 26050
     },
     {
       "epoch": 2.8140161725067383,
-      "grad_norm": 0.5590334534645081,
+      "grad_norm": 0.5527616739273071,
       "learning_rate": 0.00043168915272531023,
-      "loss": 3.5176,
+      "loss": 3.5154,
       "step": 26100
     },
     {
       "epoch": 2.8194070080862534,
-      "grad_norm": 0.585767388343811,
+      "grad_norm": 0.5766391754150391,
       "learning_rate": 0.00043136535348084183,
-      "loss": 3.5271,
+      "loss": 3.526,
       "step": 26150
     },
     {
       "epoch": 2.824797843665768,
-      "grad_norm": 0.6158841252326965,
+      "grad_norm": 0.6358306407928467,
       "learning_rate": 0.0004310415542363734,
-      "loss": 3.5225,
+      "loss": 3.5227,
       "step": 26200
     },
     {
       "epoch": 2.830188679245283,
-      "grad_norm": 0.6011167168617249,
+      "grad_norm": 0.6099495887756348,
       "learning_rate": 0.000430717754991905,
-      "loss": 3.5186,
+      "loss": 3.5171,
       "step": 26250
     },
     {
       "epoch": 2.835579514824798,
-      "grad_norm": 0.6300170421600342,
-      "learning_rate": 0.00043040043173232594,
-      "loss": 3.5321,
+      "grad_norm": 0.603064775466919,
+      "learning_rate": 0.00043039395574743654,
+      "loss": 3.5312,
       "step": 26300
     },
     {
       "epoch": 2.8409703504043127,
-      "grad_norm": 0.6018158197402954,
-      "learning_rate": 0.0004300766324878575,
-      "loss": 3.5369,
+      "grad_norm": 0.6099294424057007,
+      "learning_rate": 0.00043007015650296814,
+      "loss": 3.5359,
       "step": 26350
     },
     {
       "epoch": 2.8463611859838274,
-      "grad_norm": 0.6889051795005798,
-      "learning_rate": 0.0004297528332433891,
-      "loss": 3.524,
+      "grad_norm": 0.5590049624443054,
+      "learning_rate": 0.0004297463572584997,
+      "loss": 3.5232,
       "step": 26400
     },
     {
       "epoch": 2.8517520215633425,
-      "grad_norm": 0.5744615793228149,
-      "learning_rate": 0.00042942903399892065,
-      "loss": 3.5252,
+      "grad_norm": 0.5742785334587097,
+      "learning_rate": 0.0004294225580140313,
+      "loss": 3.523,
       "step": 26450
     },
     {
       "epoch": 2.857142857142857,
-      "grad_norm": 0.6640810966491699,
-      "learning_rate": 0.0004291052347544522,
-      "loss": 3.5226,
+      "grad_norm": 0.6284919381141663,
+      "learning_rate": 0.00042909875876956285,
+      "loss": 3.5214,
       "step": 26500
     },
     {
       "epoch": 2.862533692722372,
-      "grad_norm": 0.6115678548812866,
-      "learning_rate": 0.0004287814355099838,
-      "loss": 3.5145,
+      "grad_norm": 0.626808762550354,
+      "learning_rate": 0.0004287749595250944,
+      "loss": 3.5127,
       "step": 26550
     },
     {
       "epoch": 2.867924528301887,
-      "grad_norm": 0.5903047919273376,
-      "learning_rate": 0.0004284576362655153,
-      "loss": 3.5249,
+      "grad_norm": 0.5913968682289124,
+      "learning_rate": 0.000428451160280626,
+      "loss": 3.5246,
       "step": 26600
     },
     {
       "epoch": 2.8733153638814017,
-      "grad_norm": 0.5858241319656372,
-      "learning_rate": 0.00042813383702104696,
-      "loss": 3.5192,
+      "grad_norm": 0.5790668725967407,
+      "learning_rate": 0.0004281273610361575,
+      "loss": 3.5198,
       "step": 26650
     },
     {
       "epoch": 2.8787061994609164,
-      "grad_norm": 0.5560069680213928,
-      "learning_rate": 0.00042781003777657845,
-      "loss": 3.5193,
+      "grad_norm": 0.5755479335784912,
+      "learning_rate": 0.0004278035617916891,
+      "loss": 3.5196,
       "step": 26700
     },
     {
       "epoch": 2.884097035040431,
-      "grad_norm": 0.5569444298744202,
-      "learning_rate": 0.00042748623853211006,
-      "loss": 3.5385,
+      "grad_norm": 0.5627382397651672,
+      "learning_rate": 0.00042747976254722066,
+      "loss": 3.5359,
       "step": 26750
     },
     {
       "epoch": 2.889487870619946,
-      "grad_norm": 0.6125451922416687,
-      "learning_rate": 0.0004271624392876416,
-      "loss": 3.5248,
+      "grad_norm": 0.6645321249961853,
+      "learning_rate": 0.00042715596330275226,
+      "loss": 3.5238,
       "step": 26800
     },
     {
       "epoch": 2.894878706199461,
-      "grad_norm": 0.5783609747886658,
+      "grad_norm": 0.5590260028839111,
       "learning_rate": 0.00042683864004317316,
-      "loss": 3.52,
+      "loss": 3.517,
       "step": 26850
     },
     {
       "epoch": 2.9002695417789757,
-      "grad_norm": 0.6026281714439392,
+      "grad_norm": 0.5804741382598877,
       "learning_rate": 0.00042651484079870476,
-      "loss": 3.5427,
+      "loss": 3.5424,
       "step": 26900
     },
     {
       "epoch": 2.9056603773584904,
-      "grad_norm": 0.5993627905845642,
+      "grad_norm": 0.5705453157424927,
       "learning_rate": 0.0004261910415542363,
-      "loss": 3.5342,
+      "loss": 3.535,
       "step": 26950
     },
     {
       "epoch": 2.9110512129380055,
-      "grad_norm": 0.5433017015457153,
+      "grad_norm": 0.5481330156326294,
       "learning_rate": 0.0004258672423097679,
-      "loss": 3.5181,
+      "loss": 3.5182,
       "step": 27000
     },
     {
       "epoch": 2.9110512129380055,
-      "eval_accuracy": 0.3701280180920097,
-      "eval_loss": 3.5018980503082275,
-      "eval_runtime": 185.0799,
-      "eval_samples_per_second": 97.315,
-      "eval_steps_per_second": 6.084,
+      "eval_accuracy": 0.3699219035638256,
+      "eval_loss": 3.5017757415771484,
+      "eval_runtime": 185.4733,
+      "eval_samples_per_second": 97.108,
+      "eval_steps_per_second": 6.071,
       "step": 27000
     },
     {
       "epoch": 2.91644204851752,
-      "grad_norm": 0.6073448657989502,
+      "grad_norm": 0.5877059102058411,
       "learning_rate": 0.00042554344306529947,
-      "loss": 3.5363,
+      "loss": 3.536,
       "step": 27050
     },
     {
       "epoch": 2.921832884097035,
-      "grad_norm": 0.5769268274307251,
+      "grad_norm": 0.5492408275604248,
       "learning_rate": 0.0004252196438208311,
-      "loss": 3.5422,
+      "loss": 3.5394,
       "step": 27100
     },
     {
       "epoch": 2.92722371967655,
-      "grad_norm": 0.6756018400192261,
+      "grad_norm": 0.596275806427002,
       "learning_rate": 0.0004248958445763626,
-      "loss": 3.5257,
+      "loss": 3.524,
       "step": 27150
     },
     {
       "epoch": 2.9326145552560647,
-      "grad_norm": 0.6934609413146973,
+      "grad_norm": 0.5996169447898865,
       "learning_rate": 0.00042457204533189423,
-      "loss": 3.5172,
+      "loss": 3.5154,
       "step": 27200
     },
     {
       "epoch": 2.9380053908355794,
-      "grad_norm": 0.558773934841156,
+      "grad_norm": 0.5572853684425354,
       "learning_rate": 0.0004242482460874258,
-      "loss": 3.5289,
+      "loss": 3.5263,
       "step": 27250
     },
     {
       "epoch": 2.9433962264150946,
-      "grad_norm": 0.6366307139396667,
+      "grad_norm": 0.6095960736274719,
       "learning_rate": 0.0004239244468429573,
-      "loss": 3.5247,
+      "loss": 3.5228,
       "step": 27300
     },
     {
       "epoch": 2.9487870619946093,
-      "grad_norm": 0.6284673810005188,
+      "grad_norm": 0.6216484308242798,
       "learning_rate": 0.0004236006475984889,
-      "loss": 3.5192,
+      "loss": 3.5175,
       "step": 27350
     },
     {
       "epoch": 2.954177897574124,
-      "grad_norm": 0.5934546589851379,
+      "grad_norm": 0.5865620970726013,
       "learning_rate": 0.00042327684835402043,
-      "loss": 3.5213,
+      "loss": 3.5215,
       "step": 27400
     },
     {
       "epoch": 2.9595687331536387,
-      "grad_norm": 0.5586668848991394,
+      "grad_norm": 0.553994357585907,
       "learning_rate": 0.00042295304910955204,
-      "loss": 3.5181,
+      "loss": 3.5165,
       "step": 27450
     },
     {
       "epoch": 2.964959568733154,
-      "grad_norm": 0.5713486075401306,
+      "grad_norm": 0.557462751865387,
       "learning_rate": 0.0004226292498650836,
-      "loss": 3.5242,
+      "loss": 3.5238,
       "step": 27500
     },
     {
       "epoch": 2.9703504043126685,
-      "grad_norm": 0.5594163537025452,
+      "grad_norm": 0.5714412331581116,
       "learning_rate": 0.0004223054506206152,
-      "loss": 3.5266,
+      "loss": 3.5264,
       "step": 27550
     },
     {
       "epoch": 2.975741239892183,
-      "grad_norm": 0.5783873796463013,
+      "grad_norm": 0.5819886922836304,
       "learning_rate": 0.00042198165137614674,
-      "loss": 3.5273,
+      "loss": 3.5267,
       "step": 27600
     },
     {
       "epoch": 2.981132075471698,
-      "grad_norm": 0.5773907899856567,
+      "grad_norm": 0.5586761236190796,
       "learning_rate": 0.00042165785213167835,
-      "loss": 3.5309,
+      "loss": 3.5297,
       "step": 27650
     },
     {
       "epoch": 2.986522911051213,
-      "grad_norm": 0.6616211533546448,
+      "grad_norm": 0.6088539958000183,
       "learning_rate": 0.0004213340528872099,
-      "loss": 3.5215,
+      "loss": 3.52,
       "step": 27700
     },
     {
       "epoch": 2.9919137466307277,
-      "grad_norm": 0.5861223936080933,
+      "grad_norm": 0.597719132900238,
       "learning_rate": 0.00042101025364274145,
-      "loss": 3.5321,
+      "loss": 3.5318,
       "step": 27750
     },
     {
       "epoch": 2.9973045822102424,
-      "grad_norm": 0.6247793436050415,
+      "grad_norm": 0.6238610148429871,
       "learning_rate": 0.00042068645439827305,
-      "loss": 3.5279,
+      "loss": 3.5263,
       "step": 27800
     },
     {
       "epoch": 3.0026954177897576,
-      "grad_norm": 0.6077458262443542,
+      "grad_norm": 0.6423973441123962,
       "learning_rate": 0.0004203626551538046,
-      "loss": 3.4729,
+      "loss": 3.474,
       "step": 27850
     },
     {
       "epoch": 3.0080862533692723,
-      "grad_norm": 0.5924696326255798,
+      "grad_norm": 0.6364337801933289,
       "learning_rate": 0.0004200388559093362,
-      "loss": 3.4233,
+      "loss": 3.4241,
       "step": 27900
     },
     {
       "epoch": 3.013477088948787,
-      "grad_norm": 0.6317850351333618,
+      "grad_norm": 0.6469203233718872,
       "learning_rate": 0.0004197150566648677,
-      "loss": 3.4231,
+      "loss": 3.4202,
       "step": 27950
     },
     {
       "epoch": 3.018867924528302,
-      "grad_norm": 0.588413655757904,
+      "grad_norm": 0.557319164276123,
       "learning_rate": 0.00041939125742039936,
-      "loss": 3.4434,
+      "loss": 3.4403,
       "step": 28000
     },
     {
       "epoch": 3.018867924528302,
-      "eval_accuracy": 0.37104624364692984,
-      "eval_loss": 3.4995107650756836,
-      "eval_runtime": 184.7835,
-      "eval_samples_per_second": 97.471,
-      "eval_steps_per_second": 6.094,
+      "eval_accuracy": 0.3709357436600225,
+      "eval_loss": 3.4979920387268066,
+      "eval_runtime": 185.1473,
+      "eval_samples_per_second": 97.279,
+      "eval_steps_per_second": 6.082,
       "step": 28000
     },
     {
       "epoch": 3.024258760107817,
-      "grad_norm": 0.6233013272285461,
+      "grad_norm": 0.6403119564056396,
       "learning_rate": 0.00041906745817593086,
-      "loss": 3.4372,
+      "loss": 3.4365,
       "step": 28050
     },
     {
       "epoch": 3.0296495956873315,
-      "grad_norm": 0.6402817964553833,
+      "grad_norm": 0.5956209301948547,
       "learning_rate": 0.00041874365893146247,
-      "loss": 3.444,
+      "loss": 3.4416,
       "step": 28100
     },
     {
       "epoch": 3.035040431266846,
-      "grad_norm": 0.5920224189758301,
+      "grad_norm": 0.6170210838317871,
       "learning_rate": 0.000418419859686994,
-      "loss": 3.4329,
+      "loss": 3.4325,
       "step": 28150
     },
     {
       "epoch": 3.0404312668463613,
-      "grad_norm": 0.6073600053787231,
+      "grad_norm": 0.6147815585136414,
       "learning_rate": 0.00041809606044252557,
-      "loss": 3.4372,
+      "loss": 3.4369,
       "step": 28200
     },
     {
       "epoch": 3.045822102425876,
-      "grad_norm": 0.5874477028846741,
+      "grad_norm": 0.6116907000541687,
       "learning_rate": 0.00041777226119805717,
-      "loss": 3.4204,
+      "loss": 3.418,
       "step": 28250
     },
     {
       "epoch": 3.0512129380053907,
-      "grad_norm": 0.5746080279350281,
+      "grad_norm": 0.5721989870071411,
       "learning_rate": 0.0004174484619535887,
-      "loss": 3.4409,
+      "loss": 3.4393,
       "step": 28300
     },
     {
       "epoch": 3.056603773584906,
-      "grad_norm": 0.6111629605293274,
-      "learning_rate": 0.0004171311386940097,
-      "loss": 3.4156,
+      "grad_norm": 0.6529187560081482,
+      "learning_rate": 0.00041712466270912033,
+      "loss": 3.4149,
       "step": 28350
     },
     {
       "epoch": 3.0619946091644206,
-      "grad_norm": 0.6025534868240356,
-      "learning_rate": 0.0004168073394495413,
-      "loss": 3.4449,
+      "grad_norm": 0.5911017656326294,
+      "learning_rate": 0.0004168008634646519,
+      "loss": 3.4432,
       "step": 28400
     },
     {
       "epoch": 3.0673854447439353,
-      "grad_norm": 0.6089310050010681,
-      "learning_rate": 0.00041648354020507283,
-      "loss": 3.414,
+      "grad_norm": 0.595206618309021,
+      "learning_rate": 0.0004164770642201835,
+      "loss": 3.4124,
       "step": 28450
     },
     {
       "epoch": 3.07277628032345,
-      "grad_norm": 0.6200700402259827,
-      "learning_rate": 0.0004161597409606044,
-      "loss": 3.4394,
+      "grad_norm": 0.657249927520752,
+      "learning_rate": 0.00041615326497571503,
+      "loss": 3.4377,
       "step": 28500
     },
     {
       "epoch": 3.078167115902965,
-      "grad_norm": 0.6414934992790222,
-      "learning_rate": 0.000415835941716136,
-      "loss": 3.4516,
+      "grad_norm": 0.6252180933952332,
+      "learning_rate": 0.0004158294657312466,
+      "loss": 3.4501,
       "step": 28550
     },
     {
       "epoch": 3.08355795148248,
-      "grad_norm": 0.6194086670875549,
-      "learning_rate": 0.0004155121424716675,
-      "loss": 3.4361,
+      "grad_norm": 0.6360760927200317,
+      "learning_rate": 0.0004155056664867782,
+      "loss": 3.4355,
       "step": 28600
     },
     {
       "epoch": 3.0889487870619945,
-      "grad_norm": 0.6263707876205444,
-      "learning_rate": 0.00041518834322719914,
-      "loss": 3.4466,
+      "grad_norm": 0.6364948749542236,
+      "learning_rate": 0.0004151818672423097,
+      "loss": 3.445,
       "step": 28650
     },
     {
       "epoch": 3.0943396226415096,
-      "grad_norm": 0.598533570766449,
-      "learning_rate": 0.00041486454398273064,
-      "loss": 3.4456,
+      "grad_norm": 0.5616390109062195,
+      "learning_rate": 0.0004148580679978413,
+      "loss": 3.4433,
       "step": 28700
     },
     {
       "epoch": 3.0997304582210243,
-      "grad_norm": 0.5736040472984314,
-      "learning_rate": 0.00041454074473826224,
-      "loss": 3.4713,
+      "grad_norm": 0.6200069189071655,
+      "learning_rate": 0.00041453426875337284,
+      "loss": 3.4692,
       "step": 28750
     },
     {
       "epoch": 3.105121293800539,
-      "grad_norm": 0.6180632710456848,
-      "learning_rate": 0.0004142169454937938,
-      "loss": 3.4296,
+      "grad_norm": 0.5786812901496887,
+      "learning_rate": 0.00041421046950890445,
+      "loss": 3.4284,
       "step": 28800
     },
     {
       "epoch": 3.1105121293800537,
-      "grad_norm": 0.6590995788574219,
-      "learning_rate": 0.0004138931462493254,
-      "loss": 3.4389,
+      "grad_norm": 0.631637454032898,
+      "learning_rate": 0.000413886670264436,
+      "loss": 3.4383,
       "step": 28850
     },
     {
       "epoch": 3.115902964959569,
-      "grad_norm": 0.6105198264122009,
-      "learning_rate": 0.00041356934700485695,
-      "loss": 3.4346,
+      "grad_norm": 0.5604854822158813,
+      "learning_rate": 0.0004135628710199676,
+      "loss": 3.4327,
       "step": 28900
     },
     {
       "epoch": 3.1212938005390836,
-      "grad_norm": 0.6112799048423767,
-      "learning_rate": 0.0004132455477603885,
-      "loss": 3.4646,
+      "grad_norm": 0.5949662923812866,
+      "learning_rate": 0.00041323907177549915,
+      "loss": 3.4644,
       "step": 28950
     },
     {
       "epoch": 3.1266846361185983,
-      "grad_norm": 0.6425892114639282,
-      "learning_rate": 0.0004129217485159201,
-      "loss": 3.4374,
+      "grad_norm": 0.6039780378341675,
+      "learning_rate": 0.0004129152725310307,
+      "loss": 3.435,
       "step": 29000
     },
     {
       "epoch": 3.1266846361185983,
-      "eval_accuracy": 0.37091336116513274,
-      "eval_loss": 3.493210554122925,
-      "eval_runtime": 185.0904,
-      "eval_samples_per_second": 97.309,
-      "eval_steps_per_second": 6.084,
+      "eval_accuracy": 0.3711198016519802,
+      "eval_loss": 3.4920432567596436,
+      "eval_runtime": 185.3198,
+      "eval_samples_per_second": 97.189,
+      "eval_steps_per_second": 6.076,
       "step": 29000
     },
     {
       "epoch": 3.1320754716981134,
-      "grad_norm": 0.5860080718994141,
-      "learning_rate": 0.00041259794927145165,
-      "loss": 3.4543,
+      "grad_norm": 0.6303644776344299,
+      "learning_rate": 0.0004125914732865623,
+      "loss": 3.4524,
       "step": 29050
     },
     {
       "epoch": 3.137466307277628,
-      "grad_norm": 0.5804362893104553,
-      "learning_rate": 0.00041227415002698326,
-      "loss": 3.4552,
+      "grad_norm": 0.617788553237915,
+      "learning_rate": 0.00041226767404209386,
+      "loss": 3.4542,
       "step": 29100
     },
     {
       "epoch": 3.142857142857143,
-      "grad_norm": 0.6355499625205994,
-      "learning_rate": 0.0004119503507825148,
-      "loss": 3.4463,
+      "grad_norm": 0.5849179625511169,
+      "learning_rate": 0.00041194387479762546,
+      "loss": 3.4471,
       "step": 29150
     },
     {
       "epoch": 3.1482479784366575,
-      "grad_norm": 0.5874181389808655,
-      "learning_rate": 0.0004116265515380464,
-      "loss": 3.4556,
+      "grad_norm": 0.5979399681091309,
+      "learning_rate": 0.000411620075553157,
+      "loss": 3.4536,
       "step": 29200
     },
     {
       "epoch": 3.1536388140161726,
-      "grad_norm": 0.6099561452865601,
-      "learning_rate": 0.00041130275229357796,
-      "loss": 3.458,
+      "grad_norm": 0.6045482158660889,
+      "learning_rate": 0.0004112962763086886,
+      "loss": 3.4567,
       "step": 29250
     },
     {
       "epoch": 3.1590296495956873,
-      "grad_norm": 0.6350792646408081,
-      "learning_rate": 0.00041097895304910946,
-      "loss": 3.4627,
+      "grad_norm": 0.6165677905082703,
+      "learning_rate": 0.0004109724770642201,
+      "loss": 3.46,
       "step": 29300
     },
     {
       "epoch": 3.164420485175202,
-      "grad_norm": 0.634882390499115,
-      "learning_rate": 0.00041065515380464107,
-      "loss": 3.4526,
+      "grad_norm": 0.6779764294624329,
+      "learning_rate": 0.00041064867781975177,
+      "loss": 3.4515,
       "step": 29350
     },
     {
       "epoch": 3.169811320754717,
-      "grad_norm": 0.6106262803077698,
-      "learning_rate": 0.0004103313545601726,
-      "loss": 3.4448,
+      "grad_norm": 0.6538941860198975,
+      "learning_rate": 0.00041032487857528327,
+      "loss": 3.4451,
       "step": 29400
     },
     {
       "epoch": 3.175202156334232,
-      "grad_norm": 0.6202868223190308,
-      "learning_rate": 0.0004100075553157042,
-      "loss": 3.4313,
+      "grad_norm": 0.6297999024391174,
+      "learning_rate": 0.0004100010793308148,
+      "loss": 3.4287,
       "step": 29450
     },
     {
       "epoch": 3.1805929919137466,
-      "grad_norm": 0.5957272052764893,
-      "learning_rate": 0.00040968375607123577,
-      "loss": 3.4549,
+      "grad_norm": 0.6178961396217346,
+      "learning_rate": 0.0004096772800863464,
+      "loss": 3.4557,
       "step": 29500
     },
     {
       "epoch": 3.1859838274932613,
-      "grad_norm": 0.6086465716362,
-      "learning_rate": 0.0004093599568267674,
-      "loss": 3.4405,
+      "grad_norm": 0.582787275314331,
+      "learning_rate": 0.000409353480841878,
+      "loss": 3.4397,
       "step": 29550
     },
     {
       "epoch": 3.1913746630727764,
-      "grad_norm": 0.5802353024482727,
-      "learning_rate": 0.0004090361575822989,
-      "loss": 3.4496,
+      "grad_norm": 0.5640174150466919,
+      "learning_rate": 0.0004090296815974096,
+      "loss": 3.4464,
       "step": 29600
     },
     {
       "epoch": 3.196765498652291,
-      "grad_norm": 0.5590112805366516,
-      "learning_rate": 0.00040871235833783053,
-      "loss": 3.464,
+      "grad_norm": 0.5874976515769958,
+      "learning_rate": 0.00040870588235294113,
+      "loss": 3.4652,
       "step": 29650
     },
     {
       "epoch": 3.202156334231806,
-      "grad_norm": 0.6418723464012146,
-      "learning_rate": 0.0004083885590933621,
-      "loss": 3.4478,
+      "grad_norm": 0.6412698030471802,
+      "learning_rate": 0.00040838208310847273,
+      "loss": 3.4481,
       "step": 29700
     },
     {
       "epoch": 3.207547169811321,
-      "grad_norm": 0.5875370502471924,
-      "learning_rate": 0.00040806475984889363,
-      "loss": 3.4571,
+      "grad_norm": 0.6078434586524963,
+      "learning_rate": 0.0004080582838640043,
+      "loss": 3.4554,
       "step": 29750
     },
     {
       "epoch": 3.2129380053908356,
-      "grad_norm": 0.6010445952415466,
-      "learning_rate": 0.00040774096060442524,
-      "loss": 3.4676,
+      "grad_norm": 0.6255976557731628,
+      "learning_rate": 0.0004077344846195359,
+      "loss": 3.4665,
       "step": 29800
     },
     {
       "epoch": 3.2183288409703503,
-      "grad_norm": 0.6460816860198975,
-      "learning_rate": 0.0004074171613599568,
-      "loss": 3.4719,
+      "grad_norm": 0.6625264286994934,
+      "learning_rate": 0.00040741068537506744,
+      "loss": 3.4688,
       "step": 29850
     },
     {
       "epoch": 3.223719676549865,
-      "grad_norm": 0.6212185621261597,
-      "learning_rate": 0.0004070933621154884,
-      "loss": 3.4391,
+      "grad_norm": 0.6073241233825684,
+      "learning_rate": 0.000407086886130599,
+      "loss": 3.4371,
       "step": 29900
     },
     {
       "epoch": 3.22911051212938,
-      "grad_norm": 0.5984194278717041,
-      "learning_rate": 0.00040676956287101994,
-      "loss": 3.442,
+      "grad_norm": 0.5774091482162476,
+      "learning_rate": 0.0004067630868861306,
+      "loss": 3.4395,
       "step": 29950
     },
     {
       "epoch": 3.234501347708895,
-      "grad_norm": 0.6364345550537109,
-      "learning_rate": 0.00040644576362655155,
-      "loss": 3.4658,
+      "grad_norm": 0.5973829030990601,
+      "learning_rate": 0.0004064392876416621,
+      "loss": 3.4633,
       "step": 30000
     },
     {
       "epoch": 3.234501347708895,
-      "eval_accuracy": 0.3722194775295506,
-      "eval_loss": 3.4868123531341553,
-      "eval_runtime": 184.808,
-      "eval_samples_per_second": 97.458,
-      "eval_steps_per_second": 6.093,
+      "eval_accuracy": 0.37187917668491566,
+      "eval_loss": 3.487455368041992,
+      "eval_runtime": 185.0825,
+      "eval_samples_per_second": 97.313,
+      "eval_steps_per_second": 6.084,
       "step": 30000
     },
     {
       "epoch": 3.2398921832884096,
-      "grad_norm": 0.6432105898857117,
-      "learning_rate": 0.00040612196438208304,
-      "loss": 3.4541,
+      "grad_norm": 0.6407408714294434,
+      "learning_rate": 0.0004061154883971937,
+      "loss": 3.451,
       "step": 30050
     },
     {
       "epoch": 3.2452830188679247,
-      "grad_norm": 0.5993592739105225,
-      "learning_rate": 0.00040579816513761465,
-      "loss": 3.4567,
+      "grad_norm": 0.6154466271400452,
+      "learning_rate": 0.00040579168915272525,
+      "loss": 3.4558,
       "step": 30100
     },
     {
       "epoch": 3.2506738544474394,
-      "grad_norm": 0.6198611259460449,
-      "learning_rate": 0.0004054743658931462,
-      "loss": 3.4649,
+      "grad_norm": 0.5675138831138611,
+      "learning_rate": 0.00040546788990825685,
+      "loss": 3.4632,
       "step": 30150
     },
     {
       "epoch": 3.256064690026954,
-      "grad_norm": 0.5803235769271851,
-      "learning_rate": 0.00040515056664867775,
-      "loss": 3.4479,
+      "grad_norm": 0.6031027436256409,
+      "learning_rate": 0.0004051440906637884,
+      "loss": 3.4464,
       "step": 30200
     },
     {
       "epoch": 3.2614555256064692,
-      "grad_norm": 0.6663078665733337,
-      "learning_rate": 0.00040482676740420935,
-      "loss": 3.4527,
+      "grad_norm": 0.6430271863937378,
+      "learning_rate": 0.00040482029141931995,
+      "loss": 3.4514,
       "step": 30250
     },
     {
       "epoch": 3.266846361185984,
-      "grad_norm": 0.6873801350593567,
-      "learning_rate": 0.0004045029681597409,
-      "loss": 3.4399,
+      "grad_norm": 0.6006436944007874,
+      "learning_rate": 0.00040449649217485156,
+      "loss": 3.4387,
       "step": 30300
     },
     {
       "epoch": 3.2722371967654986,
-      "grad_norm": 0.6202357411384583,
-      "learning_rate": 0.00040418564490016186,
-      "loss": 3.4526,
+      "grad_norm": 0.6477491855621338,
+      "learning_rate": 0.0004041726929303831,
+      "loss": 3.4535,
       "step": 30350
     },
     {
       "epoch": 3.2776280323450133,
-      "grad_norm": 0.5937652587890625,
-      "learning_rate": 0.00040386184565569346,
-      "loss": 3.4651,
+      "grad_norm": 0.633296549320221,
+      "learning_rate": 0.0004038488936859147,
+      "loss": 3.4625,
       "step": 30400
     },
     {
       "epoch": 3.2830188679245285,
-      "grad_norm": 0.586410641670227,
-      "learning_rate": 0.000403538046411225,
-      "loss": 3.4547,
+      "grad_norm": 0.6047157049179077,
+      "learning_rate": 0.00040352509444144626,
+      "loss": 3.4539,
       "step": 30450
     },
     {
       "epoch": 3.288409703504043,
-      "grad_norm": 0.5957598090171814,
-      "learning_rate": 0.00040321424716675656,
-      "loss": 3.4752,
+      "grad_norm": 0.6030910015106201,
+      "learning_rate": 0.00040320129519697787,
+      "loss": 3.4739,
       "step": 30500
     },
     {
       "epoch": 3.293800539083558,
-      "grad_norm": 0.5925768613815308,
-      "learning_rate": 0.00040289044792228817,
-      "loss": 3.466,
+      "grad_norm": 0.5749094486236572,
+      "learning_rate": 0.0004028774959525094,
+      "loss": 3.4636,
       "step": 30550
     },
     {
       "epoch": 3.2991913746630726,
-      "grad_norm": 0.590333878993988,
-      "learning_rate": 0.0004025666486778197,
-      "loss": 3.4576,
+      "grad_norm": 0.6478707790374756,
+      "learning_rate": 0.000402553696708041,
+      "loss": 3.4597,
       "step": 30600
     },
     {
       "epoch": 3.3045822102425877,
-      "grad_norm": 0.5901454091072083,
-      "learning_rate": 0.0004022428494333513,
-      "loss": 3.4574,
+      "grad_norm": 0.6037589311599731,
+      "learning_rate": 0.0004022298974635726,
+      "loss": 3.4564,
       "step": 30650
     },
     {
       "epoch": 3.3099730458221024,
-      "grad_norm": 0.6296274065971375,
-      "learning_rate": 0.0004019190501888828,
-      "loss": 3.4482,
+      "grad_norm": 0.6136059761047363,
+      "learning_rate": 0.00040190609821910407,
+      "loss": 3.4472,
       "step": 30700
     },
     {
       "epoch": 3.315363881401617,
-      "grad_norm": 0.6582797765731812,
-      "learning_rate": 0.0004015952509444144,
-      "loss": 3.4542,
+      "grad_norm": 0.5998915433883667,
+      "learning_rate": 0.0004015822989746357,
+      "loss": 3.4531,
       "step": 30750
     },
     {
       "epoch": 3.3207547169811322,
-      "grad_norm": 0.6020658016204834,
-      "learning_rate": 0.000401271451699946,
-      "loss": 3.4564,
+      "grad_norm": 0.6182710528373718,
+      "learning_rate": 0.0004012584997301672,
+      "loss": 3.4565,
       "step": 30800
     },
     {
       "epoch": 3.326145552560647,
-      "grad_norm": 0.5704158544540405,
-      "learning_rate": 0.0004009476524554776,
-      "loss": 3.4486,
+      "grad_norm": 0.5568892359733582,
+      "learning_rate": 0.00040093470048569883,
+      "loss": 3.4465,
       "step": 30850
     },
     {
       "epoch": 3.3315363881401616,
-      "grad_norm": 0.6447783708572388,
-      "learning_rate": 0.0004006303291958985,
-      "loss": 3.4704,
+      "grad_norm": 0.6910562515258789,
+      "learning_rate": 0.0004006173772261198,
+      "loss": 3.4696,
       "step": 30900
     },
     {
       "epoch": 3.3369272237196768,
-      "grad_norm": 0.5830173492431641,
-      "learning_rate": 0.0004003065299514301,
-      "loss": 3.4526,
+      "grad_norm": 0.5722136497497559,
+      "learning_rate": 0.00040029357798165133,
+      "loss": 3.451,
       "step": 30950
     },
     {
       "epoch": 3.3423180592991915,
-      "grad_norm": 0.5701765418052673,
-      "learning_rate": 0.00039998273070696163,
-      "loss": 3.4715,
+      "grad_norm": 0.565493106842041,
+      "learning_rate": 0.00039996977873718294,
+      "loss": 3.4709,
       "step": 31000
     },
     {
       "epoch": 3.3423180592991915,
-      "eval_accuracy": 0.3730165551145576,
-      "eval_loss": 3.481886148452759,
-      "eval_runtime": 185.2338,
-      "eval_samples_per_second": 97.234,
-      "eval_steps_per_second": 6.079,
+      "eval_accuracy": 0.37296038157155753,
+      "eval_loss": 3.4795143604278564,
+      "eval_runtime": 185.0973,
+      "eval_samples_per_second": 97.306,
+      "eval_steps_per_second": 6.083,
       "step": 31000
     },
     {
       "epoch": 3.347708894878706,
-      "grad_norm": 0.6165177822113037,
-      "learning_rate": 0.00039965893146249324,
-      "loss": 3.4472,
+      "grad_norm": 0.6347475647926331,
+      "learning_rate": 0.0003996459794927145,
+      "loss": 3.4439,
       "step": 31050
     },
     {
       "epoch": 3.353099730458221,
-      "grad_norm": 0.6259761452674866,
-      "learning_rate": 0.0003993351322180248,
-      "loss": 3.466,
+      "grad_norm": 0.5855362415313721,
+      "learning_rate": 0.00039932218024824604,
+      "loss": 3.4663,
       "step": 31100
     },
     {
       "epoch": 3.358490566037736,
-      "grad_norm": 0.6147885918617249,
-      "learning_rate": 0.0003990113329735564,
-      "loss": 3.447,
+      "grad_norm": 0.625121533870697,
+      "learning_rate": 0.00039899838100377764,
+      "loss": 3.4442,
       "step": 31150
     },
     {
       "epoch": 3.3638814016172507,
-      "grad_norm": 0.5740400552749634,
-      "learning_rate": 0.00039868753372908794,
-      "loss": 3.4492,
+      "grad_norm": 0.5701158046722412,
+      "learning_rate": 0.0003986745817593092,
+      "loss": 3.4462,
       "step": 31200
     },
     {
       "epoch": 3.3692722371967654,
-      "grad_norm": 0.6359267830848694,
-      "learning_rate": 0.0003983637344846195,
-      "loss": 3.4654,
+      "grad_norm": 0.6462088227272034,
+      "learning_rate": 0.0003983507825148408,
+      "loss": 3.4639,
       "step": 31250
     },
     {
       "epoch": 3.37466307277628,
-      "grad_norm": 0.5735960006713867,
-      "learning_rate": 0.0003980399352401511,
-      "loss": 3.4509,
+      "grad_norm": 0.5862804055213928,
+      "learning_rate": 0.00039802698327037235,
+      "loss": 3.4482,
       "step": 31300
     },
     {
       "epoch": 3.3800539083557952,
-      "grad_norm": 0.6017249226570129,
-      "learning_rate": 0.0003977161359956826,
-      "loss": 3.4392,
+      "grad_norm": 0.592103123664856,
+      "learning_rate": 0.00039770318402590396,
+      "loss": 3.436,
       "step": 31350
     },
     {
       "epoch": 3.38544474393531,
-      "grad_norm": 0.629580020904541,
-      "learning_rate": 0.0003973923367512142,
-      "loss": 3.4468,
+      "grad_norm": 0.5994466543197632,
+      "learning_rate": 0.00039737938478143545,
+      "loss": 3.4432,
       "step": 31400
     },
     {
       "epoch": 3.3908355795148246,
-      "grad_norm": 0.5926845669746399,
-      "learning_rate": 0.00039706853750674575,
-      "loss": 3.4644,
+      "grad_norm": 0.5867735743522644,
+      "learning_rate": 0.000397055585536967,
+      "loss": 3.4651,
       "step": 31450
     },
     {
       "epoch": 3.3962264150943398,
-      "grad_norm": 0.6186175346374512,
-      "learning_rate": 0.00039674473826227736,
-      "loss": 3.4646,
+      "grad_norm": 0.5661984086036682,
+      "learning_rate": 0.0003967317862924986,
+      "loss": 3.4628,
       "step": 31500
     },
     {
       "epoch": 3.4016172506738545,
-      "grad_norm": 0.6381295919418335,
-      "learning_rate": 0.0003964209390178089,
-      "loss": 3.4516,
+      "grad_norm": 0.6031181216239929,
+      "learning_rate": 0.00039640798704803016,
+      "loss": 3.4507,
       "step": 31550
     },
     {
       "epoch": 3.407008086253369,
-      "grad_norm": 0.6201534271240234,
-      "learning_rate": 0.0003960971397733405,
-      "loss": 3.4373,
+      "grad_norm": 0.6224333047866821,
+      "learning_rate": 0.00039608418780356176,
+      "loss": 3.4377,
       "step": 31600
     },
     {
       "epoch": 3.4123989218328843,
-      "grad_norm": 0.6230754256248474,
-      "learning_rate": 0.00039577334052887206,
-      "loss": 3.4583,
+      "grad_norm": 0.6393580436706543,
+      "learning_rate": 0.0003957603885590933,
+      "loss": 3.4576,
       "step": 31650
     },
     {
       "epoch": 3.417789757412399,
-      "grad_norm": 0.6434163451194763,
-      "learning_rate": 0.0003954495412844036,
-      "loss": 3.4661,
+      "grad_norm": 0.6339957118034363,
+      "learning_rate": 0.0003954365893146249,
+      "loss": 3.4621,
       "step": 31700
     },
     {
       "epoch": 3.4231805929919137,
-      "grad_norm": 0.6090736389160156,
-      "learning_rate": 0.0003951257420399352,
-      "loss": 3.4617,
+      "grad_norm": 0.5791376829147339,
+      "learning_rate": 0.00039511279007015647,
+      "loss": 3.46,
       "step": 31750
     },
     {
       "epoch": 3.4285714285714284,
-      "grad_norm": 0.72145015001297,
-      "learning_rate": 0.00039480194279546677,
-      "loss": 3.4469,
+      "grad_norm": 0.6502759456634521,
+      "learning_rate": 0.00039478899082568807,
+      "loss": 3.4458,
       "step": 31800
     },
     {
       "epoch": 3.4339622641509435,
-      "grad_norm": 0.6008581519126892,
-      "learning_rate": 0.00039447814355099837,
-      "loss": 3.4511,
+      "grad_norm": 0.6216784715652466,
+      "learning_rate": 0.0003944651915812196,
+      "loss": 3.4494,
       "step": 31850
     },
     {
       "epoch": 3.439353099730458,
-      "grad_norm": 0.6144813895225525,
-      "learning_rate": 0.0003941543443065299,
-      "loss": 3.4622,
+      "grad_norm": 0.5869659185409546,
+      "learning_rate": 0.0003941413923367512,
+      "loss": 3.4608,
       "step": 31900
     },
     {
       "epoch": 3.444743935309973,
-      "grad_norm": 0.5947038531303406,
-      "learning_rate": 0.00039383054506206153,
-      "loss": 3.4469,
+      "grad_norm": 0.5932062864303589,
+      "learning_rate": 0.0003938175930922828,
+      "loss": 3.4443,
       "step": 31950
     },
     {
       "epoch": 3.450134770889488,
-      "grad_norm": 0.580974280834198,
-      "learning_rate": 0.000393506745817593,
-      "loss": 3.4594,
+      "grad_norm": 0.6314504742622375,
+      "learning_rate": 0.0003934937938478143,
+      "loss": 3.457,
       "step": 32000
     },
     {
       "epoch": 3.450134770889488,
-      "eval_accuracy": 0.3735301573152567,
-      "eval_loss": 3.4732370376586914,
-      "eval_runtime": 185.0385,
-      "eval_samples_per_second": 97.337,
-      "eval_steps_per_second": 6.085,
+      "eval_accuracy": 0.3736922674238759,
+      "eval_loss": 3.472475051879883,
+      "eval_runtime": 185.2489,
+      "eval_samples_per_second": 97.226,
+      "eval_steps_per_second": 6.078,
       "step": 32000
     },
     {
       "epoch": 3.4555256064690028,
-      "grad_norm": 0.6096420884132385,
-      "learning_rate": 0.0003931829465731247,
-      "loss": 3.4827,
+      "grad_norm": 0.5765188336372375,
+      "learning_rate": 0.00039317647058823523,
+      "loss": 3.4814,
       "step": 32050
     },
     {
       "epoch": 3.4609164420485174,
-      "grad_norm": 0.6233165264129639,
-      "learning_rate": 0.0003928591473286562,
-      "loss": 3.4725,
+      "grad_norm": 0.5741029381752014,
+      "learning_rate": 0.00039285267134376683,
+      "loss": 3.469,
       "step": 32100
     },
     {
       "epoch": 3.466307277628032,
-      "grad_norm": 0.6157932281494141,
-      "learning_rate": 0.00039253534808418773,
-      "loss": 3.444,
+      "grad_norm": 0.6119958758354187,
+      "learning_rate": 0.0003925288720992984,
+      "loss": 3.4415,
       "step": 32150
     },
     {
       "epoch": 3.4716981132075473,
-      "grad_norm": 0.5871294140815735,
-      "learning_rate": 0.00039221154883971933,
-      "loss": 3.4577,
+      "grad_norm": 0.5945273637771606,
+      "learning_rate": 0.00039220507285482993,
+      "loss": 3.4564,
       "step": 32200
     },
     {
       "epoch": 3.477088948787062,
-      "grad_norm": 0.6096975207328796,
-      "learning_rate": 0.0003918877495952509,
-      "loss": 3.4479,
+      "grad_norm": 0.6451990604400635,
+      "learning_rate": 0.00039188127361036154,
+      "loss": 3.4472,
       "step": 32250
     },
     {
       "epoch": 3.4824797843665767,
-      "grad_norm": 0.6300262212753296,
-      "learning_rate": 0.0003915639503507825,
-      "loss": 3.4509,
+      "grad_norm": 0.6479616761207581,
+      "learning_rate": 0.0003915574743658931,
+      "loss": 3.4494,
       "step": 32300
     },
     {
       "epoch": 3.487870619946092,
-      "grad_norm": 0.6068784594535828,
-      "learning_rate": 0.00039124015110631404,
-      "loss": 3.4435,
+      "grad_norm": 0.6039360761642456,
+      "learning_rate": 0.0003912336751214247,
+      "loss": 3.4409,
       "step": 32350
     },
     {
       "epoch": 3.4932614555256065,
-      "grad_norm": 0.652286946773529,
-      "learning_rate": 0.00039091635186184565,
-      "loss": 3.4495,
+      "grad_norm": 0.6061731576919556,
+      "learning_rate": 0.00039090987587695624,
+      "loss": 3.4462,
       "step": 32400
     },
     {
       "epoch": 3.498652291105121,
-      "grad_norm": 0.580116868019104,
-      "learning_rate": 0.0003905925526173772,
-      "loss": 3.4604,
+      "grad_norm": 0.5710239410400391,
+      "learning_rate": 0.00039058607663248785,
+      "loss": 3.4558,
       "step": 32450
     },
     {
       "epoch": 3.5040431266846364,
-      "grad_norm": 0.6080635190010071,
-      "learning_rate": 0.0003902687533729088,
-      "loss": 3.4637,
+      "grad_norm": 0.5853672623634338,
+      "learning_rate": 0.0003902622773880194,
+      "loss": 3.4638,
       "step": 32500
     },
     {
       "epoch": 3.509433962264151,
-      "grad_norm": 0.6209675073623657,
-      "learning_rate": 0.00038994495412844035,
-      "loss": 3.4659,
+      "grad_norm": 0.6254067420959473,
+      "learning_rate": 0.000389938478143551,
+      "loss": 3.4625,
       "step": 32550
     },
     {
       "epoch": 3.5148247978436657,
-      "grad_norm": 0.6277631521224976,
-      "learning_rate": 0.0003896211548839719,
-      "loss": 3.4535,
+      "grad_norm": 0.6255186796188354,
+      "learning_rate": 0.00038961467889908255,
+      "loss": 3.452,
       "step": 32600
     },
     {
       "epoch": 3.5202156334231804,
-      "grad_norm": 0.5931004285812378,
-      "learning_rate": 0.0003892973556395035,
-      "loss": 3.457,
+      "grad_norm": 0.5967075228691101,
+      "learning_rate": 0.00038929087965461405,
+      "loss": 3.4556,
       "step": 32650
     },
     {
       "epoch": 3.525606469002695,
-      "grad_norm": 0.6000095009803772,
-      "learning_rate": 0.000388973556395035,
-      "loss": 3.4726,
+      "grad_norm": 0.58425372838974,
+      "learning_rate": 0.00038896708041014566,
+      "loss": 3.4709,
       "step": 32700
     },
     {
       "epoch": 3.5309973045822103,
-      "grad_norm": 0.6495856046676636,
-      "learning_rate": 0.0003886497571505666,
-      "loss": 3.4472,
+      "grad_norm": 0.611172080039978,
+      "learning_rate": 0.0003886432811656772,
+      "loss": 3.4462,
       "step": 32750
     },
     {
       "epoch": 3.536388140161725,
-      "grad_norm": 0.6353185176849365,
-      "learning_rate": 0.00038832595790609816,
-      "loss": 3.4751,
+      "grad_norm": 0.6028887033462524,
+      "learning_rate": 0.0003883194819212088,
+      "loss": 3.473,
       "step": 32800
     },
     {
       "epoch": 3.5417789757412397,
-      "grad_norm": 0.5833432674407959,
-      "learning_rate": 0.00038800215866162976,
-      "loss": 3.475,
+      "grad_norm": 0.5676169991493225,
+      "learning_rate": 0.00038799568267674036,
+      "loss": 3.4729,
       "step": 32850
     },
     {
       "epoch": 3.547169811320755,
-      "grad_norm": 0.6292868256568909,
-      "learning_rate": 0.0003876783594171613,
-      "loss": 3.4682,
+      "grad_norm": 0.573436975479126,
+      "learning_rate": 0.00038767188343227197,
+      "loss": 3.4675,
       "step": 32900
     },
     {
       "epoch": 3.5525606469002695,
-      "grad_norm": 0.6485346555709839,
-      "learning_rate": 0.0003873545601726929,
-      "loss": 3.4554,
+      "grad_norm": 0.6455245018005371,
+      "learning_rate": 0.0003873480841878035,
+      "loss": 3.454,
       "step": 32950
     },
     {
       "epoch": 3.557951482479784,
-      "grad_norm": 0.63539057970047,
-      "learning_rate": 0.00038703076092822447,
-      "loss": 3.452,
+      "grad_norm": 0.6214224696159363,
+      "learning_rate": 0.0003870242849433351,
+      "loss": 3.4501,
       "step": 33000
     },
     {
       "epoch": 3.557951482479784,
-      "eval_accuracy": 0.3738674158790328,
-      "eval_loss": 3.4682674407958984,
-      "eval_runtime": 184.927,
-      "eval_samples_per_second": 97.395,
-      "eval_steps_per_second": 6.089,
+      "eval_accuracy": 0.37362457667476745,
+      "eval_loss": 3.466701030731201,
+      "eval_runtime": 185.1362,
+      "eval_samples_per_second": 97.285,
+      "eval_steps_per_second": 6.082,
       "step": 33000
     },
     {
       "epoch": 3.5633423180592994,
-      "grad_norm": 0.5895044803619385,
-      "learning_rate": 0.000386706961683756,
-      "loss": 3.4616,
+      "grad_norm": 0.6276306509971619,
+      "learning_rate": 0.00038670048569886667,
+      "loss": 3.4599,
       "step": 33050
     },
     {
       "epoch": 3.568733153638814,
-      "grad_norm": 0.597101092338562,
-      "learning_rate": 0.00038638963842417697,
-      "loss": 3.4612,
+      "grad_norm": 0.6255683898925781,
+      "learning_rate": 0.0003863766864543982,
+      "loss": 3.463,
       "step": 33100
     },
     {
       "epoch": 3.5741239892183287,
-      "grad_norm": 0.6554183959960938,
-      "learning_rate": 0.0003860658391797086,
-      "loss": 3.4665,
+      "grad_norm": 0.5910325050354004,
+      "learning_rate": 0.00038605288720992983,
+      "loss": 3.4655,
       "step": 33150
     },
     {
       "epoch": 3.579514824797844,
-      "grad_norm": 0.6352536678314209,
-      "learning_rate": 0.0003857420399352401,
-      "loss": 3.4622,
+      "grad_norm": 0.6093518137931824,
+      "learning_rate": 0.0003857290879654614,
+      "loss": 3.4621,
       "step": 33200
     },
     {
       "epoch": 3.5849056603773586,
-      "grad_norm": 0.6003103256225586,
-      "learning_rate": 0.00038541824069077173,
-      "loss": 3.4461,
+      "grad_norm": 0.6162254214286804,
+      "learning_rate": 0.000385405288720993,
+      "loss": 3.4442,
       "step": 33250
     },
     {
       "epoch": 3.5902964959568733,
-      "grad_norm": 0.6177358031272888,
-      "learning_rate": 0.0003850944414463033,
-      "loss": 3.4733,
+      "grad_norm": 0.6433115005493164,
+      "learning_rate": 0.00038508148947652453,
+      "loss": 3.4724,
       "step": 33300
     },
     {
       "epoch": 3.595687331536388,
-      "grad_norm": 0.6115977764129639,
-      "learning_rate": 0.0003847706422018348,
-      "loss": 3.4604,
+      "grad_norm": 0.616127073764801,
+      "learning_rate": 0.00038475769023205614,
+      "loss": 3.4594,
       "step": 33350
     },
     {
       "epoch": 3.601078167115903,
-      "grad_norm": 0.6683585047721863,
-      "learning_rate": 0.0003844468429573664,
-      "loss": 3.4566,
+      "grad_norm": 0.6631184220314026,
+      "learning_rate": 0.00038443389098758763,
+      "loss": 3.4554,
       "step": 33400
     },
     {
       "epoch": 3.606469002695418,
-      "grad_norm": 0.6268858909606934,
-      "learning_rate": 0.00038412304371289793,
-      "loss": 3.475,
+      "grad_norm": 0.6478034853935242,
+      "learning_rate": 0.00038411009174311924,
+      "loss": 3.473,
       "step": 33450
     },
     {
       "epoch": 3.6118598382749325,
-      "grad_norm": 0.56812584400177,
-      "learning_rate": 0.00038379924446842954,
-      "loss": 3.4508,
+      "grad_norm": 0.57593834400177,
+      "learning_rate": 0.0003837862924986508,
+      "loss": 3.4495,
       "step": 33500
     },
     {
       "epoch": 3.617250673854447,
-      "grad_norm": 0.6435724496841431,
-      "learning_rate": 0.0003834754452239611,
-      "loss": 3.449,
+      "grad_norm": 0.5845508575439453,
+      "learning_rate": 0.00038346249325418234,
+      "loss": 3.4489,
       "step": 33550
     },
     {
       "epoch": 3.6226415094339623,
-      "grad_norm": 0.6608956456184387,
-      "learning_rate": 0.0003831516459794927,
-      "loss": 3.4437,
+      "grad_norm": 0.6428506970405579,
+      "learning_rate": 0.00038313869400971395,
+      "loss": 3.4434,
       "step": 33600
     },
     {
       "epoch": 3.628032345013477,
-      "grad_norm": 0.6201449632644653,
-      "learning_rate": 0.00038282784673502424,
-      "loss": 3.4597,
+      "grad_norm": 0.6328097581863403,
+      "learning_rate": 0.0003828213707501349,
+      "loss": 3.4675,
       "step": 33650
     },
     {
       "epoch": 3.6334231805929917,
-      "grad_norm": 0.6143207550048828,
-      "learning_rate": 0.00038250404749055585,
-      "loss": 3.4646,
+      "grad_norm": 0.6197866201400757,
+      "learning_rate": 0.00038249757150566645,
+      "loss": 3.4638,
       "step": 33700
     },
     {
       "epoch": 3.638814016172507,
-      "grad_norm": 0.6062964200973511,
-      "learning_rate": 0.0003821802482460874,
-      "loss": 3.4443,
+      "grad_norm": 0.6091422438621521,
+      "learning_rate": 0.00038217377226119805,
+      "loss": 3.4442,
       "step": 33750
     },
     {
       "epoch": 3.6442048517520216,
-      "grad_norm": 0.6019686460494995,
-      "learning_rate": 0.00038185644900161895,
-      "loss": 3.4499,
+      "grad_norm": 0.6277755498886108,
+      "learning_rate": 0.0003818499730167296,
+      "loss": 3.449,
       "step": 33800
     },
     {
       "epoch": 3.6495956873315363,
-      "grad_norm": 0.5941645503044128,
-      "learning_rate": 0.00038153264975715056,
-      "loss": 3.434,
+      "grad_norm": 0.5895088315010071,
+      "learning_rate": 0.00038152617377226115,
+      "loss": 3.4322,
       "step": 33850
     },
     {
       "epoch": 3.6549865229110514,
-      "grad_norm": 0.6274322867393494,
-      "learning_rate": 0.0003812088505126821,
-      "loss": 3.4256,
+      "grad_norm": 0.6150822043418884,
+      "learning_rate": 0.00038120237452779276,
+      "loss": 3.4265,
       "step": 33900
     },
     {
       "epoch": 3.660377358490566,
-      "grad_norm": 0.6215175986289978,
-      "learning_rate": 0.0003808850512682137,
-      "loss": 3.4534,
+      "grad_norm": 0.6281339526176453,
+      "learning_rate": 0.0003808785752833243,
+      "loss": 3.4533,
       "step": 33950
     },
     {
       "epoch": 3.665768194070081,
-      "grad_norm": 0.6265103816986084,
-      "learning_rate": 0.0003805612520237452,
-      "loss": 3.4498,
+      "grad_norm": 0.6197679042816162,
+      "learning_rate": 0.0003805547760388559,
+      "loss": 3.4477,
       "step": 34000
     },
     {
       "epoch": 3.665768194070081,
-      "eval_accuracy": 0.3748275814486494,
-      "eval_loss": 3.4626383781433105,
-      "eval_runtime": 185.0442,
-      "eval_samples_per_second": 97.334,
-      "eval_steps_per_second": 6.085,
+      "eval_accuracy": 0.3746061468632835,
+      "eval_loss": 3.461383104324341,
+      "eval_runtime": 185.1927,
+      "eval_samples_per_second": 97.255,
+      "eval_steps_per_second": 6.08,
       "step": 34000
     },
     {
       "epoch": 3.671159029649596,
-      "grad_norm": 0.6977486610412598,
-      "learning_rate": 0.00038023745277927687,
+      "grad_norm": 0.6397350430488586,
+      "learning_rate": 0.0003802309767943874,
       "loss": 3.4512,
       "step": 34050
     },
     {
       "epoch": 3.6765498652291106,
-      "grad_norm": 0.6424899697303772,
-      "learning_rate": 0.00037991365353480836,
-      "loss": 3.4514,
+      "grad_norm": 0.6090516448020935,
+      "learning_rate": 0.000379907177549919,
+      "loss": 3.4506,
       "step": 34100
     },
     {
       "epoch": 3.6819407008086253,
-      "grad_norm": 0.591140627861023,
-      "learning_rate": 0.0003795898542903399,
-      "loss": 3.4467,
+      "grad_norm": 0.6119341254234314,
+      "learning_rate": 0.00037958337830545057,
+      "loss": 3.4448,
       "step": 34150
     },
     {
       "epoch": 3.68733153638814,
-      "grad_norm": 0.6942396759986877,
-      "learning_rate": 0.0003792660550458715,
-      "loss": 3.4584,
+      "grad_norm": 0.6494148969650269,
+      "learning_rate": 0.00037925957906098217,
+      "loss": 3.4572,
       "step": 34200
     },
     {
       "epoch": 3.6927223719676547,
-      "grad_norm": 0.6438764333724976,
-      "learning_rate": 0.00037894225580140307,
-      "loss": 3.4478,
+      "grad_norm": 0.6527368426322937,
+      "learning_rate": 0.0003789357798165137,
+      "loss": 3.4459,
       "step": 34250
     },
     {
       "epoch": 3.69811320754717,
-      "grad_norm": 0.6779077649116516,
-      "learning_rate": 0.00037861845655693467,
-      "loss": 3.4519,
+      "grad_norm": 0.6111875772476196,
+      "learning_rate": 0.00037861198057204527,
+      "loss": 3.4508,
       "step": 34300
     },
     {
       "epoch": 3.7035040431266846,
-      "grad_norm": 0.6349784135818481,
-      "learning_rate": 0.0003782946573124662,
-      "loss": 3.4495,
+      "grad_norm": 0.6030789017677307,
+      "learning_rate": 0.0003782881813275769,
+      "loss": 3.4492,
       "step": 34350
     },
     {
       "epoch": 3.7088948787061993,
-      "grad_norm": 0.6402055025100708,
-      "learning_rate": 0.00037797085806799783,
-      "loss": 3.4582,
+      "grad_norm": 0.6276407241821289,
+      "learning_rate": 0.0003779643820831084,
+      "loss": 3.4563,
       "step": 34400
     },
     {
       "epoch": 3.7142857142857144,
-      "grad_norm": 0.6125994920730591,
-      "learning_rate": 0.0003776470588235294,
-      "loss": 3.4278,
+      "grad_norm": 0.6185351014137268,
+      "learning_rate": 0.00037764058283864003,
+      "loss": 3.4279,
       "step": 34450
     },
     {
       "epoch": 3.719676549865229,
-      "grad_norm": 0.658770740032196,
-      "learning_rate": 0.000377323259579061,
-      "loss": 3.4702,
+      "grad_norm": 0.6208236217498779,
+      "learning_rate": 0.0003773167835941716,
+      "loss": 3.4682,
       "step": 34500
     },
     {
       "epoch": 3.725067385444744,
-      "grad_norm": 0.6636422276496887,
-      "learning_rate": 0.00037699946033459253,
-      "loss": 3.448,
+      "grad_norm": 0.6041314601898193,
+      "learning_rate": 0.0003769929843497032,
+      "loss": 3.4465,
       "step": 34550
     },
     {
       "epoch": 3.730458221024259,
-      "grad_norm": 0.6237731575965881,
-      "learning_rate": 0.0003766756610901241,
-      "loss": 3.453,
+      "grad_norm": 0.5724568963050842,
+      "learning_rate": 0.00037666918510523474,
+      "loss": 3.4498,
       "step": 34600
     },
     {
       "epoch": 3.7358490566037736,
-      "grad_norm": 0.7565545439720154,
-      "learning_rate": 0.0003763518618456557,
-      "loss": 3.4553,
+      "grad_norm": 0.5927848815917969,
+      "learning_rate": 0.00037634538586076634,
+      "loss": 3.4556,
       "step": 34650
     },
     {
       "epoch": 3.7412398921832883,
-      "grad_norm": 0.6257597804069519,
-      "learning_rate": 0.0003760280626011872,
-      "loss": 3.4623,
+      "grad_norm": 0.6263810396194458,
+      "learning_rate": 0.00037602158661629784,
+      "loss": 3.46,
       "step": 34700
     },
     {
       "epoch": 3.7466307277628035,
-      "grad_norm": 0.636355996131897,
-      "learning_rate": 0.0003757042633567188,
-      "loss": 3.4671,
+      "grad_norm": 0.6283044815063477,
+      "learning_rate": 0.0003756977873718294,
+      "loss": 3.4649,
       "step": 34750
     },
     {
       "epoch": 3.752021563342318,
-      "grad_norm": 0.6237640976905823,
-      "learning_rate": 0.00037538046411225034,
-      "loss": 3.4505,
+      "grad_norm": 0.5919647812843323,
+      "learning_rate": 0.000375373988127361,
+      "loss": 3.4495,
       "step": 34800
     },
     {
       "epoch": 3.757412398921833,
-      "grad_norm": 0.6493251919746399,
-      "learning_rate": 0.00037505666486778195,
-      "loss": 3.4604,
+      "grad_norm": 0.64276522397995,
+      "learning_rate": 0.00037505018888289254,
+      "loss": 3.4591,
       "step": 34850
     },
     {
       "epoch": 3.7628032345013476,
-      "grad_norm": 0.6321045756340027,
-      "learning_rate": 0.0003747328656233135,
-      "loss": 3.4504,
+      "grad_norm": 0.6236562132835388,
+      "learning_rate": 0.00037472638963842415,
+      "loss": 3.4511,
       "step": 34900
     },
     {
       "epoch": 3.7681940700808623,
-      "grad_norm": 0.6456063389778137,
-      "learning_rate": 0.0003744090663788451,
-      "loss": 3.4446,
+      "grad_norm": 0.645114004611969,
+      "learning_rate": 0.0003744025903939557,
+      "loss": 3.4443,
       "step": 34950
     },
     {
       "epoch": 3.7735849056603774,
-      "grad_norm": 0.7021212577819824,
-      "learning_rate": 0.00037408526713437665,
-      "loss": 3.4671,
+      "grad_norm": 0.6953789591789246,
+      "learning_rate": 0.0003740787911494873,
+      "loss": 3.4652,
       "step": 35000
     },
     {
       "epoch": 3.7735849056603774,
-      "eval_accuracy": 0.3753076099070138,
-      "eval_loss": 3.457233428955078,
-      "eval_runtime": 184.8636,
-      "eval_samples_per_second": 97.429,
-      "eval_steps_per_second": 6.091,
+      "eval_accuracy": 0.3752516536697893,
+      "eval_loss": 3.454463243484497,
+      "eval_runtime": 185.0118,
+      "eval_samples_per_second": 97.351,
+      "eval_steps_per_second": 6.086,
       "step": 35000
     },
     {
       "epoch": 3.778975741239892,
-      "grad_norm": 0.6232941746711731,
-      "learning_rate": 0.0003737614678899082,
-      "loss": 3.4515,
+      "grad_norm": 0.615135669708252,
+      "learning_rate": 0.00037375499190501885,
+      "loss": 3.4494,
       "step": 35050
     },
     {
       "epoch": 3.784366576819407,
-      "grad_norm": 0.6385236978530884,
-      "learning_rate": 0.0003734376686454398,
-      "loss": 3.4544,
+      "grad_norm": 0.6264244914054871,
+      "learning_rate": 0.0003734311926605504,
+      "loss": 3.4529,
       "step": 35100
     },
     {
       "epoch": 3.789757412398922,
-      "grad_norm": 0.6696701049804688,
-      "learning_rate": 0.00037311386940097136,
-      "loss": 3.4392,
+      "grad_norm": 0.6474460959434509,
+      "learning_rate": 0.000373107393416082,
+      "loss": 3.4373,
       "step": 35150
     },
     {
       "epoch": 3.7951482479784366,
-      "grad_norm": 0.6041948795318604,
-      "learning_rate": 0.00037279007015650296,
-      "loss": 3.4452,
+      "grad_norm": 0.5701635479927063,
+      "learning_rate": 0.00037278359417161356,
+      "loss": 3.4437,
       "step": 35200
     },
     {
       "epoch": 3.8005390835579513,
-      "grad_norm": 0.6724011898040771,
-      "learning_rate": 0.0003724662709120345,
-      "loss": 3.453,
+      "grad_norm": 0.6054803133010864,
+      "learning_rate": 0.00037245979492714517,
+      "loss": 3.4508,
       "step": 35250
     },
     {
       "epoch": 3.8059299191374665,
-      "grad_norm": 0.6799139976501465,
-      "learning_rate": 0.0003721424716675661,
-      "loss": 3.4457,
+      "grad_norm": 0.6329179406166077,
+      "learning_rate": 0.0003721359956826767,
+      "loss": 3.4436,
       "step": 35300
     },
     {
       "epoch": 3.811320754716981,
-      "grad_norm": 0.6157107353210449,
-      "learning_rate": 0.0003718186724230976,
-      "loss": 3.4671,
+      "grad_norm": 0.6441768407821655,
+      "learning_rate": 0.0003718121964382083,
+      "loss": 3.4643,
       "step": 35350
     },
     {
       "epoch": 3.816711590296496,
-      "grad_norm": 0.6035497188568115,
-      "learning_rate": 0.0003714948731786293,
-      "loss": 3.4475,
+      "grad_norm": 0.5799574255943298,
+      "learning_rate": 0.0003714883971937398,
+      "loss": 3.4452,
       "step": 35400
     },
     {
       "epoch": 3.822102425876011,
-      "grad_norm": 0.6183246374130249,
-      "learning_rate": 0.00037117107393416077,
-      "loss": 3.4539,
+      "grad_norm": 0.6319050192832947,
+      "learning_rate": 0.0003711645979492714,
+      "loss": 3.4519,
       "step": 35450
     },
     {
       "epoch": 3.8274932614555257,
-      "grad_norm": 0.6753723621368408,
-      "learning_rate": 0.0003708472746896923,
-      "loss": 3.4443,
+      "grad_norm": 0.6138501167297363,
+      "learning_rate": 0.00037084079870480297,
+      "loss": 3.4435,
       "step": 35500
     },
     {
       "epoch": 3.8328840970350404,
-      "grad_norm": 0.6167969107627869,
-      "learning_rate": 0.0003705234754452239,
-      "loss": 3.4475,
+      "grad_norm": 0.6051216721534729,
+      "learning_rate": 0.0003705169994603345,
+      "loss": 3.4481,
       "step": 35550
     },
     {
       "epoch": 3.838274932614555,
-      "grad_norm": 0.6387770175933838,
-      "learning_rate": 0.0003701996762007555,
-      "loss": 3.4591,
+      "grad_norm": 0.6242967844009399,
+      "learning_rate": 0.00037019320021586613,
+      "loss": 3.4572,
       "step": 35600
     },
     {
       "epoch": 3.8436657681940702,
-      "grad_norm": 0.6343386173248291,
-      "learning_rate": 0.0003698758769562871,
-      "loss": 3.4443,
+      "grad_norm": 0.627966582775116,
+      "learning_rate": 0.0003698694009713977,
+      "loss": 3.4424,
       "step": 35650
     },
     {
       "epoch": 3.849056603773585,
-      "grad_norm": 0.6071075201034546,
-      "learning_rate": 0.00036955207771181863,
-      "loss": 3.4436,
+      "grad_norm": 0.5960062146186829,
+      "learning_rate": 0.0003695456017269293,
+      "loss": 3.4426,
       "step": 35700
     },
     {
       "epoch": 3.8544474393530996,
-      "grad_norm": 0.596243143081665,
-      "learning_rate": 0.00036922827846735024,
-      "loss": 3.4529,
+      "grad_norm": 0.5962764024734497,
+      "learning_rate": 0.00036922180248246083,
+      "loss": 3.4543,
       "step": 35750
     },
     {
       "epoch": 3.8598382749326143,
-      "grad_norm": 0.6391444206237793,
-      "learning_rate": 0.0003689044792228818,
-      "loss": 3.4403,
+      "grad_norm": 0.6414475440979004,
+      "learning_rate": 0.00036889800323799244,
+      "loss": 3.4405,
       "step": 35800
     },
     {
       "epoch": 3.8652291105121295,
-      "grad_norm": 0.6206788420677185,
+      "grad_norm": 0.6472159028053284,
       "learning_rate": 0.0003685806799784134,
-      "loss": 3.4506,
+      "loss": 3.4513,
       "step": 35850
     },
     {
       "epoch": 3.870619946091644,
-      "grad_norm": 0.652442455291748,
+      "grad_norm": 0.63205885887146,
       "learning_rate": 0.00036825688073394494,
-      "loss": 3.4626,
+      "loss": 3.4625,
       "step": 35900
     },
     {
       "epoch": 3.876010781671159,
-      "grad_norm": 0.6262994408607483,
+      "grad_norm": 0.6096854209899902,
       "learning_rate": 0.0003679330814894765,
-      "loss": 3.4641,
+      "loss": 3.4624,
       "step": 35950
     },
     {
       "epoch": 3.881401617250674,
-      "grad_norm": 0.6360272765159607,
+      "grad_norm": 0.6115530133247375,
       "learning_rate": 0.0003676092822450081,
-      "loss": 3.4334,
+      "loss": 3.4316,
       "step": 36000
     },
     {
       "epoch": 3.881401617250674,
-      "eval_accuracy": 0.37577340483707444,
-      "eval_loss": 3.450922966003418,
-      "eval_runtime": 185.1357,
-      "eval_samples_per_second": 97.285,
-      "eval_steps_per_second": 6.082,
+      "eval_accuracy": 0.37572016492204535,
+      "eval_loss": 3.448432445526123,
+      "eval_runtime": 185.5015,
+      "eval_samples_per_second": 97.094,
+      "eval_steps_per_second": 6.07,
       "step": 36000
     },
     {
       "epoch": 3.8867924528301887,
-      "grad_norm": 0.6211079359054565,
+      "grad_norm": 0.6192970871925354,
       "learning_rate": 0.0003672854830005396,
-      "loss": 3.47,
+      "loss": 3.4669,
       "step": 36050
     },
     {
       "epoch": 3.8921832884097034,
-      "grad_norm": 0.6081632375717163,
+      "grad_norm": 0.6395145654678345,
       "learning_rate": 0.0003669616837560712,
-      "loss": 3.4335,
+      "loss": 3.4336,
       "step": 36100
     },
     {
       "epoch": 3.8975741239892185,
-      "grad_norm": 0.6606460213661194,
+      "grad_norm": 0.6128035187721252,
       "learning_rate": 0.00036663788451160275,
-      "loss": 3.4427,
+      "loss": 3.4405,
       "step": 36150
     },
     {
       "epoch": 3.9029649595687332,
-      "grad_norm": 0.6085197925567627,
+      "grad_norm": 0.5994486808776855,
       "learning_rate": 0.00036631408526713435,
-      "loss": 3.4481,
+      "loss": 3.4473,
       "step": 36200
     },
     {
       "epoch": 3.908355795148248,
-      "grad_norm": 0.5942933559417725,
+      "grad_norm": 0.598868191242218,
       "learning_rate": 0.0003659902860226659,
-      "loss": 3.4558,
+      "loss": 3.4512,
       "step": 36250
     },
     {
       "epoch": 3.913746630727763,
-      "grad_norm": 0.5989931225776672,
+      "grad_norm": 0.6117761731147766,
       "learning_rate": 0.00036566648677819745,
-      "loss": 3.4572,
+      "loss": 3.453,
       "step": 36300
     },
     {
       "epoch": 3.9191374663072778,
-      "grad_norm": 0.6112847328186035,
+      "grad_norm": 0.6310285329818726,
       "learning_rate": 0.00036534268753372906,
-      "loss": 3.4703,
+      "loss": 3.4659,
       "step": 36350
     },
     {
       "epoch": 3.9245283018867925,
-      "grad_norm": 0.5879990458488464,
+      "grad_norm": 0.5916323065757751,
       "learning_rate": 0.0003650188882892606,
-      "loss": 3.4374,
+      "loss": 3.4365,
       "step": 36400
     },
     {
       "epoch": 3.929919137466307,
-      "grad_norm": 0.6718971729278564,
+      "grad_norm": 0.6302815079689026,
       "learning_rate": 0.0003646950890447922,
-      "loss": 3.4428,
+      "loss": 3.4415,
       "step": 36450
     },
     {
       "epoch": 3.935309973045822,
-      "grad_norm": 0.5937814116477966,
+      "grad_norm": 0.6169242262840271,
       "learning_rate": 0.00036437128980032376,
-      "loss": 3.4565,
+      "loss": 3.4535,
       "step": 36500
     },
     {
       "epoch": 3.940700808625337,
-      "grad_norm": 0.6202378869056702,
+      "grad_norm": 0.6036542057991028,
       "learning_rate": 0.00036404749055585537,
-      "loss": 3.4696,
+      "loss": 3.4697,
       "step": 36550
     },
     {
       "epoch": 3.9460916442048517,
-      "grad_norm": 0.6717408895492554,
+      "grad_norm": 0.6433455944061279,
       "learning_rate": 0.0003637236913113869,
-      "loss": 3.4239,
+      "loss": 3.4215,
       "step": 36600
     },
     {
       "epoch": 3.9514824797843664,
-      "grad_norm": 0.6636652946472168,
+      "grad_norm": 0.6333649754524231,
       "learning_rate": 0.0003633998920669185,
-      "loss": 3.4406,
+      "loss": 3.439,
       "step": 36650
     },
     {
       "epoch": 3.9568733153638815,
-      "grad_norm": 0.6985486149787903,
+      "grad_norm": 0.6479489207267761,
       "learning_rate": 0.0003630760928224501,
-      "loss": 3.4461,
+      "loss": 3.4444,
       "step": 36700
     },
     {
       "epoch": 3.9622641509433962,
-      "grad_norm": 0.639927089214325,
+      "grad_norm": 0.5945091247558594,
       "learning_rate": 0.00036275229357798157,
-      "loss": 3.4282,
+      "loss": 3.4266,
       "step": 36750
     },
     {
       "epoch": 3.967654986522911,
-      "grad_norm": 0.6031926870346069,
+      "grad_norm": 0.5627322793006897,
       "learning_rate": 0.0003624284943335132,
-      "loss": 3.4406,
+      "loss": 3.439,
       "step": 36800
     },
     {
       "epoch": 3.973045822102426,
-      "grad_norm": 0.6068711876869202,
+      "grad_norm": 0.5948078036308289,
       "learning_rate": 0.00036210469508904473,
-      "loss": 3.45,
+      "loss": 3.449,
       "step": 36850
     },
     {
       "epoch": 3.9784366576819408,
-      "grad_norm": 0.6099409461021423,
+      "grad_norm": 0.5916422605514526,
       "learning_rate": 0.00036178089584457633,
-      "loss": 3.4486,
+      "loss": 3.4465,
       "step": 36900
     },
     {
       "epoch": 3.9838274932614555,
-      "grad_norm": 0.6189759373664856,
+      "grad_norm": 0.5898805856704712,
       "learning_rate": 0.0003614570966001079,
-      "loss": 3.4265,
+      "loss": 3.4251,
       "step": 36950
     },
     {
       "epoch": 3.9892183288409706,
-      "grad_norm": 0.6519210338592529,
+      "grad_norm": 0.5998630523681641,
       "learning_rate": 0.0003611332973556395,
-      "loss": 3.416,
+      "loss": 3.4147,
       "step": 37000
     },
     {
       "epoch": 3.9892183288409706,
-      "eval_accuracy": 0.3765111579453348,
-      "eval_loss": 3.4432153701782227,
-      "eval_runtime": 184.8939,
-      "eval_samples_per_second": 97.413,
-      "eval_steps_per_second": 6.09,
+      "eval_accuracy": 0.37643064615546423,
+      "eval_loss": 3.441873073577881,
+      "eval_runtime": 184.939,
+      "eval_samples_per_second": 97.389,
+      "eval_steps_per_second": 6.088,
       "step": 37000
     },
     {
       "epoch": 3.9946091644204853,
-      "grad_norm": 0.6266780495643616,
+      "grad_norm": 0.616155743598938,
       "learning_rate": 0.00036080949811117104,
-      "loss": 3.4672,
+      "loss": 3.4655,
       "step": 37050
     },
     {
       "epoch": 4.0,
-      "grad_norm": 1.3627803325653076,
+      "grad_norm": 1.3538962602615356,
       "learning_rate": 0.00036048569886670264,
-      "loss": 3.4304,
+      "loss": 3.4306,
       "step": 37100
     },
     {
       "epoch": 4.005390835579515,
-      "grad_norm": 0.6425639390945435,
-      "learning_rate": 0.00036016837560712354,
-      "loss": 3.3615,
+      "grad_norm": 0.6368972659111023,
+      "learning_rate": 0.0003601618996222342,
+      "loss": 3.3594,
       "step": 37150
     },
     {
       "epoch": 4.010781671159029,
-      "grad_norm": 0.594615638256073,
-      "learning_rate": 0.00035984457636265515,
-      "loss": 3.3569,
+      "grad_norm": 0.6396125555038452,
+      "learning_rate": 0.00035983810037776574,
+      "loss": 3.3545,
       "step": 37200
     },
     {
       "epoch": 4.0161725067385445,
-      "grad_norm": 0.671310544013977,
-      "learning_rate": 0.0003595207771181867,
-      "loss": 3.3522,
+      "grad_norm": 0.6328780055046082,
+      "learning_rate": 0.00035951430113329735,
+      "loss": 3.3507,
       "step": 37250
     },
     {
       "epoch": 4.02156334231806,
-      "grad_norm": 0.6261488795280457,
-      "learning_rate": 0.0003591969778737183,
-      "loss": 3.357,
+      "grad_norm": 0.5838497877120972,
+      "learning_rate": 0.0003591905018888289,
+      "loss": 3.355,
       "step": 37300
     },
     {
       "epoch": 4.026954177897574,
-      "grad_norm": 0.63107830286026,
-      "learning_rate": 0.00035887317862924985,
-      "loss": 3.3399,
+      "grad_norm": 0.5783930420875549,
+      "learning_rate": 0.0003588667026443605,
+      "loss": 3.336,
       "step": 37350
     },
     {
       "epoch": 4.032345013477089,
-      "grad_norm": 0.6028571128845215,
-      "learning_rate": 0.00035854937938478146,
-      "loss": 3.3433,
+      "grad_norm": 0.597440242767334,
+      "learning_rate": 0.000358542903399892,
+      "loss": 3.3412,
       "step": 37400
     },
     {
       "epoch": 4.037735849056604,
-      "grad_norm": 0.6417591571807861,
-      "learning_rate": 0.00035822558014031295,
-      "loss": 3.353,
+      "grad_norm": 0.6205726861953735,
+      "learning_rate": 0.0003582191041554236,
+      "loss": 3.3508,
       "step": 37450
     },
     {
       "epoch": 4.0431266846361185,
-      "grad_norm": 0.6593953967094421,
-      "learning_rate": 0.0003579017808958445,
-      "loss": 3.3535,
+      "grad_norm": 0.6443513631820679,
+      "learning_rate": 0.00035789530491095516,
+      "loss": 3.3509,
       "step": 37500
     },
     {
       "epoch": 4.048517520215634,
-      "grad_norm": 0.6569228172302246,
-      "learning_rate": 0.0003575779816513761,
-      "loss": 3.3623,
+      "grad_norm": 0.6111722588539124,
+      "learning_rate": 0.00035757150566648676,
+      "loss": 3.3604,
       "step": 37550
     },
     {
       "epoch": 4.053908355795148,
-      "grad_norm": 0.6970022320747375,
-      "learning_rate": 0.00035725418240690766,
-      "loss": 3.357,
+      "grad_norm": 0.6462409496307373,
+      "learning_rate": 0.0003572477064220183,
+      "loss": 3.3548,
       "step": 37600
     },
     {
       "epoch": 4.059299191374663,
-      "grad_norm": 0.652712345123291,
-      "learning_rate": 0.00035693038316243926,
-      "loss": 3.3645,
+      "grad_norm": 0.6218355298042297,
+      "learning_rate": 0.00035692390717754986,
+      "loss": 3.3616,
       "step": 37650
     },
     {
       "epoch": 4.064690026954178,
-      "grad_norm": 0.6411773562431335,
-      "learning_rate": 0.0003566065839179708,
-      "loss": 3.3643,
+      "grad_norm": 0.6028040647506714,
+      "learning_rate": 0.00035660010793308147,
+      "loss": 3.3614,
       "step": 37700
     },
     {
       "epoch": 4.070080862533692,
-      "grad_norm": 0.6141267418861389,
-      "learning_rate": 0.0003562827846735024,
-      "loss": 3.3716,
+      "grad_norm": 0.5973382592201233,
+      "learning_rate": 0.000356276308688613,
+      "loss": 3.3686,
       "step": 37750
     },
     {
       "epoch": 4.0754716981132075,
-      "grad_norm": 0.61728835105896,
-      "learning_rate": 0.00035595898542903397,
-      "loss": 3.362,
+      "grad_norm": 0.5949469804763794,
+      "learning_rate": 0.0003559525094441446,
+      "loss": 3.36,
       "step": 37800
     },
     {
       "epoch": 4.080862533692723,
-      "grad_norm": 0.6423006653785706,
-      "learning_rate": 0.0003556351861845656,
-      "loss": 3.368,
+      "grad_norm": 0.6174900531768799,
+      "learning_rate": 0.00035562871019967617,
+      "loss": 3.3676,
       "step": 37850
     },
     {
       "epoch": 4.086253369272237,
-      "grad_norm": 0.6371752619743347,
-      "learning_rate": 0.0003553113869400971,
-      "loss": 3.3778,
+      "grad_norm": 0.6042484641075134,
+      "learning_rate": 0.0003553049109552078,
+      "loss": 3.3756,
       "step": 37900
     },
     {
       "epoch": 4.091644204851752,
-      "grad_norm": 0.6404188871383667,
-      "learning_rate": 0.0003549875876956287,
-      "loss": 3.3784,
+      "grad_norm": 0.610124409198761,
+      "learning_rate": 0.00035498111171073933,
+      "loss": 3.3738,
       "step": 37950
     },
     {
       "epoch": 4.097035040431267,
-      "grad_norm": 0.6130540370941162,
-      "learning_rate": 0.0003546637884511603,
-      "loss": 3.3571,
+      "grad_norm": 0.6482214331626892,
+      "learning_rate": 0.0003546573124662708,
+      "loss": 3.3564,
       "step": 38000
     },
     {
       "epoch": 4.097035040431267,
-      "eval_accuracy": 0.37693935897621167,
-      "eval_loss": 3.4457244873046875,
-      "eval_runtime": 184.614,
-      "eval_samples_per_second": 97.56,
-      "eval_steps_per_second": 6.099,
+      "eval_accuracy": 0.37664436638579535,
+      "eval_loss": 3.4471352100372314,
+      "eval_runtime": 185.4167,
+      "eval_samples_per_second": 97.138,
+      "eval_steps_per_second": 6.073,
       "step": 38000
     },
     {
       "epoch": 4.1024258760107815,
-      "grad_norm": 0.6331332921981812,
-      "learning_rate": 0.0003543399892066918,
-      "loss": 3.3762,
+      "grad_norm": 0.674815833568573,
+      "learning_rate": 0.0003543335132218025,
+      "loss": 3.3728,
       "step": 38050
     },
     {
       "epoch": 4.107816711590297,
-      "grad_norm": 0.6457241177558899,
-      "learning_rate": 0.0003540161899622234,
-      "loss": 3.3599,
+      "grad_norm": 0.659717857837677,
+      "learning_rate": 0.000354009713977334,
+      "loss": 3.3575,
       "step": 38100
     },
     {
       "epoch": 4.113207547169812,
-      "grad_norm": 0.6348027586936951,
-      "learning_rate": 0.00035369239071775493,
-      "loss": 3.3749,
+      "grad_norm": 0.6627160310745239,
+      "learning_rate": 0.0003536859147328656,
+      "loss": 3.3714,
       "step": 38150
     },
     {
       "epoch": 4.118598382749326,
-      "grad_norm": 0.650617778301239,
-      "learning_rate": 0.00035336859147328654,
-      "loss": 3.3783,
+      "grad_norm": 0.6379284858703613,
+      "learning_rate": 0.00035336211548839713,
+      "loss": 3.3761,
       "step": 38200
     },
     {
       "epoch": 4.123989218328841,
-      "grad_norm": 0.6565114855766296,
-      "learning_rate": 0.0003530447922288181,
-      "loss": 3.366,
+      "grad_norm": 0.610173761844635,
+      "learning_rate": 0.00035303831624392874,
+      "loss": 3.3633,
       "step": 38250
     },
     {
       "epoch": 4.129380053908355,
-      "grad_norm": 0.6230000853538513,
-      "learning_rate": 0.0003527209929843497,
-      "loss": 3.3705,
+      "grad_norm": 0.6131238341331482,
+      "learning_rate": 0.0003527145169994603,
+      "loss": 3.3686,
       "step": 38300
     },
     {
       "epoch": 4.1347708894878705,
-      "grad_norm": 0.6349278092384338,
-      "learning_rate": 0.00035239719373988124,
-      "loss": 3.3536,
+      "grad_norm": 0.6268836259841919,
+      "learning_rate": 0.0003523907177549919,
+      "loss": 3.353,
       "step": 38350
     },
     {
       "epoch": 4.140161725067386,
-      "grad_norm": 0.6567692756652832,
-      "learning_rate": 0.0003520733944954128,
-      "loss": 3.3749,
+      "grad_norm": 0.6279870271682739,
+      "learning_rate": 0.00035206691851052345,
+      "loss": 3.3732,
       "step": 38400
     },
     {
       "epoch": 4.1455525606469,
-      "grad_norm": 0.6406522989273071,
-      "learning_rate": 0.0003517495952509444,
-      "loss": 3.3725,
+      "grad_norm": 0.624947726726532,
+      "learning_rate": 0.000351743119266055,
+      "loss": 3.37,
       "step": 38450
     },
     {
       "epoch": 4.150943396226415,
-      "grad_norm": 0.611667275428772,
-      "learning_rate": 0.00035142579600647595,
-      "loss": 3.3662,
+      "grad_norm": 0.6435061693191528,
+      "learning_rate": 0.0003514193200215866,
+      "loss": 3.3646,
       "step": 38500
     },
     {
       "epoch": 4.15633423180593,
-      "grad_norm": 0.608488917350769,
-      "learning_rate": 0.00035110199676200755,
-      "loss": 3.3649,
+      "grad_norm": 0.6154115796089172,
+      "learning_rate": 0.00035109552077711815,
+      "loss": 3.3634,
       "step": 38550
     },
     {
       "epoch": 4.1617250673854445,
-      "grad_norm": 0.6489611268043518,
-      "learning_rate": 0.0003507781975175391,
-      "loss": 3.3649,
+      "grad_norm": 0.6166911125183105,
+      "learning_rate": 0.00035077172153264976,
+      "loss": 3.3637,
       "step": 38600
     },
     {
       "epoch": 4.16711590296496,
-      "grad_norm": 0.7167544364929199,
-      "learning_rate": 0.0003504543982730707,
-      "loss": 3.3698,
+      "grad_norm": 0.6432587504386902,
+      "learning_rate": 0.0003504479222881813,
+      "loss": 3.3691,
       "step": 38650
     },
     {
       "epoch": 4.172506738544475,
-      "grad_norm": 0.6684973239898682,
-      "learning_rate": 0.00035013059902860226,
-      "loss": 3.3851,
+      "grad_norm": 0.6543565392494202,
+      "learning_rate": 0.0003501241230437129,
+      "loss": 3.3833,
       "step": 38700
     },
     {
       "epoch": 4.177897574123989,
-      "grad_norm": 0.6070584654808044,
-      "learning_rate": 0.00034980679978413375,
-      "loss": 3.3609,
+      "grad_norm": 0.6134627461433411,
+      "learning_rate": 0.0003498003237992444,
+      "loss": 3.3596,
       "step": 38750
     },
     {
       "epoch": 4.183288409703504,
-      "grad_norm": 0.6427111029624939,
-      "learning_rate": 0.00034948300053966536,
-      "loss": 3.3758,
+      "grad_norm": 0.5936211943626404,
+      "learning_rate": 0.000349476524554776,
+      "loss": 3.3745,
       "step": 38800
     },
     {
       "epoch": 4.188679245283019,
-      "grad_norm": 0.6557640433311462,
-      "learning_rate": 0.0003491592012951969,
-      "loss": 3.3859,
+      "grad_norm": 0.617832362651825,
+      "learning_rate": 0.00034915272531030756,
+      "loss": 3.384,
       "step": 38850
     },
     {
       "epoch": 4.1940700808625335,
-      "grad_norm": 0.6033419966697693,
-      "learning_rate": 0.0003488354020507285,
-      "loss": 3.4005,
+      "grad_norm": 0.5889009237289429,
+      "learning_rate": 0.0003488289260658391,
+      "loss": 3.3997,
       "step": 38900
     },
     {
       "epoch": 4.199460916442049,
-      "grad_norm": 0.6868317127227783,
-      "learning_rate": 0.00034851160280626007,
-      "loss": 3.3661,
+      "grad_norm": 0.6840178370475769,
+      "learning_rate": 0.0003485051268213707,
+      "loss": 3.3653,
       "step": 38950
     },
     {
       "epoch": 4.204851752021563,
-      "grad_norm": 0.651969313621521,
-      "learning_rate": 0.00034818780356179167,
-      "loss": 3.3826,
+      "grad_norm": 0.664037823677063,
+      "learning_rate": 0.00034818132757690227,
+      "loss": 3.3819,
       "step": 39000
     },
     {
       "epoch": 4.204851752021563,
-      "eval_accuracy": 0.37738494446913884,
-      "eval_loss": 3.4434092044830322,
-      "eval_runtime": 185.0308,
-      "eval_samples_per_second": 97.341,
-      "eval_steps_per_second": 6.085,
+      "eval_accuracy": 0.377053335855529,
+      "eval_loss": 3.4432272911071777,
+      "eval_runtime": 184.8506,
+      "eval_samples_per_second": 97.435,
+      "eval_steps_per_second": 6.091,
       "step": 39000
     },
     {
       "epoch": 4.210242587601078,
-      "grad_norm": 0.6915486454963684,
-      "learning_rate": 0.0003478640043173232,
-      "loss": 3.386,
+      "grad_norm": 0.6409573554992676,
+      "learning_rate": 0.0003478575283324339,
+      "loss": 3.3856,
       "step": 39050
     },
     {
       "epoch": 4.215633423180593,
-      "grad_norm": 0.639582097530365,
-      "learning_rate": 0.0003475402050728548,
-      "loss": 3.3905,
+      "grad_norm": 0.6263582706451416,
+      "learning_rate": 0.0003475337290879654,
+      "loss": 3.3883,
       "step": 39100
     },
     {
       "epoch": 4.2210242587601075,
-      "grad_norm": 0.6311129927635193,
-      "learning_rate": 0.0003472228818132757,
-      "loss": 3.3763,
+      "grad_norm": 0.6016018986701965,
+      "learning_rate": 0.00034720992984349703,
+      "loss": 3.3752,
       "step": 39150
     },
     {
       "epoch": 4.226415094339623,
-      "grad_norm": 0.7378162145614624,
-      "learning_rate": 0.00034689908256880733,
-      "loss": 3.3681,
+      "grad_norm": 0.70400071144104,
+      "learning_rate": 0.0003468861305990286,
+      "loss": 3.3671,
       "step": 39200
     },
     {
       "epoch": 4.231805929919138,
-      "grad_norm": 0.6093368530273438,
-      "learning_rate": 0.0003465752833243389,
-      "loss": 3.3713,
+      "grad_norm": 0.6123245358467102,
+      "learning_rate": 0.0003465623313545602,
+      "loss": 3.3699,
       "step": 39250
     },
     {
       "epoch": 4.237196765498652,
-      "grad_norm": 0.6382229328155518,
-      "learning_rate": 0.0003462514840798705,
-      "loss": 3.3736,
+      "grad_norm": 0.6317894458770752,
+      "learning_rate": 0.00034623853211009173,
+      "loss": 3.3728,
       "step": 39300
     },
     {
       "epoch": 4.242587601078167,
-      "grad_norm": 0.5930299758911133,
-      "learning_rate": 0.00034592768483540203,
-      "loss": 3.3852,
+      "grad_norm": 0.6047124862670898,
+      "learning_rate": 0.00034591473286562323,
+      "loss": 3.3843,
       "step": 39350
     },
     {
       "epoch": 4.247978436657682,
-      "grad_norm": 0.6670295000076294,
-      "learning_rate": 0.00034560388559093364,
-      "loss": 3.3791,
+      "grad_norm": 0.6490882039070129,
+      "learning_rate": 0.0003455909336211549,
+      "loss": 3.3801,
       "step": 39400
     },
     {
       "epoch": 4.2533692722371965,
-      "grad_norm": 0.6960169076919556,
-      "learning_rate": 0.00034528008634646514,
-      "loss": 3.3902,
+      "grad_norm": 0.6895270347595215,
+      "learning_rate": 0.0003452671343766864,
+      "loss": 3.3879,
       "step": 39450
     },
     {
       "epoch": 4.258760107816712,
-      "grad_norm": 0.6310935020446777,
-      "learning_rate": 0.00034495628710199674,
-      "loss": 3.3771,
+      "grad_norm": 0.6247091889381409,
+      "learning_rate": 0.000344943335132218,
+      "loss": 3.3764,
       "step": 39500
     },
     {
       "epoch": 4.264150943396227,
-      "grad_norm": 0.6121526956558228,
-      "learning_rate": 0.0003446324878575283,
-      "loss": 3.3639,
+      "grad_norm": 0.6222687363624573,
+      "learning_rate": 0.00034461953588774954,
+      "loss": 3.3631,
       "step": 39550
     },
     {
       "epoch": 4.269541778975741,
-      "grad_norm": 0.627255916595459,
-      "learning_rate": 0.00034430868861305984,
-      "loss": 3.3805,
+      "grad_norm": 0.6222398281097412,
+      "learning_rate": 0.00034429573664328115,
+      "loss": 3.3788,
       "step": 39600
     },
     {
       "epoch": 4.274932614555256,
-      "grad_norm": 0.6354991793632507,
-      "learning_rate": 0.00034398488936859145,
-      "loss": 3.3886,
+      "grad_norm": 0.6109077334403992,
+      "learning_rate": 0.0003439719373988127,
+      "loss": 3.3883,
       "step": 39650
     },
     {
       "epoch": 4.280323450134771,
-      "grad_norm": 0.6785920262336731,
-      "learning_rate": 0.000343661090124123,
-      "loss": 3.372,
+      "grad_norm": 0.6322901248931885,
+      "learning_rate": 0.00034364813815434425,
+      "loss": 3.3713,
       "step": 39700
     },
     {
       "epoch": 4.285714285714286,
-      "grad_norm": 0.6545200347900391,
-      "learning_rate": 0.0003433372908796546,
-      "loss": 3.395,
+      "grad_norm": 0.6394525170326233,
+      "learning_rate": 0.00034332433890987585,
+      "loss": 3.3939,
       "step": 39750
     },
     {
       "epoch": 4.291105121293801,
-      "grad_norm": 0.6170735359191895,
-      "learning_rate": 0.00034301349163518615,
-      "loss": 3.3826,
+      "grad_norm": 0.6161048412322998,
+      "learning_rate": 0.0003430005396654074,
+      "loss": 3.3808,
       "step": 39800
     },
     {
       "epoch": 4.296495956873315,
-      "grad_norm": 0.6551817059516907,
-      "learning_rate": 0.00034268969239071776,
-      "loss": 3.3834,
+      "grad_norm": 0.6096895337104797,
+      "learning_rate": 0.000342676740420939,
+      "loss": 3.3812,
       "step": 39850
     },
     {
       "epoch": 4.30188679245283,
-      "grad_norm": 0.6523256897926331,
-      "learning_rate": 0.0003423658931462493,
-      "loss": 3.384,
+      "grad_norm": 0.6321001648902893,
+      "learning_rate": 0.00034235294117647056,
+      "loss": 3.3828,
       "step": 39900
     },
     {
       "epoch": 4.307277628032345,
-      "grad_norm": 0.6886041164398193,
-      "learning_rate": 0.00034204209390178086,
-      "loss": 3.3928,
+      "grad_norm": 0.6452771425247192,
+      "learning_rate": 0.00034202914193200216,
+      "loss": 3.3883,
       "step": 39950
     },
     {
       "epoch": 4.3126684636118595,
-      "grad_norm": 0.6848386526107788,
-      "learning_rate": 0.00034171829465731246,
-      "loss": 3.3971,
+      "grad_norm": 0.6669614911079407,
+      "learning_rate": 0.0003417053426875337,
+      "loss": 3.3957,
       "step": 40000
     },
     {
       "epoch": 4.3126684636118595,
-      "eval_accuracy": 0.3776180049135009,
-      "eval_loss": 3.4378561973571777,
-      "eval_runtime": 185.0073,
-      "eval_samples_per_second": 97.353,
-      "eval_steps_per_second": 6.086,
+      "eval_accuracy": 0.37731149511697626,
+      "eval_loss": 3.437464952468872,
+      "eval_runtime": 185.0409,
+      "eval_samples_per_second": 97.335,
+      "eval_steps_per_second": 6.085,
       "step": 40000
     },
     {
       "epoch": 4.318059299191375,
-      "grad_norm": 0.6221829056739807,
-      "learning_rate": 0.00034139449541284396,
-      "loss": 3.3875,
+      "grad_norm": 0.6292645335197449,
+      "learning_rate": 0.0003413815434430653,
+      "loss": 3.3865,
       "step": 40050
     },
     {
       "epoch": 4.32345013477089,
-      "grad_norm": 0.6784347891807556,
-      "learning_rate": 0.00034107069616837556,
-      "loss": 3.3911,
+      "grad_norm": 0.6619394421577454,
+      "learning_rate": 0.0003410577441985968,
+      "loss": 3.3892,
       "step": 40100
     },
     {
       "epoch": 4.328840970350404,
-      "grad_norm": 0.6379362344741821,
-      "learning_rate": 0.0003407468969239071,
-      "loss": 3.372,
+      "grad_norm": 0.608463704586029,
+      "learning_rate": 0.00034073394495412837,
+      "loss": 3.3712,
       "step": 40150
     },
     {
       "epoch": 4.334231805929919,
-      "grad_norm": 0.6500546932220459,
-      "learning_rate": 0.0003404230976794387,
-      "loss": 3.4049,
+      "grad_norm": 0.6409887671470642,
+      "learning_rate": 0.00034041014570965997,
+      "loss": 3.4038,
       "step": 40200
     },
     {
       "epoch": 4.339622641509434,
-      "grad_norm": 0.6693824529647827,
-      "learning_rate": 0.00034009929843497027,
-      "loss": 3.4065,
+      "grad_norm": 0.6474012136459351,
+      "learning_rate": 0.0003400863464651915,
+      "loss": 3.4056,
       "step": 40250
     },
     {
       "epoch": 4.345013477088949,
-      "grad_norm": 0.6409181952476501,
-      "learning_rate": 0.0003397754991905019,
-      "loss": 3.3808,
+      "grad_norm": 0.6828320622444153,
+      "learning_rate": 0.0003397625472207231,
+      "loss": 3.3789,
       "step": 40300
     },
     {
       "epoch": 4.350404312668464,
-      "grad_norm": 0.6256203651428223,
-      "learning_rate": 0.0003394516999460334,
-      "loss": 3.392,
+      "grad_norm": 0.659954845905304,
+      "learning_rate": 0.0003394387479762547,
+      "loss": 3.3883,
       "step": 40350
     },
     {
       "epoch": 4.355795148247978,
-      "grad_norm": 0.6128562688827515,
-      "learning_rate": 0.000339127900701565,
-      "loss": 3.3865,
+      "grad_norm": 0.6052781939506531,
+      "learning_rate": 0.0003391149487317863,
+      "loss": 3.3847,
       "step": 40400
     },
     {
       "epoch": 4.361185983827493,
-      "grad_norm": 0.638848066329956,
-      "learning_rate": 0.0003388041014570966,
-      "loss": 3.3884,
+      "grad_norm": 0.6608632802963257,
+      "learning_rate": 0.00033879114948731783,
+      "loss": 3.3881,
       "step": 40450
     },
     {
       "epoch": 4.366576819407008,
-      "grad_norm": 0.6677626371383667,
-      "learning_rate": 0.00033848030221262813,
-      "loss": 3.3821,
+      "grad_norm": 0.6373227834701538,
+      "learning_rate": 0.00033846735024284944,
+      "loss": 3.3809,
       "step": 40500
     },
     {
       "epoch": 4.3719676549865225,
-      "grad_norm": 0.6529048085212708,
-      "learning_rate": 0.00033815650296815974,
-      "loss": 3.388,
+      "grad_norm": 0.6604944467544556,
+      "learning_rate": 0.000338143550998381,
+      "loss": 3.3869,
       "step": 40550
     },
     {
       "epoch": 4.377358490566038,
-      "grad_norm": 0.6577892303466797,
-      "learning_rate": 0.0003378327037236913,
-      "loss": 3.3945,
+      "grad_norm": 0.619690477848053,
+      "learning_rate": 0.00033781975175391254,
+      "loss": 3.3931,
       "step": 40600
     },
     {
       "epoch": 4.382749326145553,
-      "grad_norm": 0.6787473559379578,
-      "learning_rate": 0.0003375089044792229,
-      "loss": 3.3849,
+      "grad_norm": 0.6604430675506592,
+      "learning_rate": 0.00033749595250944414,
+      "loss": 3.3844,
       "step": 40650
     },
     {
       "epoch": 4.388140161725067,
-      "grad_norm": 0.6604834794998169,
-      "learning_rate": 0.00033718510523475444,
-      "loss": 3.3991,
+      "grad_norm": 0.6418233513832092,
+      "learning_rate": 0.00033717215326497564,
+      "loss": 3.397,
       "step": 40700
     },
     {
       "epoch": 4.393530997304582,
-      "grad_norm": 0.6834147572517395,
-      "learning_rate": 0.00033686130599028605,
-      "loss": 3.4014,
+      "grad_norm": 0.6358850002288818,
+      "learning_rate": 0.0003368483540205073,
+      "loss": 3.3994,
       "step": 40750
     },
     {
       "epoch": 4.398921832884097,
-      "grad_norm": 0.6034276485443115,
-      "learning_rate": 0.00033653750674581754,
-      "loss": 3.389,
+      "grad_norm": 0.6532427072525024,
+      "learning_rate": 0.0003365245547760388,
+      "loss": 3.3877,
       "step": 40800
     },
     {
       "epoch": 4.404312668463612,
-      "grad_norm": 0.6377930045127869,
-      "learning_rate": 0.0003362137075013491,
-      "loss": 3.3799,
+      "grad_norm": 0.6354030966758728,
+      "learning_rate": 0.0003362007555315704,
+      "loss": 3.3772,
       "step": 40850
     },
     {
       "epoch": 4.409703504043127,
-      "grad_norm": 0.663975179195404,
-      "learning_rate": 0.0003358899082568807,
-      "loss": 3.3802,
+      "grad_norm": 0.6793525815010071,
+      "learning_rate": 0.00033587695628710195,
+      "loss": 3.3799,
       "step": 40900
     },
     {
       "epoch": 4.415094339622642,
-      "grad_norm": 0.9682093858718872,
-      "learning_rate": 0.00033556610901241225,
-      "loss": 3.377,
+      "grad_norm": 0.9723051190376282,
+      "learning_rate": 0.00033555315704263355,
+      "loss": 3.3773,
       "step": 40950
     },
     {
       "epoch": 4.420485175202156,
-      "grad_norm": 0.6749927997589111,
-      "learning_rate": 0.00033524230976794385,
-      "loss": 3.411,
+      "grad_norm": 0.6340115666389465,
+      "learning_rate": 0.0003352293577981651,
+      "loss": 3.4119,
       "step": 41000
     },
     {
       "epoch": 4.420485175202156,
-      "eval_accuracy": 0.3784913568257537,
-      "eval_loss": 3.4337351322174072,
-      "eval_runtime": 184.8697,
-      "eval_samples_per_second": 97.425,
-      "eval_steps_per_second": 6.091,
+      "eval_accuracy": 0.3779828613107819,
+      "eval_loss": 3.4332971572875977,
+      "eval_runtime": 185.1533,
+      "eval_samples_per_second": 97.276,
+      "eval_steps_per_second": 6.081,
       "step": 41000
     },
     {
       "epoch": 4.425876010781671,
-      "grad_norm": 0.628652036190033,
-      "learning_rate": 0.0003349185105234754,
-      "loss": 3.3868,
+      "grad_norm": 0.6561840176582336,
+      "learning_rate": 0.00033490555855369665,
+      "loss": 3.3857,
       "step": 41050
     },
     {
       "epoch": 4.431266846361186,
-      "grad_norm": 0.6534876823425293,
-      "learning_rate": 0.000334594711279007,
-      "loss": 3.3965,
+      "grad_norm": 0.6575011610984802,
+      "learning_rate": 0.00033458175930922826,
+      "loss": 3.3953,
       "step": 41100
     },
     {
       "epoch": 4.436657681940701,
-      "grad_norm": 0.6127163171768188,
-      "learning_rate": 0.00033427091203453856,
-      "loss": 3.4026,
+      "grad_norm": 0.6666945815086365,
+      "learning_rate": 0.0003342579600647598,
+      "loss": 3.4005,
       "step": 41150
     },
     {
       "epoch": 4.442048517520216,
-      "grad_norm": 0.6120777726173401,
-      "learning_rate": 0.0003339535887749595,
-      "loss": 3.3872,
+      "grad_norm": 0.5999481678009033,
+      "learning_rate": 0.0003339341608202914,
+      "loss": 3.3865,
       "step": 41200
     },
     {
       "epoch": 4.44743935309973,
-      "grad_norm": 0.6257559061050415,
-      "learning_rate": 0.00033362978953049106,
-      "loss": 3.4045,
+      "grad_norm": 0.6154288649559021,
+      "learning_rate": 0.00033361036157582297,
+      "loss": 3.4036,
       "step": 41250
     },
     {
       "epoch": 4.452830188679245,
-      "grad_norm": 0.6158122420310974,
-      "learning_rate": 0.00033330599028602267,
-      "loss": 3.3767,
+      "grad_norm": 0.6239913702011108,
+      "learning_rate": 0.00033328656233135457,
+      "loss": 3.3756,
       "step": 41300
     },
     {
       "epoch": 4.45822102425876,
-      "grad_norm": 0.6703405976295471,
-      "learning_rate": 0.0003329821910415542,
-      "loss": 3.382,
+      "grad_norm": 0.6489045023918152,
+      "learning_rate": 0.0003329627630868861,
+      "loss": 3.3797,
       "step": 41350
     },
     {
       "epoch": 4.463611859838275,
-      "grad_norm": 0.6126678586006165,
-      "learning_rate": 0.0003326583917970858,
-      "loss": 3.4067,
+      "grad_norm": 0.6149885058403015,
+      "learning_rate": 0.0003326389638424177,
+      "loss": 3.4028,
       "step": 41400
     },
     {
       "epoch": 4.46900269541779,
-      "grad_norm": 0.6338332295417786,
-      "learning_rate": 0.0003323345925526173,
-      "loss": 3.3812,
+      "grad_norm": 0.6494463682174683,
+      "learning_rate": 0.0003323151645979492,
+      "loss": 3.3799,
       "step": 41450
     },
     {
       "epoch": 4.474393530997305,
-      "grad_norm": 1.067539095878601,
-      "learning_rate": 0.0003320107933081489,
-      "loss": 3.3943,
+      "grad_norm": 0.8456867933273315,
+      "learning_rate": 0.00033199136535348077,
+      "loss": 3.3919,
       "step": 41500
     },
     {
       "epoch": 4.479784366576819,
-      "grad_norm": 0.6305244565010071,
-      "learning_rate": 0.0003316869940636805,
-      "loss": 3.3855,
+      "grad_norm": 0.6378749012947083,
+      "learning_rate": 0.0003316675661090124,
+      "loss": 3.3854,
       "step": 41550
     },
     {
       "epoch": 4.485175202156334,
-      "grad_norm": 0.6363754868507385,
-      "learning_rate": 0.000331363194819212,
-      "loss": 3.3964,
+      "grad_norm": 0.6068414449691772,
+      "learning_rate": 0.00033134376686454393,
+      "loss": 3.3958,
       "step": 41600
     },
     {
       "epoch": 4.490566037735849,
-      "grad_norm": 0.6011024117469788,
-      "learning_rate": 0.00033103939557474363,
-      "loss": 3.3798,
+      "grad_norm": 0.6004335284233093,
+      "learning_rate": 0.00033101996762007553,
+      "loss": 3.3789,
       "step": 41650
     },
     {
       "epoch": 4.495956873315364,
-      "grad_norm": 0.6291077136993408,
-      "learning_rate": 0.0003307155963302752,
-      "loss": 3.4022,
+      "grad_norm": 0.6522519588470459,
+      "learning_rate": 0.0003306961683756071,
+      "loss": 3.4029,
       "step": 41700
     },
     {
       "epoch": 4.501347708894879,
-      "grad_norm": 0.6409448385238647,
-      "learning_rate": 0.0003303917970858068,
-      "loss": 3.3858,
+      "grad_norm": 0.645179271697998,
+      "learning_rate": 0.0003303723691311387,
+      "loss": 3.3842,
       "step": 41750
     },
     {
       "epoch": 4.506738544474393,
-      "grad_norm": 0.6451292037963867,
-      "learning_rate": 0.00033006799784133833,
-      "loss": 3.391,
+      "grad_norm": 0.6851235032081604,
+      "learning_rate": 0.00033004856988667024,
+      "loss": 3.3892,
       "step": 41800
     },
     {
       "epoch": 4.512129380053908,
-      "grad_norm": 0.6577849388122559,
-      "learning_rate": 0.00032974419859686994,
-      "loss": 3.3801,
+      "grad_norm": 0.6213834881782532,
+      "learning_rate": 0.0003297247706422018,
+      "loss": 3.3784,
       "step": 41850
     },
     {
       "epoch": 4.517520215633423,
-      "grad_norm": 0.6435655355453491,
-      "learning_rate": 0.0003294203993524015,
-      "loss": 3.3903,
+      "grad_norm": 0.6451056599617004,
+      "learning_rate": 0.00032940744738262274,
+      "loss": 3.3886,
       "step": 41900
     },
     {
       "epoch": 4.5229110512129385,
-      "grad_norm": 0.6412240266799927,
-      "learning_rate": 0.0003290966001079331,
-      "loss": 3.3722,
+      "grad_norm": 0.6113904714584351,
+      "learning_rate": 0.00032908364813815435,
+      "loss": 3.3711,
       "step": 41950
     },
     {
       "epoch": 4.528301886792453,
-      "grad_norm": 0.622590184211731,
-      "learning_rate": 0.00032877280086346465,
-      "loss": 3.386,
+      "grad_norm": 0.605117917060852,
+      "learning_rate": 0.0003287598488936859,
+      "loss": 3.3842,
       "step": 42000
     },
     {
       "epoch": 4.528301886792453,
-      "eval_accuracy": 0.37879612817607333,
-      "eval_loss": 3.4283080101013184,
-      "eval_runtime": 185.1161,
-      "eval_samples_per_second": 97.296,
-      "eval_steps_per_second": 6.083,
+      "eval_accuracy": 0.37882416062112945,
+      "eval_loss": 3.427164316177368,
+      "eval_runtime": 184.9685,
+      "eval_samples_per_second": 97.373,
+      "eval_steps_per_second": 6.088,
       "step": 42000
     },
     {
       "epoch": 4.533692722371968,
-      "grad_norm": 0.6410642266273499,
-      "learning_rate": 0.00032844900161899614,
-      "loss": 3.3828,
+      "grad_norm": 0.6477672457695007,
+      "learning_rate": 0.0003284360496492175,
+      "loss": 3.3816,
       "step": 42050
     },
     {
       "epoch": 4.539083557951482,
-      "grad_norm": 0.6771649718284607,
-      "learning_rate": 0.0003281252023745278,
-      "loss": 3.3897,
+      "grad_norm": 0.6513396501541138,
+      "learning_rate": 0.000328112250404749,
+      "loss": 3.3891,
       "step": 42100
     },
     {
       "epoch": 4.544474393530997,
-      "grad_norm": 0.6513844728469849,
-      "learning_rate": 0.0003278014031300593,
-      "loss": 3.4196,
+      "grad_norm": 0.6318349242210388,
+      "learning_rate": 0.00032778845116028066,
+      "loss": 3.4179,
       "step": 42150
     },
     {
       "epoch": 4.549865229110512,
-      "grad_norm": 0.6491783857345581,
-      "learning_rate": 0.0003274776038855909,
-      "loss": 3.387,
+      "grad_norm": 0.6026633977890015,
+      "learning_rate": 0.00032746465191581215,
+      "loss": 3.3851,
       "step": 42200
     },
     {
       "epoch": 4.555256064690027,
-      "grad_norm": 0.643081545829773,
-      "learning_rate": 0.00032715380464112245,
-      "loss": 3.3852,
+      "grad_norm": 0.6635650396347046,
+      "learning_rate": 0.0003271408526713437,
+      "loss": 3.3838,
       "step": 42250
     },
     {
       "epoch": 4.560646900269542,
-      "grad_norm": 0.6979756951332092,
-      "learning_rate": 0.00032683000539665406,
-      "loss": 3.3863,
+      "grad_norm": 0.6454138159751892,
+      "learning_rate": 0.0003268170534268753,
+      "loss": 3.3865,
       "step": 42300
     },
     {
       "epoch": 4.566037735849057,
-      "grad_norm": 0.6504450440406799,
-      "learning_rate": 0.0003265062061521856,
-      "loss": 3.3895,
+      "grad_norm": 0.6208834052085876,
+      "learning_rate": 0.00032649325418240686,
+      "loss": 3.3868,
       "step": 42350
     },
     {
       "epoch": 4.571428571428571,
-      "grad_norm": 0.6485116481781006,
-      "learning_rate": 0.0003261824069077172,
-      "loss": 3.378,
+      "grad_norm": 0.6189976930618286,
+      "learning_rate": 0.00032616945493793846,
+      "loss": 3.3757,
       "step": 42400
     },
     {
       "epoch": 4.576819407008086,
-      "grad_norm": 0.686379075050354,
-      "learning_rate": 0.00032585860766324876,
-      "loss": 3.3909,
+      "grad_norm": 0.6465439796447754,
+      "learning_rate": 0.00032584565569347,
+      "loss": 3.3898,
       "step": 42450
     },
     {
       "epoch": 4.5822102425876015,
-      "grad_norm": 0.624584436416626,
-      "learning_rate": 0.0003255348084187803,
-      "loss": 3.3864,
+      "grad_norm": 0.6091396808624268,
+      "learning_rate": 0.0003255218564490016,
+      "loss": 3.3861,
       "step": 42500
     },
     {
       "epoch": 4.587601078167116,
-      "grad_norm": 0.7264350652694702,
-      "learning_rate": 0.0003252110091743119,
-      "loss": 3.3879,
+      "grad_norm": 0.6565108895301819,
+      "learning_rate": 0.00032519805720453317,
+      "loss": 3.3858,
       "step": 42550
     },
     {
       "epoch": 4.592991913746631,
-      "grad_norm": 0.6314030289649963,
-      "learning_rate": 0.00032488720992984347,
-      "loss": 3.3723,
+      "grad_norm": 0.6114659309387207,
+      "learning_rate": 0.0003248742579600647,
+      "loss": 3.3709,
       "step": 42600
     },
     {
       "epoch": 4.598382749326145,
-      "grad_norm": 0.7158349752426147,
-      "learning_rate": 0.0003245634106853751,
-      "loss": 3.3733,
+      "grad_norm": 0.6839783787727356,
+      "learning_rate": 0.0003245504587155963,
+      "loss": 3.3721,
       "step": 42650
     },
     {
       "epoch": 4.60377358490566,
-      "grad_norm": 0.6618641018867493,
-      "learning_rate": 0.0003242396114409066,
-      "loss": 3.3827,
+      "grad_norm": 0.664556086063385,
+      "learning_rate": 0.0003242266594711278,
+      "loss": 3.3804,
       "step": 42700
     },
     {
       "epoch": 4.609164420485175,
-      "grad_norm": 0.643898606300354,
-      "learning_rate": 0.00032391581219643823,
-      "loss": 3.3991,
+      "grad_norm": 0.6460453271865845,
+      "learning_rate": 0.0003239093362115488,
+      "loss": 3.3993,
       "step": 42750
     },
     {
       "epoch": 4.6145552560646905,
-      "grad_norm": 0.6714193820953369,
-      "learning_rate": 0.0003235920129519697,
-      "loss": 3.3718,
+      "grad_norm": 0.6806046366691589,
+      "learning_rate": 0.00032358553696708043,
+      "loss": 3.371,
       "step": 42800
     },
     {
       "epoch": 4.619946091644205,
-      "grad_norm": 0.650656521320343,
-      "learning_rate": 0.0003232682137075013,
-      "loss": 3.3804,
+      "grad_norm": 0.6139158606529236,
+      "learning_rate": 0.00032326173772261193,
+      "loss": 3.3778,
       "step": 42850
     },
     {
       "epoch": 4.62533692722372,
-      "grad_norm": 0.6801531314849854,
-      "learning_rate": 0.0003229444144630329,
-      "loss": 3.3771,
+      "grad_norm": 0.6241970062255859,
+      "learning_rate": 0.00032293793847814353,
+      "loss": 3.377,
       "step": 42900
     },
     {
       "epoch": 4.630727762803234,
-      "grad_norm": 0.6805683374404907,
-      "learning_rate": 0.00032262061521856443,
-      "loss": 3.4009,
+      "grad_norm": 0.678596019744873,
+      "learning_rate": 0.0003226141392336751,
+      "loss": 3.3996,
       "step": 42950
     },
     {
       "epoch": 4.636118598382749,
-      "grad_norm": 0.6824569702148438,
-      "learning_rate": 0.00032229681597409604,
-      "loss": 3.4062,
+      "grad_norm": 0.6589379906654358,
+      "learning_rate": 0.00032229033998920663,
+      "loss": 3.4039,
       "step": 43000
     },
     {
       "epoch": 4.636118598382749,
-      "eval_accuracy": 0.3791326261696348,
-      "eval_loss": 3.4244093894958496,
-      "eval_runtime": 184.8967,
-      "eval_samples_per_second": 97.411,
-      "eval_steps_per_second": 6.09,
+      "eval_accuracy": 0.3791835843740197,
+      "eval_loss": 3.422538995742798,
+      "eval_runtime": 185.1391,
+      "eval_samples_per_second": 97.284,
+      "eval_steps_per_second": 6.082,
       "step": 43000
     },
     {
       "epoch": 4.6415094339622645,
-      "grad_norm": 0.6415202021598816,
-      "learning_rate": 0.0003219730167296276,
-      "loss": 3.3891,
+      "grad_norm": 0.6241201758384705,
+      "learning_rate": 0.00032196654074473824,
+      "loss": 3.3885,
       "step": 43050
     },
     {
       "epoch": 4.646900269541779,
-      "grad_norm": 0.6398457884788513,
-      "learning_rate": 0.0003216492174851592,
-      "loss": 3.3942,
+      "grad_norm": 0.6069834232330322,
+      "learning_rate": 0.0003216427415002698,
+      "loss": 3.3944,
       "step": 43100
     },
     {
       "epoch": 4.652291105121294,
-      "grad_norm": 0.7170093059539795,
-      "learning_rate": 0.00032132541824069074,
-      "loss": 3.409,
+      "grad_norm": 0.6450423002243042,
+      "learning_rate": 0.0003213189422558014,
+      "loss": 3.4062,
       "step": 43150
     },
     {
       "epoch": 4.657681940700809,
-      "grad_norm": 0.6552037596702576,
-      "learning_rate": 0.0003210080949811117,
-      "loss": 3.3803,
+      "grad_norm": 0.6572805643081665,
+      "learning_rate": 0.00032099514301133295,
+      "loss": 3.3788,
       "step": 43200
     },
     {
       "epoch": 4.663072776280323,
-      "grad_norm": 0.6536373496055603,
-      "learning_rate": 0.00032068429573664324,
-      "loss": 3.3793,
+      "grad_norm": 0.6346420645713806,
+      "learning_rate": 0.00032067134376686455,
+      "loss": 3.3781,
       "step": 43250
     },
     {
       "epoch": 4.668463611859838,
-      "grad_norm": 0.6051586866378784,
-      "learning_rate": 0.00032036049649217485,
-      "loss": 3.3903,
+      "grad_norm": 0.6331166625022888,
+      "learning_rate": 0.0003203475445223961,
+      "loss": 3.3912,
       "step": 43300
     },
     {
       "epoch": 4.6738544474393535,
-      "grad_norm": 0.665641725063324,
-      "learning_rate": 0.0003200366972477064,
+      "grad_norm": 0.6622107028961182,
+      "learning_rate": 0.0003200237452779277,
       "loss": 3.3884,
       "step": 43350
     },
     {
       "epoch": 4.679245283018868,
-      "grad_norm": 0.6546832323074341,
-      "learning_rate": 0.000319712898003238,
-      "loss": 3.4008,
+      "grad_norm": 0.6595224738121033,
+      "learning_rate": 0.00031969994603345926,
+      "loss": 3.4004,
       "step": 43400
     },
     {
       "epoch": 4.684636118598383,
-      "grad_norm": 0.6022357940673828,
-      "learning_rate": 0.0003193890987587695,
-      "loss": 3.3881,
+      "grad_norm": 0.6058201789855957,
+      "learning_rate": 0.00031937614678899075,
+      "loss": 3.3886,
       "step": 43450
     },
     {
       "epoch": 4.690026954177897,
-      "grad_norm": 0.6649525165557861,
-      "learning_rate": 0.0003190652995143011,
+      "grad_norm": 0.6582033038139343,
+      "learning_rate": 0.00031905234754452236,
       "loss": 3.4009,
       "step": 43500
     },
     {
       "epoch": 4.695417789757412,
-      "grad_norm": 0.6134843826293945,
-      "learning_rate": 0.00031874150026983266,
-      "loss": 3.3842,
+      "grad_norm": 0.6451115608215332,
+      "learning_rate": 0.0003187285483000539,
+      "loss": 3.384,
       "step": 43550
     },
     {
       "epoch": 4.7008086253369274,
-      "grad_norm": 0.7057274580001831,
-      "learning_rate": 0.0003184177010253642,
-      "loss": 3.3762,
+      "grad_norm": 0.6477821469306946,
+      "learning_rate": 0.0003184047490555855,
+      "loss": 3.3735,
       "step": 43600
     },
     {
       "epoch": 4.706199460916442,
-      "grad_norm": 0.659426748752594,
-      "learning_rate": 0.0003180939017808958,
-      "loss": 3.3923,
+      "grad_norm": 0.6032214164733887,
+      "learning_rate": 0.00031808094981111706,
+      "loss": 3.3908,
       "step": 43650
     },
     {
       "epoch": 4.711590296495957,
-      "grad_norm": 0.6758320331573486,
-      "learning_rate": 0.00031777010253642736,
-      "loss": 3.3784,
+      "grad_norm": 0.6578019261360168,
+      "learning_rate": 0.00031775715056664867,
+      "loss": 3.3762,
       "step": 43700
     },
     {
       "epoch": 4.716981132075472,
-      "grad_norm": 0.6479880213737488,
-      "learning_rate": 0.00031744630329195897,
-      "loss": 3.3966,
+      "grad_norm": 0.6483471393585205,
+      "learning_rate": 0.0003174333513221802,
+      "loss": 3.396,
       "step": 43750
     },
     {
       "epoch": 4.722371967654986,
-      "grad_norm": 0.6481097936630249,
-      "learning_rate": 0.0003171225040474905,
-      "loss": 3.3905,
+      "grad_norm": 0.6229943633079529,
+      "learning_rate": 0.00031710955207771177,
+      "loss": 3.3894,
       "step": 43800
     },
     {
       "epoch": 4.727762803234501,
-      "grad_norm": 0.8242167830467224,
-      "learning_rate": 0.0003167987048030221,
-      "loss": 3.3923,
+      "grad_norm": 0.6678479909896851,
+      "learning_rate": 0.0003167857528332434,
+      "loss": 3.3916,
       "step": 43850
     },
     {
       "epoch": 4.7331536388140165,
-      "grad_norm": 0.6138800978660583,
-      "learning_rate": 0.00031647490555855367,
-      "loss": 3.396,
+      "grad_norm": 0.5916961431503296,
+      "learning_rate": 0.0003164619535887749,
+      "loss": 3.3943,
       "step": 43900
     },
     {
       "epoch": 4.738544474393531,
-      "grad_norm": 0.6652048230171204,
-      "learning_rate": 0.0003161511063140853,
-      "loss": 3.3855,
+      "grad_norm": 0.6245575547218323,
+      "learning_rate": 0.00031613815434430653,
+      "loss": 3.3859,
       "step": 43950
     },
     {
       "epoch": 4.743935309973046,
-      "grad_norm": 0.6460886597633362,
-      "learning_rate": 0.00031582730706961683,
-      "loss": 3.3905,
+      "grad_norm": 0.6339305639266968,
+      "learning_rate": 0.0003158143550998381,
+      "loss": 3.3886,
       "step": 44000
     },
     {
       "epoch": 4.743935309973046,
-      "eval_accuracy": 0.3798925444670092,
-      "eval_loss": 3.4189040660858154,
-      "eval_runtime": 185.0852,
-      "eval_samples_per_second": 97.312,
-      "eval_steps_per_second": 6.084,
+      "eval_accuracy": 0.37981963837928573,
+      "eval_loss": 3.416712760925293,
+      "eval_runtime": 185.1327,
+      "eval_samples_per_second": 97.287,
+      "eval_steps_per_second": 6.082,
       "step": 44000
     },
     {
       "epoch": 4.74932614555256,
-      "grad_norm": 0.6555764675140381,
-      "learning_rate": 0.0003155035078251483,
-      "loss": 3.397,
+      "grad_norm": 0.64893639087677,
+      "learning_rate": 0.0003154905558553697,
+      "loss": 3.3989,
       "step": 44050
     },
     {
       "epoch": 4.754716981132075,
-      "grad_norm": 0.6533858776092529,
-      "learning_rate": 0.00031517970858068,
-      "loss": 3.3928,
+      "grad_norm": 0.6587449312210083,
+      "learning_rate": 0.0003151667566109012,
+      "loss": 3.3939,
       "step": 44100
     },
     {
       "epoch": 4.7601078167115904,
-      "grad_norm": 0.6652816534042358,
-      "learning_rate": 0.0003148559093362115,
-      "loss": 3.4021,
+      "grad_norm": 0.6603413820266724,
+      "learning_rate": 0.00031484295736643284,
+      "loss": 3.4024,
       "step": 44150
     },
     {
       "epoch": 4.765498652291106,
-      "grad_norm": 0.6321492195129395,
-      "learning_rate": 0.00031453858607663243,
-      "loss": 3.4016,
+      "grad_norm": 0.6770880818367004,
+      "learning_rate": 0.00031451915812196434,
+      "loss": 3.4005,
       "step": 44200
     },
     {
       "epoch": 4.77088948787062,
-      "grad_norm": 0.632645308971405,
-      "learning_rate": 0.00031421478683216404,
-      "loss": 3.3598,
+      "grad_norm": 0.5969046950340271,
+      "learning_rate": 0.0003141953588774959,
+      "loss": 3.3597,
       "step": 44250
     },
     {
       "epoch": 4.776280323450135,
-      "grad_norm": 0.6430423855781555,
-      "learning_rate": 0.0003138909875876956,
-      "loss": 3.387,
+      "grad_norm": 0.6240620017051697,
+      "learning_rate": 0.0003138715596330275,
+      "loss": 3.3875,
       "step": 44300
     },
     {
       "epoch": 4.781671159029649,
-      "grad_norm": 0.6638597249984741,
-      "learning_rate": 0.0003135671883432272,
-      "loss": 3.3917,
+      "grad_norm": 0.6881016492843628,
+      "learning_rate": 0.00031354776038855904,
+      "loss": 3.3925,
       "step": 44350
     },
     {
       "epoch": 4.787061994609164,
-      "grad_norm": 0.6225000023841858,
-      "learning_rate": 0.00031324338909875874,
-      "loss": 3.3763,
+      "grad_norm": 0.602745532989502,
+      "learning_rate": 0.00031322396114409065,
+      "loss": 3.3754,
       "step": 44400
     },
     {
       "epoch": 4.7924528301886795,
-      "grad_norm": 0.6393536329269409,
-      "learning_rate": 0.0003129195898542903,
-      "loss": 3.3982,
+      "grad_norm": 0.6551032066345215,
+      "learning_rate": 0.0003129001618996222,
+      "loss": 3.3987,
       "step": 44450
     },
     {
       "epoch": 4.797843665768194,
-      "grad_norm": 0.7148990035057068,
-      "learning_rate": 0.0003125957906098219,
-      "loss": 3.3954,
+      "grad_norm": 0.715237557888031,
+      "learning_rate": 0.0003125763626551538,
+      "loss": 3.3968,
       "step": 44500
     },
     {
       "epoch": 4.803234501347709,
-      "grad_norm": 0.6571036577224731,
-      "learning_rate": 0.00031227199136535345,
-      "loss": 3.3703,
+      "grad_norm": 0.6164499521255493,
+      "learning_rate": 0.00031225256341068535,
+      "loss": 3.3697,
       "step": 44550
     },
     {
       "epoch": 4.808625336927224,
-      "grad_norm": 0.6597515940666199,
-      "learning_rate": 0.00031194819212088505,
-      "loss": 3.3852,
+      "grad_norm": 0.6373364925384521,
+      "learning_rate": 0.00031192876416621696,
+      "loss": 3.3842,
       "step": 44600
     },
     {
       "epoch": 4.814016172506738,
-      "grad_norm": 0.6399565935134888,
-      "learning_rate": 0.0003116243928764166,
-      "loss": 3.3852,
+      "grad_norm": 0.6444586515426636,
+      "learning_rate": 0.0003116049649217485,
+      "loss": 3.3847,
       "step": 44650
     },
     {
       "epoch": 4.819407008086253,
-      "grad_norm": 0.6244239807128906,
-      "learning_rate": 0.0003113005936319482,
-      "loss": 3.3862,
+      "grad_norm": 0.6503430604934692,
+      "learning_rate": 0.00031128116567728,
+      "loss": 3.3859,
       "step": 44700
     },
     {
       "epoch": 4.824797843665769,
-      "grad_norm": 0.6903212070465088,
-      "learning_rate": 0.00031097679438747976,
-      "loss": 3.4,
+      "grad_norm": 0.6720568537712097,
+      "learning_rate": 0.00031095736643281166,
+      "loss": 3.4001,
       "step": 44750
     },
     {
       "epoch": 4.830188679245283,
-      "grad_norm": 0.722104549407959,
-      "learning_rate": 0.00031065299514301126,
-      "loss": 3.3689,
+      "grad_norm": 0.7140028476715088,
+      "learning_rate": 0.00031063356718834316,
+      "loss": 3.3695,
       "step": 44800
     },
     {
       "epoch": 4.835579514824798,
-      "grad_norm": 0.6965301036834717,
-      "learning_rate": 0.00031032919589854286,
-      "loss": 3.385,
+      "grad_norm": 0.6931409239768982,
+      "learning_rate": 0.00031030976794387476,
+      "loss": 3.3834,
       "step": 44850
     },
     {
       "epoch": 4.840970350404312,
-      "grad_norm": 0.6514486074447632,
-      "learning_rate": 0.0003100053966540744,
-      "loss": 3.362,
+      "grad_norm": 0.6551147699356079,
+      "learning_rate": 0.0003099859686994063,
+      "loss": 3.3616,
       "step": 44900
     },
     {
       "epoch": 4.846361185983827,
-      "grad_norm": 0.6319965720176697,
-      "learning_rate": 0.000309681597409606,
-      "loss": 3.4001,
+      "grad_norm": 0.6414363384246826,
+      "learning_rate": 0.0003096621694549379,
+      "loss": 3.3987,
       "step": 44950
     },
     {
       "epoch": 4.8517520215633425,
-      "grad_norm": 0.6779524683952332,
-      "learning_rate": 0.00030935779816513757,
-      "loss": 3.3846,
+      "grad_norm": 0.658860981464386,
+      "learning_rate": 0.00030933837021046947,
+      "loss": 3.3862,
       "step": 45000
     },
     {
       "epoch": 4.8517520215633425,
-      "eval_accuracy": 0.3805246869683139,
-      "eval_loss": 3.41042423248291,
-      "eval_runtime": 184.8977,
-      "eval_samples_per_second": 97.411,
-      "eval_steps_per_second": 6.09,
+      "eval_accuracy": 0.3804288551212615,
+      "eval_loss": 3.4101946353912354,
+      "eval_runtime": 185.1294,
+      "eval_samples_per_second": 97.289,
+      "eval_steps_per_second": 6.082,
       "step": 45000
     },
     {
       "epoch": 4.857142857142857,
-      "grad_norm": 0.6845847964286804,
-      "learning_rate": 0.00030903399892066917,
+      "grad_norm": 0.673330545425415,
+      "learning_rate": 0.0003090145709660011,
       "loss": 3.3955,
       "step": 45050
     },
     {
       "epoch": 4.862533692722372,
-      "grad_norm": 0.6388900279998779,
-      "learning_rate": 0.0003087101996762007,
-      "loss": 3.4001,
+      "grad_norm": 0.6617267727851868,
+      "learning_rate": 0.0003086907717215326,
+      "loss": 3.3992,
       "step": 45100
     },
     {
       "epoch": 4.867924528301887,
-      "grad_norm": 0.6666938066482544,
-      "learning_rate": 0.0003083864004317323,
-      "loss": 3.3707,
+      "grad_norm": 0.6610994338989258,
+      "learning_rate": 0.0003083669724770642,
+      "loss": 3.3697,
       "step": 45150
     },
     {
       "epoch": 4.873315363881401,
-      "grad_norm": 0.6598957180976868,
-      "learning_rate": 0.0003080626011872639,
-      "loss": 3.3805,
+      "grad_norm": 0.60479336977005,
+      "learning_rate": 0.0003080431732325958,
+      "loss": 3.3779,
       "step": 45200
     },
     {
       "epoch": 4.878706199460916,
-      "grad_norm": 0.6968781352043152,
-      "learning_rate": 0.00030773880194279543,
-      "loss": 3.3946,
+      "grad_norm": 0.7236313223838806,
+      "learning_rate": 0.00030771937398812733,
+      "loss": 3.3945,
       "step": 45250
     },
     {
       "epoch": 4.884097035040432,
-      "grad_norm": 0.6397393345832825,
-      "learning_rate": 0.00030741500269832703,
+      "grad_norm": 0.6632745862007141,
+      "learning_rate": 0.00030739557474365894,
       "loss": 3.4061,
       "step": 45300
     },
     {
       "epoch": 4.889487870619946,
-      "grad_norm": 0.6977748870849609,
-      "learning_rate": 0.0003070912034538586,
-      "loss": 3.3841,
+      "grad_norm": 0.6762586236000061,
+      "learning_rate": 0.0003070717754991905,
+      "loss": 3.3828,
       "step": 45350
     },
     {
       "epoch": 4.894878706199461,
-      "grad_norm": 0.6801537275314331,
-      "learning_rate": 0.0003067674042093902,
-      "loss": 3.3938,
+      "grad_norm": 0.6466242074966431,
+      "learning_rate": 0.0003067479762547221,
+      "loss": 3.3951,
       "step": 45400
     },
     {
       "epoch": 4.900269541778976,
-      "grad_norm": 0.6654176115989685,
-      "learning_rate": 0.0003064436049649217,
-      "loss": 3.392,
+      "grad_norm": 0.6276098489761353,
+      "learning_rate": 0.0003064241770102536,
+      "loss": 3.3896,
       "step": 45450
     },
     {
       "epoch": 4.90566037735849,
-      "grad_norm": 0.6970499157905579,
-      "learning_rate": 0.0003061198057204533,
-      "loss": 3.3884,
+      "grad_norm": 0.6750917434692383,
+      "learning_rate": 0.00030610037776578514,
+      "loss": 3.3901,
       "step": 45500
     },
     {
       "epoch": 4.9110512129380055,
-      "grad_norm": 0.6450934410095215,
-      "learning_rate": 0.00030579600647598484,
-      "loss": 3.3903,
+      "grad_norm": 0.6179811954498291,
+      "learning_rate": 0.00030577657852131674,
+      "loss": 3.3895,
       "step": 45550
     },
     {
       "epoch": 4.916442048517521,
-      "grad_norm": 0.6576349139213562,
-      "learning_rate": 0.00030547220723151644,
-      "loss": 3.3872,
+      "grad_norm": 0.6492365002632141,
+      "learning_rate": 0.0003054527792768483,
+      "loss": 3.3835,
       "step": 45600
     },
     {
       "epoch": 4.921832884097035,
-      "grad_norm": 0.6595627069473267,
-      "learning_rate": 0.000305148407987048,
-      "loss": 3.3894,
+      "grad_norm": 0.6423570513725281,
+      "learning_rate": 0.0003051289800323799,
+      "loss": 3.3883,
       "step": 45650
     },
     {
       "epoch": 4.92722371967655,
-      "grad_norm": 0.701417088508606,
-      "learning_rate": 0.00030482460874257955,
-      "loss": 3.3649,
+      "grad_norm": 0.6584356427192688,
+      "learning_rate": 0.00030480518078791145,
+      "loss": 3.3633,
       "step": 45700
     },
     {
       "epoch": 4.932614555256064,
-      "grad_norm": 0.6728206872940063,
-      "learning_rate": 0.00030450080949811115,
-      "loss": 3.3825,
+      "grad_norm": 0.6494635939598083,
+      "learning_rate": 0.00030448138154344305,
+      "loss": 3.3824,
       "step": 45750
     },
     {
       "epoch": 4.938005390835579,
-      "grad_norm": 0.7555509805679321,
-      "learning_rate": 0.0003041770102536427,
-      "loss": 3.4,
+      "grad_norm": 0.6741239428520203,
+      "learning_rate": 0.0003041575822989746,
+      "loss": 3.3991,
       "step": 45800
     },
     {
       "epoch": 4.943396226415095,
-      "grad_norm": 0.725534975528717,
-      "learning_rate": 0.0003038532110091743,
-      "loss": 3.3784,
+      "grad_norm": 0.7292436957359314,
+      "learning_rate": 0.0003038337830545062,
+      "loss": 3.3769,
       "step": 45850
     },
     {
       "epoch": 4.948787061994609,
-      "grad_norm": 0.658740222454071,
-      "learning_rate": 0.00030352941176470586,
-      "loss": 3.3868,
+      "grad_norm": 0.6789447665214539,
+      "learning_rate": 0.00030350998381003776,
+      "loss": 3.3864,
       "step": 45900
     },
     {
       "epoch": 4.954177897574124,
-      "grad_norm": 0.6139926314353943,
-      "learning_rate": 0.00030320561252023746,
-      "loss": 3.377,
+      "grad_norm": 0.6103675961494446,
+      "learning_rate": 0.0003031861845655693,
+      "loss": 3.3783,
       "step": 45950
     },
     {
       "epoch": 4.959568733153639,
-      "grad_norm": 0.6422691345214844,
-      "learning_rate": 0.000302881813275769,
-      "loss": 3.3849,
+      "grad_norm": 0.6245911717414856,
+      "learning_rate": 0.0003028623853211009,
+      "loss": 3.3834,
       "step": 46000
     },
     {
       "epoch": 4.959568733153639,
-      "eval_accuracy": 0.3808301102359604,
-      "eval_loss": 3.4091789722442627,
-      "eval_runtime": 185.1173,
-      "eval_samples_per_second": 97.295,
-      "eval_steps_per_second": 6.083,
+      "eval_accuracy": 0.38076980788322334,
+      "eval_loss": 3.4084885120391846,
+      "eval_runtime": 185.2341,
+      "eval_samples_per_second": 97.234,
+      "eval_steps_per_second": 6.079,
       "step": 46000
     },
     {
       "epoch": 4.964959568733153,
-      "grad_norm": 0.6328104138374329,
-      "learning_rate": 0.0003025580140313006,
-      "loss": 3.3936,
+      "grad_norm": 0.631356954574585,
+      "learning_rate": 0.00030253858607663247,
+      "loss": 3.3948,
       "step": 46050
     },
     {
       "epoch": 4.9703504043126685,
-      "grad_norm": 0.6444491147994995,
-      "learning_rate": 0.00030223421478683217,
-      "loss": 3.3657,
+      "grad_norm": 0.6470606923103333,
+      "learning_rate": 0.00030221478683216407,
+      "loss": 3.3663,
       "step": 46100
     },
     {
       "epoch": 4.975741239892184,
-      "grad_norm": 0.6250355243682861,
-      "learning_rate": 0.00030191041554236366,
-      "loss": 3.3731,
+      "grad_norm": 0.672898530960083,
+      "learning_rate": 0.00030189098758769557,
+      "loss": 3.372,
       "step": 46150
     },
     {
       "epoch": 4.981132075471698,
-      "grad_norm": 0.7009178996086121,
-      "learning_rate": 0.00030158661629789527,
-      "loss": 3.3837,
+      "grad_norm": 0.7242577075958252,
+      "learning_rate": 0.00030156718834322717,
+      "loss": 3.3835,
       "step": 46200
     },
     {
       "epoch": 4.986522911051213,
-      "grad_norm": 0.6598658561706543,
-      "learning_rate": 0.0003012628170534268,
-      "loss": 3.4008,
+      "grad_norm": 0.7056680917739868,
+      "learning_rate": 0.0003012433890987587,
+      "loss": 3.401,
       "step": 46250
     },
     {
       "epoch": 4.991913746630727,
-      "grad_norm": 0.6467294692993164,
-      "learning_rate": 0.0003009390178089584,
-      "loss": 3.3979,
+      "grad_norm": 0.6546418070793152,
+      "learning_rate": 0.0003009195898542903,
+      "loss": 3.395,
       "step": 46300
     },
     {
       "epoch": 4.997304582210242,
-      "grad_norm": 0.6867441534996033,
-      "learning_rate": 0.00030061521856449,
-      "loss": 3.3752,
+      "grad_norm": 0.6470614671707153,
+      "learning_rate": 0.0003005957906098219,
+      "loss": 3.3757,
       "step": 46350
     },
     {
       "epoch": 5.002695417789758,
-      "grad_norm": 0.6885562539100647,
-      "learning_rate": 0.0003002914193200216,
-      "loss": 3.3244,
+      "grad_norm": 0.6490879654884338,
+      "learning_rate": 0.00030027199136535343,
+      "loss": 3.3243,
       "step": 46400
     },
     {
       "epoch": 5.008086253369272,
-      "grad_norm": 0.7105883359909058,
-      "learning_rate": 0.00029996762007555313,
-      "loss": 3.3023,
+      "grad_norm": 0.6910776495933533,
+      "learning_rate": 0.00029994819212088503,
+      "loss": 3.3026,
       "step": 46450
     },
     {
       "epoch": 5.013477088948787,
-      "grad_norm": 0.6555173397064209,
-      "learning_rate": 0.0002996438208310847,
-      "loss": 3.3038,
+      "grad_norm": 0.6369443535804749,
+      "learning_rate": 0.0002996243928764166,
+      "loss": 3.3021,
       "step": 46500
     },
     {
       "epoch": 5.018867924528302,
-      "grad_norm": 0.6800537705421448,
-      "learning_rate": 0.0002993200215866163,
-      "loss": 3.3036,
+      "grad_norm": 0.7196043729782104,
+      "learning_rate": 0.0002993005936319482,
+      "loss": 3.3039,
       "step": 46550
     },
     {
       "epoch": 5.024258760107816,
-      "grad_norm": 0.6979748606681824,
-      "learning_rate": 0.00029899622234214783,
-      "loss": 3.2893,
+      "grad_norm": 0.6704365015029907,
+      "learning_rate": 0.00029897679438747974,
+      "loss": 3.2892,
       "step": 46600
     },
     {
       "epoch": 5.0296495956873315,
-      "grad_norm": 0.6974326372146606,
-      "learning_rate": 0.00029867242309767944,
-      "loss": 3.2922,
+      "grad_norm": 0.6854135990142822,
+      "learning_rate": 0.0002986529951430113,
+      "loss": 3.2931,
       "step": 46650
     },
     {
       "epoch": 5.035040431266847,
-      "grad_norm": 0.6461212635040283,
-      "learning_rate": 0.000298348623853211,
-      "loss": 3.2947,
+      "grad_norm": 0.6373289227485657,
+      "learning_rate": 0.0002983291958985429,
+      "loss": 3.2935,
       "step": 46700
     },
     {
       "epoch": 5.040431266846361,
-      "grad_norm": 0.672947108745575,
-      "learning_rate": 0.00029802482460874254,
-      "loss": 3.3011,
+      "grad_norm": 0.6976351737976074,
+      "learning_rate": 0.00029801187263896385,
+      "loss": 3.3017,
       "step": 46750
     },
     {
       "epoch": 5.045822102425876,
-      "grad_norm": 0.7461097240447998,
-      "learning_rate": 0.0002977010253642741,
-      "loss": 3.3097,
+      "grad_norm": 0.6690797209739685,
+      "learning_rate": 0.0002976880733944954,
+      "loss": 3.3089,
       "step": 46800
     },
     {
       "epoch": 5.051212938005391,
-      "grad_norm": 0.6764471530914307,
-      "learning_rate": 0.0002973772261198057,
-      "loss": 3.3087,
+      "grad_norm": 0.7078843116760254,
+      "learning_rate": 0.00029736427415002695,
+      "loss": 3.3085,
       "step": 46850
     },
     {
       "epoch": 5.056603773584905,
-      "grad_norm": 0.6477118134498596,
-      "learning_rate": 0.00029705342687533725,
-      "loss": 3.3035,
+      "grad_norm": 0.6783893704414368,
+      "learning_rate": 0.0002970404749055585,
+      "loss": 3.3026,
       "step": 46900
     },
     {
       "epoch": 5.061994609164421,
-      "grad_norm": 0.7262566685676575,
-      "learning_rate": 0.00029672962763086885,
-      "loss": 3.3022,
+      "grad_norm": 0.7261761426925659,
+      "learning_rate": 0.0002967166756610901,
+      "loss": 3.3008,
       "step": 46950
     },
     {
       "epoch": 5.067385444743936,
-      "grad_norm": 0.7069379091262817,
-      "learning_rate": 0.0002964058283864004,
-      "loss": 3.311,
+      "grad_norm": 0.6983113884925842,
+      "learning_rate": 0.00029639287641662165,
+      "loss": 3.3112,
       "step": 47000
     },
     {
       "epoch": 5.067385444743936,
-      "eval_accuracy": 0.38110000400929156,
-      "eval_loss": 3.411581039428711,
-      "eval_runtime": 184.7171,
-      "eval_samples_per_second": 97.506,
-      "eval_steps_per_second": 6.096,
+      "eval_accuracy": 0.3807572041482369,
+      "eval_loss": 3.411543369293213,
+      "eval_runtime": 184.8806,
+      "eval_samples_per_second": 97.42,
+      "eval_steps_per_second": 6.09,
       "step": 47000
     },
     {
       "epoch": 5.07277628032345,
-      "grad_norm": 0.684544026851654,
-      "learning_rate": 0.000296082029141932,
-      "loss": 3.3039,
+      "grad_norm": 0.7106756567955017,
+      "learning_rate": 0.00029606907717215326,
+      "loss": 3.3048,
       "step": 47050
     },
     {
       "epoch": 5.078167115902965,
-      "grad_norm": 0.6550946235656738,
-      "learning_rate": 0.00029575822989746356,
-      "loss": 3.2982,
+      "grad_norm": 0.6234951615333557,
+      "learning_rate": 0.0002957452779276848,
+      "loss": 3.2983,
       "step": 47100
     },
     {
       "epoch": 5.083557951482479,
-      "grad_norm": 0.6490722894668579,
-      "learning_rate": 0.0002954344306529951,
-      "loss": 3.3105,
+      "grad_norm": 0.6151743531227112,
+      "learning_rate": 0.00029542147868321636,
+      "loss": 3.3121,
       "step": 47150
     },
     {
       "epoch": 5.0889487870619945,
-      "grad_norm": 0.6412487030029297,
-      "learning_rate": 0.00029511063140852666,
-      "loss": 3.3115,
+      "grad_norm": 0.6277012825012207,
+      "learning_rate": 0.00029509767943874796,
+      "loss": 3.3105,
       "step": 47200
     },
     {
       "epoch": 5.09433962264151,
-      "grad_norm": 0.6618337035179138,
-      "learning_rate": 0.00029478683216405826,
-      "loss": 3.2947,
+      "grad_norm": 0.6742247343063354,
+      "learning_rate": 0.0002947738801942795,
+      "loss": 3.2964,
       "step": 47250
     },
     {
       "epoch": 5.099730458221024,
-      "grad_norm": 0.7021799087524414,
-      "learning_rate": 0.0002944630329195898,
-      "loss": 3.3195,
+      "grad_norm": 0.6989336609840393,
+      "learning_rate": 0.0002944500809498111,
+      "loss": 3.3197,
       "step": 47300
     },
     {
       "epoch": 5.105121293800539,
-      "grad_norm": 0.6509613394737244,
-      "learning_rate": 0.0002941392336751214,
-      "loss": 3.32,
+      "grad_norm": 0.6480673551559448,
+      "learning_rate": 0.00029412628170534267,
+      "loss": 3.3195,
       "step": 47350
     },
     {
       "epoch": 5.110512129380054,
-      "grad_norm": 0.6623237133026123,
-      "learning_rate": 0.00029381543443065297,
-      "loss": 3.301,
+      "grad_norm": 0.6779480576515198,
+      "learning_rate": 0.0002938024824608742,
+      "loss": 3.3009,
       "step": 47400
     },
     {
       "epoch": 5.115902964959568,
-      "grad_norm": 0.6974384784698486,
-      "learning_rate": 0.0002934916351861846,
-      "loss": 3.3038,
+      "grad_norm": 0.6613697409629822,
+      "learning_rate": 0.00029347868321640577,
+      "loss": 3.3048,
       "step": 47450
     },
     {
       "epoch": 5.121293800539084,
-      "grad_norm": 0.6917511820793152,
-      "learning_rate": 0.0002931678359417161,
-      "loss": 3.2947,
+      "grad_norm": 0.6762946248054504,
+      "learning_rate": 0.0002931548839719374,
+      "loss": 3.2942,
       "step": 47500
     },
     {
       "epoch": 5.126684636118599,
-      "grad_norm": 0.6955335736274719,
-      "learning_rate": 0.0002928440366972477,
-      "loss": 3.3021,
+      "grad_norm": 0.6452640295028687,
+      "learning_rate": 0.0002928310847274689,
+      "loss": 3.3027,
       "step": 47550
     },
     {
       "epoch": 5.132075471698113,
-      "grad_norm": 0.7200063467025757,
-      "learning_rate": 0.0002925202374527792,
-      "loss": 3.2928,
+      "grad_norm": 0.7308632731437683,
+      "learning_rate": 0.00029250728548300053,
+      "loss": 3.2931,
       "step": 47600
     },
     {
       "epoch": 5.137466307277628,
-      "grad_norm": 0.716500461101532,
-      "learning_rate": 0.00029219643820831083,
-      "loss": 3.308,
+      "grad_norm": 0.7426350712776184,
+      "learning_rate": 0.0002921834862385321,
+      "loss": 3.3076,
       "step": 47650
     },
     {
       "epoch": 5.142857142857143,
-      "grad_norm": 0.6869317889213562,
-      "learning_rate": 0.0002918791149487318,
-      "loss": 3.3217,
+      "grad_norm": 0.6810452342033386,
+      "learning_rate": 0.0002918596869940637,
+      "loss": 3.3227,
       "step": 47700
     },
     {
       "epoch": 5.1482479784366575,
-      "grad_norm": 0.7114856243133545,
-      "learning_rate": 0.00029155531570426333,
-      "loss": 3.3117,
+      "grad_norm": 0.6850289702415466,
+      "learning_rate": 0.0002915358877495952,
+      "loss": 3.3086,
       "step": 47750
     },
     {
       "epoch": 5.153638814016173,
-      "grad_norm": 0.6355905532836914,
-      "learning_rate": 0.00029123151645979494,
-      "loss": 3.2979,
+      "grad_norm": 0.6442118287086487,
+      "learning_rate": 0.0002912120885051268,
+      "loss": 3.2973,
       "step": 47800
     },
     {
       "epoch": 5.159029649595688,
-      "grad_norm": 0.6989576816558838,
-      "learning_rate": 0.0002909077172153265,
-      "loss": 3.3042,
+      "grad_norm": 0.702198326587677,
+      "learning_rate": 0.00029088828926065834,
+      "loss": 3.3049,
       "step": 47850
     },
     {
       "epoch": 5.164420485175202,
-      "grad_norm": 0.716057300567627,
-      "learning_rate": 0.00029058391797085804,
-      "loss": 3.3303,
+      "grad_norm": 0.692348062992096,
+      "learning_rate": 0.00029056449001618994,
+      "loss": 3.3313,
       "step": 47900
     },
     {
       "epoch": 5.169811320754717,
-      "grad_norm": 0.6772635579109192,
-      "learning_rate": 0.0002902601187263896,
-      "loss": 3.3113,
+      "grad_norm": 0.6892920136451721,
+      "learning_rate": 0.0002902406907717215,
+      "loss": 3.3101,
       "step": 47950
     },
     {
       "epoch": 5.175202156334231,
-      "grad_norm": 0.6645113229751587,
-      "learning_rate": 0.0002899363194819212,
-      "loss": 3.3052,
+      "grad_norm": 0.7259092330932617,
+      "learning_rate": 0.0002899168915272531,
+      "loss": 3.3048,
       "step": 48000
     },
     {
       "epoch": 5.175202156334231,
-      "eval_accuracy": 0.3809431092392874,
-      "eval_loss": 3.411557912826538,
-      "eval_runtime": 184.8975,
-      "eval_samples_per_second": 97.411,
-      "eval_steps_per_second": 6.09,
+      "eval_accuracy": 0.3808885654896046,
+      "eval_loss": 3.4122819900512695,
+      "eval_runtime": 185.1757,
+      "eval_samples_per_second": 97.264,
+      "eval_steps_per_second": 6.081,
       "step": 48000
     },
     {
       "epoch": 5.180592991913747,
-      "grad_norm": 0.6688160300254822,
-      "learning_rate": 0.00028961252023745274,
-      "loss": 3.3126,
+      "grad_norm": 0.6801049113273621,
+      "learning_rate": 0.00028959956826767405,
+      "loss": 3.3147,
       "step": 48050
     },
     {
       "epoch": 5.185983827493262,
-      "grad_norm": 0.6707350611686707,
-      "learning_rate": 0.00028928872099298435,
-      "loss": 3.3017,
+      "grad_norm": 0.7039002776145935,
+      "learning_rate": 0.00028927576902320555,
+      "loss": 3.3004,
       "step": 48100
     },
     {
       "epoch": 5.191374663072776,
-      "grad_norm": 0.6549227833747864,
-      "learning_rate": 0.0002889649217485159,
-      "loss": 3.3112,
+      "grad_norm": 0.6632753610610962,
+      "learning_rate": 0.00028895196977873715,
+      "loss": 3.31,
       "step": 48150
     },
     {
       "epoch": 5.196765498652291,
-      "grad_norm": 0.6969558596611023,
-      "learning_rate": 0.00028864112250404745,
-      "loss": 3.3247,
+      "grad_norm": 0.6418648362159729,
+      "learning_rate": 0.0002886281705342687,
+      "loss": 3.326,
       "step": 48200
     },
     {
       "epoch": 5.202156334231806,
-      "grad_norm": 0.7048711180686951,
-      "learning_rate": 0.00028831732325957906,
-      "loss": 3.317,
+      "grad_norm": 0.6770094037055969,
+      "learning_rate": 0.0002883043712898003,
+      "loss": 3.3181,
       "step": 48250
     },
     {
       "epoch": 5.2075471698113205,
-      "grad_norm": 0.7173691987991333,
-      "learning_rate": 0.0002879935240151106,
-      "loss": 3.3182,
+      "grad_norm": 0.6800588965415955,
+      "learning_rate": 0.00028798057204533186,
+      "loss": 3.3185,
       "step": 48300
     },
     {
       "epoch": 5.212938005390836,
-      "grad_norm": 0.6918731927871704,
-      "learning_rate": 0.00028766972477064216,
-      "loss": 3.3264,
+      "grad_norm": 0.649732768535614,
+      "learning_rate": 0.00028765677280086346,
+      "loss": 3.3252,
       "step": 48350
     },
     {
       "epoch": 5.218328840970351,
-      "grad_norm": 0.6857321262359619,
-      "learning_rate": 0.00028734592552617376,
-      "loss": 3.3196,
+      "grad_norm": 0.714310348033905,
+      "learning_rate": 0.000287332973556395,
+      "loss": 3.3191,
       "step": 48400
     },
     {
       "epoch": 5.223719676549865,
-      "grad_norm": 0.6688578724861145,
-      "learning_rate": 0.0002870221262817053,
-      "loss": 3.3114,
+      "grad_norm": 0.6347412467002869,
+      "learning_rate": 0.0002870091743119266,
+      "loss": 3.3109,
       "step": 48450
     },
     {
       "epoch": 5.22911051212938,
-      "grad_norm": 0.7027651071548462,
-      "learning_rate": 0.00028669832703723686,
-      "loss": 3.3448,
+      "grad_norm": 0.6917827725410461,
+      "learning_rate": 0.00028668537506745817,
+      "loss": 3.3453,
       "step": 48500
     },
     {
       "epoch": 5.234501347708895,
-      "grad_norm": 0.6578187346458435,
-      "learning_rate": 0.00028637452779276847,
-      "loss": 3.3182,
+      "grad_norm": 0.650833249092102,
+      "learning_rate": 0.0002863615758229897,
+      "loss": 3.3185,
       "step": 48550
     },
     {
       "epoch": 5.2398921832884096,
-      "grad_norm": 0.6483734846115112,
-      "learning_rate": 0.0002860507285483,
-      "loss": 3.3184,
+      "grad_norm": 0.6491532325744629,
+      "learning_rate": 0.00028603777657852127,
+      "loss": 3.3186,
       "step": 48600
     },
     {
       "epoch": 5.245283018867925,
-      "grad_norm": 0.6719345450401306,
-      "learning_rate": 0.0002857269293038316,
-      "loss": 3.3152,
+      "grad_norm": 0.6467685699462891,
+      "learning_rate": 0.0002857139773340529,
+      "loss": 3.3169,
       "step": 48650
     },
     {
       "epoch": 5.250673854447439,
-      "grad_norm": 0.7231610417366028,
-      "learning_rate": 0.0002854031300593632,
-      "loss": 3.3191,
+      "grad_norm": 0.7819921970367432,
+      "learning_rate": 0.0002853966540744738,
+      "loss": 3.3204,
       "step": 48700
     },
     {
       "epoch": 5.256064690026954,
-      "grad_norm": 0.7005861401557922,
-      "learning_rate": 0.0002850793308148947,
-      "loss": 3.3382,
+      "grad_norm": 0.6711828112602234,
+      "learning_rate": 0.0002850728548300054,
+      "loss": 3.3389,
       "step": 48750
     },
     {
       "epoch": 5.261455525606469,
-      "grad_norm": 0.6894075274467468,
-      "learning_rate": 0.0002847555315704263,
-      "loss": 3.3371,
+      "grad_norm": 0.6818621158599854,
+      "learning_rate": 0.000284749055585537,
+      "loss": 3.3368,
       "step": 48800
     },
     {
       "epoch": 5.2668463611859835,
-      "grad_norm": 0.6957369446754456,
-      "learning_rate": 0.0002844317323259579,
-      "loss": 3.3153,
+      "grad_norm": 0.6671029925346375,
+      "learning_rate": 0.0002844252563410685,
+      "loss": 3.3145,
       "step": 48850
     },
     {
       "epoch": 5.272237196765499,
-      "grad_norm": 0.7194564342498779,
-      "learning_rate": 0.00028410793308148943,
-      "loss": 3.313,
+      "grad_norm": 0.7544181942939758,
+      "learning_rate": 0.0002841014570966001,
+      "loss": 3.314,
       "step": 48900
     },
     {
       "epoch": 5.277628032345014,
-      "grad_norm": 0.677614688873291,
-      "learning_rate": 0.00028378413383702103,
-      "loss": 3.3354,
+      "grad_norm": 0.6877387762069702,
+      "learning_rate": 0.00028377765785213163,
+      "loss": 3.3361,
       "step": 48950
     },
     {
       "epoch": 5.283018867924528,
-      "grad_norm": 0.6809502243995667,
-      "learning_rate": 0.0002834603345925526,
-      "loss": 3.3259,
+      "grad_norm": 0.6355829834938049,
+      "learning_rate": 0.00028345385860766324,
+      "loss": 3.3249,
       "step": 49000
     },
     {
       "epoch": 5.283018867924528,
-      "eval_accuracy": 0.38146768538165576,
-      "eval_loss": 3.406904458999634,
-      "eval_runtime": 185.2899,
-      "eval_samples_per_second": 97.204,
-      "eval_steps_per_second": 6.077,
+      "eval_accuracy": 0.38155525960923425,
+      "eval_loss": 3.4067444801330566,
+      "eval_runtime": 184.7957,
+      "eval_samples_per_second": 97.464,
+      "eval_steps_per_second": 6.093,
       "step": 49000
     },
     {
       "epoch": 5.288409703504043,
-      "grad_norm": 0.661638617515564,
-      "learning_rate": 0.0002831365353480842,
-      "loss": 3.3349,
+      "grad_norm": 0.6431439518928528,
+      "learning_rate": 0.0002831300593631948,
+      "loss": 3.3335,
       "step": 49050
     },
     {
       "epoch": 5.293800539083558,
-      "grad_norm": 0.6964898109436035,
-      "learning_rate": 0.00028281273610361574,
-      "loss": 3.3309,
+      "grad_norm": 0.6457442045211792,
+      "learning_rate": 0.0002828062601187264,
+      "loss": 3.3295,
       "step": 49100
     },
     {
       "epoch": 5.2991913746630726,
-      "grad_norm": 0.6699468493461609,
-      "learning_rate": 0.00028248893685914734,
-      "loss": 3.3279,
+      "grad_norm": 0.7099213600158691,
+      "learning_rate": 0.00028248246087425794,
+      "loss": 3.3284,
       "step": 49150
     },
     {
       "epoch": 5.304582210242588,
-      "grad_norm": 0.6527925729751587,
-      "learning_rate": 0.00028216513761467884,
-      "loss": 3.3377,
+      "grad_norm": 0.6772319674491882,
+      "learning_rate": 0.0002821586616297895,
+      "loss": 3.3365,
       "step": 49200
     },
     {
       "epoch": 5.309973045822103,
-      "grad_norm": 0.6300584673881531,
-      "learning_rate": 0.00028184133837021045,
-      "loss": 3.3239,
+      "grad_norm": 0.6409128308296204,
+      "learning_rate": 0.0002818348623853211,
+      "loss": 3.3249,
       "step": 49250
     },
     {
       "epoch": 5.315363881401617,
-      "grad_norm": 0.727450430393219,
-      "learning_rate": 0.000281517539125742,
-      "loss": 3.3253,
+      "grad_norm": 0.7182660102844238,
+      "learning_rate": 0.00028151106314085265,
+      "loss": 3.3255,
       "step": 49300
     },
     {
       "epoch": 5.320754716981132,
-      "grad_norm": 0.7141486406326294,
-      "learning_rate": 0.0002811937398812736,
-      "loss": 3.3347,
+      "grad_norm": 0.7561542987823486,
+      "learning_rate": 0.0002811872638963842,
+      "loss": 3.3359,
       "step": 49350
     },
     {
       "epoch": 5.3261455525606465,
-      "grad_norm": 0.8665236830711365,
-      "learning_rate": 0.00028086994063680515,
-      "loss": 3.3233,
+      "grad_norm": 0.675321638584137,
+      "learning_rate": 0.0002808634646519158,
+      "loss": 3.3237,
       "step": 49400
     },
     {
       "epoch": 5.331536388140162,
-      "grad_norm": 0.6957372426986694,
-      "learning_rate": 0.00028054614139233676,
-      "loss": 3.3173,
+      "grad_norm": 0.6842005252838135,
+      "learning_rate": 0.00028053966540744736,
+      "loss": 3.3151,
       "step": 49450
     },
     {
       "epoch": 5.336927223719677,
-      "grad_norm": 0.6600592732429504,
-      "learning_rate": 0.0002802223421478683,
-      "loss": 3.3405,
+      "grad_norm": 0.7084938287734985,
+      "learning_rate": 0.0002802158661629789,
+      "loss": 3.3432,
       "step": 49500
     },
     {
       "epoch": 5.342318059299191,
-      "grad_norm": 0.6827237010002136,
-      "learning_rate": 0.00027989854290339986,
-      "loss": 3.3152,
+      "grad_norm": 0.6873745322227478,
+      "learning_rate": 0.0002798920669185105,
+      "loss": 3.3164,
       "step": 49550
     },
     {
       "epoch": 5.347708894878706,
-      "grad_norm": 0.6884130835533142,
-      "learning_rate": 0.0002795747436589314,
-      "loss": 3.3326,
+      "grad_norm": 0.6905918121337891,
+      "learning_rate": 0.00027956826767404206,
+      "loss": 3.3339,
       "step": 49600
     },
     {
       "epoch": 5.353099730458221,
-      "grad_norm": 0.701141357421875,
-      "learning_rate": 0.000279250944414463,
-      "loss": 3.3293,
+      "grad_norm": 0.6891156435012817,
+      "learning_rate": 0.00027924446842957367,
+      "loss": 3.3292,
       "step": 49650
     },
     {
       "epoch": 5.3584905660377355,
-      "grad_norm": 0.6687909960746765,
-      "learning_rate": 0.00027892714516999456,
-      "loss": 3.3375,
+      "grad_norm": 0.6603860855102539,
+      "learning_rate": 0.0002789206691851052,
+      "loss": 3.3368,
       "step": 49700
     },
     {
       "epoch": 5.363881401617251,
-      "grad_norm": 0.6871972680091858,
-      "learning_rate": 0.00027860334592552617,
-      "loss": 3.3208,
+      "grad_norm": 0.676149308681488,
+      "learning_rate": 0.00027859686994063677,
+      "loss": 3.3218,
       "step": 49750
     },
     {
       "epoch": 5.369272237196766,
-      "grad_norm": 0.712066113948822,
-      "learning_rate": 0.0002782795466810577,
-      "loss": 3.3236,
+      "grad_norm": 0.7008675932884216,
+      "learning_rate": 0.0002782730706961683,
+      "loss": 3.3254,
       "step": 49800
     },
     {
       "epoch": 5.37466307277628,
-      "grad_norm": 0.6459314823150635,
-      "learning_rate": 0.00027795574743658927,
-      "loss": 3.3382,
+      "grad_norm": 0.63350909948349,
+      "learning_rate": 0.0002779492714516999,
+      "loss": 3.3386,
       "step": 49850
     },
     {
       "epoch": 5.380053908355795,
-      "grad_norm": 0.7322667241096497,
-      "learning_rate": 0.0002776384241770102,
-      "loss": 3.3321,
+      "grad_norm": 0.6737155914306641,
+      "learning_rate": 0.00027762547220723147,
+      "loss": 3.3339,
       "step": 49900
     },
     {
       "epoch": 5.38544474393531,
-      "grad_norm": 0.6739450097084045,
-      "learning_rate": 0.00027731462493254177,
-      "loss": 3.3187,
+      "grad_norm": 0.6853422522544861,
+      "learning_rate": 0.0002773016729627631,
+      "loss": 3.3183,
       "step": 49950
     },
     {
       "epoch": 5.390835579514825,
-      "grad_norm": 0.6529873013496399,
-      "learning_rate": 0.0002769908256880734,
-      "loss": 3.3392,
+      "grad_norm": 0.6315034627914429,
+      "learning_rate": 0.00027697787371829463,
+      "loss": 3.339,
       "step": 50000
     },
     {
       "epoch": 5.390835579514825,
-      "eval_accuracy": 0.3821004798002873,
-      "eval_loss": 3.400057792663574,
-      "eval_runtime": 184.8218,
-      "eval_samples_per_second": 97.451,
-      "eval_steps_per_second": 6.092,
+      "eval_accuracy": 0.3820815741978076,
+      "eval_loss": 3.399562358856201,
+      "eval_runtime": 185.1089,
+      "eval_samples_per_second": 97.299,
+      "eval_steps_per_second": 6.083,
       "step": 50000
     },
     {
       "epoch": 5.39622641509434,
-      "grad_norm": 0.7031570076942444,
-      "learning_rate": 0.00027666702644360493,
-      "loss": 3.3487,
+      "grad_norm": 0.6739052534103394,
+      "learning_rate": 0.00027665407447382623,
+      "loss": 3.3508,
       "step": 50050
     },
     {
       "epoch": 5.401617250673855,
-      "grad_norm": 0.6449311375617981,
-      "learning_rate": 0.00027634322719913653,
-      "loss": 3.3411,
+      "grad_norm": 0.6296254992485046,
+      "learning_rate": 0.0002763302752293578,
+      "loss": 3.3394,
       "step": 50100
     },
     {
       "epoch": 5.407008086253369,
-      "grad_norm": 0.6960560083389282,
-      "learning_rate": 0.0002760194279546681,
-      "loss": 3.3318,
+      "grad_norm": 0.7053967118263245,
+      "learning_rate": 0.00027600647598488933,
+      "loss": 3.3309,
       "step": 50150
     },
     {
       "epoch": 5.412398921832884,
-      "grad_norm": 0.6679825186729431,
-      "learning_rate": 0.00027569562871019963,
-      "loss": 3.3136,
+      "grad_norm": 0.6510531902313232,
+      "learning_rate": 0.0002756826767404209,
+      "loss": 3.3147,
       "step": 50200
     },
     {
       "epoch": 5.4177897574123985,
-      "grad_norm": 0.6533336639404297,
-      "learning_rate": 0.00027537182946573124,
-      "loss": 3.3423,
+      "grad_norm": 0.6612011194229126,
+      "learning_rate": 0.0002753588774959525,
+      "loss": 3.3438,
       "step": 50250
     },
     {
       "epoch": 5.423180592991914,
-      "grad_norm": 0.6472018361091614,
-      "learning_rate": 0.0002750480302212628,
-      "loss": 3.3375,
+      "grad_norm": 0.6302815079689026,
+      "learning_rate": 0.00027503507825148404,
+      "loss": 3.3371,
       "step": 50300
     },
     {
       "epoch": 5.428571428571429,
-      "grad_norm": 0.7485306859016418,
-      "learning_rate": 0.0002747242309767944,
-      "loss": 3.325,
+      "grad_norm": 0.6635833382606506,
+      "learning_rate": 0.00027471127900701564,
+      "loss": 3.3237,
       "step": 50350
     },
     {
       "epoch": 5.433962264150943,
-      "grad_norm": 1.076799750328064,
-      "learning_rate": 0.00027440043173232594,
-      "loss": 3.3463,
+      "grad_norm": 0.7546477317810059,
+      "learning_rate": 0.0002743874797625472,
+      "loss": 3.3456,
       "step": 50400
     },
     {
       "epoch": 5.439353099730458,
-      "grad_norm": 0.7454240322113037,
-      "learning_rate": 0.0002740766324878575,
-      "loss": 3.3443,
+      "grad_norm": 0.73973548412323,
+      "learning_rate": 0.0002740636805180788,
+      "loss": 3.3419,
       "step": 50450
     },
     {
       "epoch": 5.444743935309973,
-      "grad_norm": 0.6593639254570007,
-      "learning_rate": 0.00027375283324338905,
+      "grad_norm": 0.6448124051094055,
+      "learning_rate": 0.00027373988127361035,
       "loss": 3.3284,
       "step": 50500
     },
     {
       "epoch": 5.450134770889488,
-      "grad_norm": 0.657081663608551,
-      "learning_rate": 0.00027342903399892065,
-      "loss": 3.324,
+      "grad_norm": 0.6900261044502258,
+      "learning_rate": 0.0002734160820291419,
+      "loss": 3.3253,
       "step": 50550
     },
     {
       "epoch": 5.455525606469003,
-      "grad_norm": 0.7052240371704102,
-      "learning_rate": 0.0002731052347544522,
-      "loss": 3.3394,
+      "grad_norm": 0.6863948106765747,
+      "learning_rate": 0.00027309228278467345,
+      "loss": 3.3399,
       "step": 50600
     },
     {
       "epoch": 5.460916442048518,
-      "grad_norm": 0.8208935260772705,
-      "learning_rate": 0.0002727814355099838,
-      "loss": 3.3323,
+      "grad_norm": 0.8710711598396301,
+      "learning_rate": 0.00027276848354020506,
+      "loss": 3.3339,
       "step": 50650
     },
     {
       "epoch": 5.466307277628032,
-      "grad_norm": 0.6855742931365967,
-      "learning_rate": 0.00027245763626551536,
-      "loss": 3.3259,
+      "grad_norm": 0.7047181129455566,
+      "learning_rate": 0.0002724446842957366,
+      "loss": 3.3241,
       "step": 50700
     },
     {
       "epoch": 5.471698113207547,
-      "grad_norm": 0.6824098229408264,
-      "learning_rate": 0.00027213383702104696,
-      "loss": 3.3114,
+      "grad_norm": 0.6998323202133179,
+      "learning_rate": 0.0002721208850512682,
+      "loss": 3.3125,
       "step": 50750
     },
     {
       "epoch": 5.4770889487870615,
-      "grad_norm": 0.7104268074035645,
-      "learning_rate": 0.00027181003777657846,
-      "loss": 3.3402,
+      "grad_norm": 0.6888628005981445,
+      "learning_rate": 0.00027179708580679976,
+      "loss": 3.3405,
       "step": 50800
     },
     {
       "epoch": 5.482479784366577,
-      "grad_norm": 0.7086119651794434,
-      "learning_rate": 0.00027148623853211006,
-      "loss": 3.3259,
+      "grad_norm": 0.6762120723724365,
+      "learning_rate": 0.0002714732865623313,
+      "loss": 3.3244,
       "step": 50850
     },
     {
       "epoch": 5.487870619946092,
-      "grad_norm": 0.6901703476905823,
-      "learning_rate": 0.0002711624392876416,
-      "loss": 3.3348,
+      "grad_norm": 0.6589097380638123,
+      "learning_rate": 0.0002711494873178629,
+      "loss": 3.335,
       "step": 50900
     },
     {
       "epoch": 5.493261455525606,
-      "grad_norm": 0.70135098695755,
-      "learning_rate": 0.0002708386400431732,
-      "loss": 3.349,
+      "grad_norm": 0.6884141564369202,
+      "learning_rate": 0.00027082568807339447,
+      "loss": 3.3512,
       "step": 50950
     },
     {
       "epoch": 5.498652291105121,
-      "grad_norm": 0.6997105479240417,
-      "learning_rate": 0.00027051484079870477,
+      "grad_norm": 0.6888891458511353,
+      "learning_rate": 0.000270501888828926,
       "loss": 3.3135,
       "step": 51000
     },
     {
       "epoch": 5.498652291105121,
-      "eval_accuracy": 0.3822508553970225,
-      "eval_loss": 3.3968350887298584,
-      "eval_runtime": 185.1499,
-      "eval_samples_per_second": 97.278,
-      "eval_steps_per_second": 6.082,
+      "eval_accuracy": 0.3822205412413223,
+      "eval_loss": 3.397444248199463,
+      "eval_runtime": 185.2714,
+      "eval_samples_per_second": 97.214,
+      "eval_steps_per_second": 6.078,
       "step": 51000
     },
     {
       "epoch": 5.504043126684636,
-      "grad_norm": 0.7074891328811646,
-      "learning_rate": 0.00027019104155423637,
-      "loss": 3.3215,
+      "grad_norm": 0.6714994311332703,
+      "learning_rate": 0.0002701780895844576,
+      "loss": 3.321,
       "step": 51050
     },
     {
       "epoch": 5.509433962264151,
-      "grad_norm": 0.6949896216392517,
-      "learning_rate": 0.0002698672423097679,
-      "loss": 3.3343,
+      "grad_norm": 0.6546480655670166,
+      "learning_rate": 0.0002698542903399892,
+      "loss": 3.3323,
       "step": 51100
     },
     {
       "epoch": 5.514824797843666,
-      "grad_norm": 0.717146635055542,
-      "learning_rate": 0.00026954344306529953,
-      "loss": 3.349,
+      "grad_norm": 0.696523129940033,
+      "learning_rate": 0.0002695304910955207,
+      "loss": 3.3496,
       "step": 51150
     },
     {
       "epoch": 5.520215633423181,
-      "grad_norm": 0.673259437084198,
-      "learning_rate": 0.0002692196438208311,
-      "loss": 3.3233,
+      "grad_norm": 0.6788317561149597,
+      "learning_rate": 0.00026920669185105233,
+      "loss": 3.3245,
       "step": 51200
     },
     {
       "epoch": 5.525606469002695,
-      "grad_norm": 0.7613580226898193,
-      "learning_rate": 0.00026889584457636263,
-      "loss": 3.3286,
+      "grad_norm": 0.7704463601112366,
+      "learning_rate": 0.0002688828926065839,
+      "loss": 3.3297,
       "step": 51250
     },
     {
       "epoch": 5.53099730458221,
-      "grad_norm": 0.6700328588485718,
-      "learning_rate": 0.0002685720453318942,
-      "loss": 3.3417,
+      "grad_norm": 0.6638984680175781,
+      "learning_rate": 0.0002685590933621155,
+      "loss": 3.3427,
       "step": 51300
     },
     {
       "epoch": 5.536388140161725,
-      "grad_norm": 0.6604465842247009,
-      "learning_rate": 0.0002682482460874258,
-      "loss": 3.321,
+      "grad_norm": 0.6343899369239807,
+      "learning_rate": 0.00026823529411764704,
+      "loss": 3.3216,
       "step": 51350
     },
     {
       "epoch": 5.54177897574124,
-      "grad_norm": 0.7246283888816833,
-      "learning_rate": 0.00026792444684295733,
-      "loss": 3.3297,
+      "grad_norm": 0.7129838466644287,
+      "learning_rate": 0.0002679114948731786,
+      "loss": 3.33,
       "step": 51400
     },
     {
       "epoch": 5.547169811320755,
-      "grad_norm": 0.7392826676368713,
-      "learning_rate": 0.00026760064759848894,
-      "loss": 3.3475,
+      "grad_norm": 0.717201828956604,
+      "learning_rate": 0.0002675876956287102,
+      "loss": 3.3476,
       "step": 51450
     },
     {
       "epoch": 5.55256064690027,
-      "grad_norm": 0.6978980302810669,
-      "learning_rate": 0.0002672768483540205,
-      "loss": 3.3311,
+      "grad_norm": 0.6938988566398621,
+      "learning_rate": 0.00026726389638424174,
+      "loss": 3.334,
       "step": 51500
     },
     {
       "epoch": 5.557951482479784,
-      "grad_norm": 0.7010505199432373,
-      "learning_rate": 0.00026695304910955204,
-      "loss": 3.3245,
+      "grad_norm": 0.6860924363136292,
+      "learning_rate": 0.0002669400971397733,
+      "loss": 3.3264,
       "step": 51550
     },
     {
       "epoch": 5.563342318059299,
-      "grad_norm": 0.7545852661132812,
-      "learning_rate": 0.00026662924986508365,
-      "loss": 3.3501,
+      "grad_norm": 0.6723926663398743,
+      "learning_rate": 0.0002666162978953049,
+      "loss": 3.3517,
       "step": 51600
     },
     {
       "epoch": 5.568733153638814,
-      "grad_norm": 0.6517020463943481,
-      "learning_rate": 0.0002663054506206152,
-      "loss": 3.3475,
+      "grad_norm": 0.634928822517395,
+      "learning_rate": 0.00026629249865083645,
+      "loss": 3.3487,
       "step": 51650
     },
     {
       "epoch": 5.574123989218329,
-      "grad_norm": 0.7191069722175598,
-      "learning_rate": 0.00026598165137614675,
-      "loss": 3.3301,
+      "grad_norm": 0.709865391254425,
+      "learning_rate": 0.00026596869940636805,
+      "loss": 3.3307,
       "step": 51700
     },
     {
       "epoch": 5.579514824797844,
-      "grad_norm": 0.6981732845306396,
-      "learning_rate": 0.00026565785213167835,
-      "loss": 3.3249,
+      "grad_norm": 0.6830937266349792,
+      "learning_rate": 0.0002656449001618996,
+      "loss": 3.3239,
       "step": 51750
     },
     {
       "epoch": 5.584905660377358,
-      "grad_norm": 0.7021121382713318,
-      "learning_rate": 0.0002653340528872099,
-      "loss": 3.3381,
+      "grad_norm": 0.7171959280967712,
+      "learning_rate": 0.0002653211009174312,
+      "loss": 3.3398,
       "step": 51800
     },
     {
       "epoch": 5.590296495956873,
-      "grad_norm": 0.7237430810928345,
-      "learning_rate": 0.00026501025364274145,
-      "loss": 3.3342,
+      "grad_norm": 0.6931149363517761,
+      "learning_rate": 0.0002649973016729627,
+      "loss": 3.336,
       "step": 51850
     },
     {
       "epoch": 5.595687331536388,
-      "grad_norm": 0.6834338903427124,
-      "learning_rate": 0.00026468645439827306,
-      "loss": 3.3419,
+      "grad_norm": 0.6825995445251465,
+      "learning_rate": 0.0002646735024284943,
+      "loss": 3.3417,
       "step": 51900
     },
     {
       "epoch": 5.601078167115903,
-      "grad_norm": 0.6926736831665039,
-      "learning_rate": 0.0002643626551538046,
-      "loss": 3.3299,
+      "grad_norm": 0.6648308038711548,
+      "learning_rate": 0.00026434970318402586,
+      "loss": 3.3304,
       "step": 51950
     },
     {
       "epoch": 5.606469002695418,
-      "grad_norm": 0.696631908416748,
-      "learning_rate": 0.0002640388559093362,
-      "loss": 3.3244,
+      "grad_norm": 0.6847530007362366,
+      "learning_rate": 0.00026402590393955746,
+      "loss": 3.3253,
       "step": 52000
     },
     {
       "epoch": 5.606469002695418,
-      "eval_accuracy": 0.38270100431123794,
-      "eval_loss": 3.3929481506347656,
-      "eval_runtime": 184.8201,
-      "eval_samples_per_second": 97.452,
-      "eval_steps_per_second": 6.092,
+      "eval_accuracy": 0.382617232934733,
+      "eval_loss": 3.393908739089966,
+      "eval_runtime": 185.0037,
+      "eval_samples_per_second": 97.355,
+      "eval_steps_per_second": 6.086,
       "step": 52000
     },
     {
       "epoch": 5.611859838274933,
-      "grad_norm": 0.6704844832420349,
-      "learning_rate": 0.00026371505666486776,
-      "loss": 3.3505,
+      "grad_norm": 0.6547165513038635,
+      "learning_rate": 0.000263702104695089,
+      "loss": 3.3494,
       "step": 52050
     },
     {
       "epoch": 5.617250673854447,
-      "grad_norm": 0.7229808568954468,
-      "learning_rate": 0.0002633912574203993,
-      "loss": 3.3271,
+      "grad_norm": 0.6862826347351074,
+      "learning_rate": 0.0002633783054506206,
+      "loss": 3.3276,
       "step": 52100
     },
     {
       "epoch": 5.622641509433962,
-      "grad_norm": 0.7184265851974487,
-      "learning_rate": 0.00026306745817593086,
-      "loss": 3.3306,
+      "grad_norm": 0.7386119365692139,
+      "learning_rate": 0.00026305450620615217,
+      "loss": 3.3311,
       "step": 52150
     },
     {
       "epoch": 5.628032345013477,
-      "grad_norm": 0.6913803219795227,
-      "learning_rate": 0.00026274365893146247,
-      "loss": 3.326,
+      "grad_norm": 0.6849868297576904,
+      "learning_rate": 0.0002627307069616837,
+      "loss": 3.3285,
       "step": 52200
     },
     {
       "epoch": 5.633423180592992,
-      "grad_norm": 0.6871593594551086,
-      "learning_rate": 0.000262419859686994,
-      "loss": 3.328,
+      "grad_norm": 0.7046032547950745,
+      "learning_rate": 0.0002624069077172153,
+      "loss": 3.3305,
       "step": 52250
     },
     {
       "epoch": 5.638814016172507,
-      "grad_norm": 0.6831674575805664,
-      "learning_rate": 0.0002620960604425256,
-      "loss": 3.3284,
+      "grad_norm": 0.6594682335853577,
+      "learning_rate": 0.0002620831084727469,
+      "loss": 3.3306,
       "step": 52300
     },
     {
       "epoch": 5.644204851752022,
-      "grad_norm": 0.7181283235549927,
-      "learning_rate": 0.0002617722611980572,
+      "grad_norm": 0.6664849519729614,
+      "learning_rate": 0.0002617593092282784,
       "loss": 3.3381,
       "step": 52350
     },
     {
       "epoch": 5.649595687331536,
-      "grad_norm": 0.6951797604560852,
-      "learning_rate": 0.0002614484619535888,
-      "loss": 3.3242,
+      "grad_norm": 0.6926441192626953,
+      "learning_rate": 0.00026143550998381003,
+      "loss": 3.3279,
       "step": 52400
     },
     {
       "epoch": 5.654986522911051,
-      "grad_norm": 0.7090012431144714,
-      "learning_rate": 0.00026112466270912033,
-      "loss": 3.354,
+      "grad_norm": 0.7188628315925598,
+      "learning_rate": 0.0002611117107393416,
+      "loss": 3.3547,
       "step": 52450
     },
     {
       "epoch": 5.660377358490566,
-      "grad_norm": 0.6859129071235657,
-      "learning_rate": 0.0002608008634646519,
-      "loss": 3.3293,
+      "grad_norm": 0.7127479314804077,
+      "learning_rate": 0.00026078791149487313,
+      "loss": 3.3288,
       "step": 52500
     },
     {
       "epoch": 5.665768194070081,
-      "grad_norm": 0.7087217569351196,
-      "learning_rate": 0.00026047706422018343,
-      "loss": 3.3462,
+      "grad_norm": 0.6943938732147217,
+      "learning_rate": 0.00026046411225040474,
+      "loss": 3.3464,
       "step": 52550
     },
     {
       "epoch": 5.671159029649596,
-      "grad_norm": 0.6836449503898621,
-      "learning_rate": 0.00026015326497571504,
-      "loss": 3.3587,
+      "grad_norm": 0.6763001084327698,
+      "learning_rate": 0.0002601403130059363,
+      "loss": 3.3593,
       "step": 52600
     },
     {
       "epoch": 5.67654986522911,
-      "grad_norm": 0.7609317302703857,
-      "learning_rate": 0.0002598294657312466,
-      "loss": 3.3302,
+      "grad_norm": 0.7764798998832703,
+      "learning_rate": 0.0002598165137614679,
+      "loss": 3.3311,
       "step": 52650
     },
     {
       "epoch": 5.681940700808625,
-      "grad_norm": 0.7099915146827698,
-      "learning_rate": 0.0002595056664867782,
-      "loss": 3.3192,
+      "grad_norm": 0.7057187557220459,
+      "learning_rate": 0.00025949271451699944,
+      "loss": 3.3212,
       "step": 52700
     },
     {
       "epoch": 5.6873315363881405,
-      "grad_norm": 0.7150067090988159,
-      "learning_rate": 0.00025918186724230974,
-      "loss": 3.3365,
+      "grad_norm": 0.6973100900650024,
+      "learning_rate": 0.000259168915272531,
+      "loss": 3.3382,
       "step": 52750
     },
     {
       "epoch": 5.692722371967655,
-      "grad_norm": 0.6776279211044312,
-      "learning_rate": 0.00025885806799784135,
-      "loss": 3.3202,
+      "grad_norm": 0.6737171411514282,
+      "learning_rate": 0.0002588451160280626,
+      "loss": 3.3209,
       "step": 52800
     },
     {
       "epoch": 5.69811320754717,
-      "grad_norm": 0.6856902837753296,
-      "learning_rate": 0.0002585342687533729,
-      "loss": 3.3162,
+      "grad_norm": 0.715227484703064,
+      "learning_rate": 0.00025852131678359415,
+      "loss": 3.3182,
       "step": 52850
     },
     {
       "epoch": 5.703504043126685,
-      "grad_norm": 0.7150585651397705,
-      "learning_rate": 0.00025821046950890445,
-      "loss": 3.3215,
+      "grad_norm": 0.7033119797706604,
+      "learning_rate": 0.0002581975175391257,
+      "loss": 3.3229,
       "step": 52900
     },
     {
       "epoch": 5.708894878706199,
-      "grad_norm": 0.7075424194335938,
-      "learning_rate": 0.000257886670264436,
-      "loss": 3.3247,
+      "grad_norm": 0.7333863377571106,
+      "learning_rate": 0.0002578737182946573,
+      "loss": 3.3259,
       "step": 52950
     },
     {
       "epoch": 5.714285714285714,
-      "grad_norm": 0.7501097917556763,
-      "learning_rate": 0.0002575628710199676,
-      "loss": 3.3379,
+      "grad_norm": 0.7076955437660217,
+      "learning_rate": 0.00025754991905018885,
+      "loss": 3.3385,
       "step": 53000
     },
     {
       "epoch": 5.714285714285714,
-      "eval_accuracy": 0.38347841572355057,
-      "eval_loss": 3.3866093158721924,
-      "eval_runtime": 185.0709,
-      "eval_samples_per_second": 97.319,
-      "eval_steps_per_second": 6.084,
+      "eval_accuracy": 0.38299588824876646,
+      "eval_loss": 3.389537811279297,
+      "eval_runtime": 185.2976,
+      "eval_samples_per_second": 97.2,
+      "eval_steps_per_second": 6.077,
       "step": 53000
     },
     {
       "epoch": 5.719676549865229,
-      "grad_norm": 0.6975317597389221,
-      "learning_rate": 0.00025724554776038856,
-      "loss": 3.3276,
+      "grad_norm": 0.6770291328430176,
+      "learning_rate": 0.0002572325957906098,
+      "loss": 3.3304,
       "step": 53050
     },
     {
       "epoch": 5.725067385444744,
-      "grad_norm": 0.7414888143539429,
-      "learning_rate": 0.0002569217485159201,
-      "loss": 3.3394,
+      "grad_norm": 0.7202932834625244,
+      "learning_rate": 0.00025690879654614136,
+      "loss": 3.3392,
       "step": 53100
     },
     {
       "epoch": 5.730458221024259,
-      "grad_norm": 0.6874222159385681,
-      "learning_rate": 0.0002565979492714517,
-      "loss": 3.3305,
+      "grad_norm": 0.6683873534202576,
+      "learning_rate": 0.0002565849973016729,
+      "loss": 3.3322,
       "step": 53150
     },
     {
       "epoch": 5.735849056603773,
-      "grad_norm": 0.6985239386558533,
-      "learning_rate": 0.00025627415002698326,
-      "loss": 3.3424,
+      "grad_norm": 0.7561058402061462,
+      "learning_rate": 0.0002562611980572045,
+      "loss": 3.3437,
       "step": 53200
     },
     {
       "epoch": 5.741239892183288,
-      "grad_norm": 0.7149052619934082,
-      "learning_rate": 0.0002559503507825148,
-      "loss": 3.3156,
+      "grad_norm": 0.651046097278595,
+      "learning_rate": 0.00025593739881273606,
+      "loss": 3.3163,
       "step": 53250
     },
     {
       "epoch": 5.7466307277628035,
-      "grad_norm": 0.7191247940063477,
-      "learning_rate": 0.00025562655153804636,
-      "loss": 3.3569,
+      "grad_norm": 0.7331181764602661,
+      "learning_rate": 0.00025561359956826767,
+      "loss": 3.3574,
       "step": 53300
     },
     {
       "epoch": 5.752021563342318,
-      "grad_norm": 0.6939454674720764,
-      "learning_rate": 0.00025530275229357797,
-      "loss": 3.3385,
+      "grad_norm": 0.6629624366760254,
+      "learning_rate": 0.0002552898003237992,
+      "loss": 3.3406,
       "step": 53350
     },
     {
       "epoch": 5.757412398921833,
-      "grad_norm": 0.6978759765625,
-      "learning_rate": 0.0002549789530491095,
-      "loss": 3.3435,
+      "grad_norm": 0.6959452629089355,
+      "learning_rate": 0.0002549660010793308,
+      "loss": 3.345,
       "step": 53400
     },
     {
       "epoch": 5.762803234501348,
-      "grad_norm": 0.6790081858634949,
-      "learning_rate": 0.0002546551538046411,
-      "loss": 3.3399,
+      "grad_norm": 0.6279221773147583,
+      "learning_rate": 0.0002546422018348624,
+      "loss": 3.3411,
       "step": 53450
     },
     {
       "epoch": 5.768194070080862,
-      "grad_norm": 0.7253314852714539,
-      "learning_rate": 0.0002543313545601727,
-      "loss": 3.321,
+      "grad_norm": 0.6780901551246643,
+      "learning_rate": 0.0002543184025903939,
+      "loss": 3.3237,
       "step": 53500
     },
     {
       "epoch": 5.773584905660377,
-      "grad_norm": 0.6867673397064209,
-      "learning_rate": 0.0002540075553157042,
-      "loss": 3.3211,
+      "grad_norm": 0.702332079410553,
+      "learning_rate": 0.0002539946033459255,
+      "loss": 3.3236,
       "step": 53550
     },
     {
       "epoch": 5.7789757412398925,
-      "grad_norm": 0.7390033602714539,
-      "learning_rate": 0.00025368375607123583,
-      "loss": 3.3414,
+      "grad_norm": 0.7578160166740417,
+      "learning_rate": 0.0002536708041014571,
+      "loss": 3.3433,
       "step": 53600
     },
     {
       "epoch": 5.784366576819407,
-      "grad_norm": 0.6785998940467834,
-      "learning_rate": 0.0002533599568267674,
-      "loss": 3.3245,
+      "grad_norm": 0.6509333252906799,
+      "learning_rate": 0.00025334700485698863,
+      "loss": 3.3264,
       "step": 53650
     },
     {
       "epoch": 5.789757412398922,
-      "grad_norm": 0.6617134213447571,
-      "learning_rate": 0.00025303615758229893,
-      "loss": 3.3451,
+      "grad_norm": 0.7034610509872437,
+      "learning_rate": 0.00025302320561252023,
+      "loss": 3.3471,
       "step": 53700
     },
     {
       "epoch": 5.795148247978437,
-      "grad_norm": 0.7111865282058716,
-      "learning_rate": 0.00025271235833783053,
-      "loss": 3.3338,
+      "grad_norm": 0.688968300819397,
+      "learning_rate": 0.0002526994063680518,
+      "loss": 3.334,
       "step": 53750
     },
     {
       "epoch": 5.800539083557951,
-      "grad_norm": 0.7098289728164673,
-      "learning_rate": 0.0002523885590933621,
-      "loss": 3.3222,
+      "grad_norm": 0.6626862287521362,
+      "learning_rate": 0.0002523756071235834,
+      "loss": 3.3253,
       "step": 53800
     },
     {
       "epoch": 5.8059299191374665,
-      "grad_norm": 0.7283722162246704,
-      "learning_rate": 0.00025206475984889364,
-      "loss": 3.3266,
+      "grad_norm": 0.715716540813446,
+      "learning_rate": 0.00025205180787911494,
+      "loss": 3.3286,
       "step": 53850
     },
     {
       "epoch": 5.811320754716981,
-      "grad_norm": 0.7388856410980225,
-      "learning_rate": 0.00025174096060442524,
-      "loss": 3.3276,
+      "grad_norm": 0.6780540347099304,
+      "learning_rate": 0.0002517280086346465,
+      "loss": 3.329,
       "step": 53900
     },
     {
       "epoch": 5.816711590296496,
-      "grad_norm": 0.7053267955780029,
-      "learning_rate": 0.0002514171613599568,
-      "loss": 3.3259,
+      "grad_norm": 0.7091286182403564,
+      "learning_rate": 0.00025140420939017804,
+      "loss": 3.3265,
       "step": 53950
     },
     {
       "epoch": 5.822102425876011,
-      "grad_norm": 0.6764585971832275,
-      "learning_rate": 0.0002510933621154884,
-      "loss": 3.3211,
+      "grad_norm": 0.7116252779960632,
+      "learning_rate": 0.00025108041014570965,
+      "loss": 3.3229,
       "step": 54000
     },
     {
       "epoch": 5.822102425876011,
-      "eval_accuracy": 0.3836551939720247,
-      "eval_loss": 3.3842101097106934,
-      "eval_runtime": 184.7838,
-      "eval_samples_per_second": 97.471,
+      "eval_accuracy": 0.3835450199437808,
+      "eval_loss": 3.384857416152954,
+      "eval_runtime": 184.7842,
+      "eval_samples_per_second": 97.47,
       "eval_steps_per_second": 6.094,
       "step": 54000
     },
     {
       "epoch": 5.827493261455525,
-      "grad_norm": 0.6842856407165527,
-      "learning_rate": 0.00025076956287101995,
-      "loss": 3.3321,
+      "grad_norm": 0.6713538765907288,
+      "learning_rate": 0.0002507566109012412,
+      "loss": 3.3324,
       "step": 54050
     },
     {
       "epoch": 5.83288409703504,
-      "grad_norm": 0.6792701482772827,
-      "learning_rate": 0.00025044576362655155,
-      "loss": 3.32,
+      "grad_norm": 0.7192128896713257,
+      "learning_rate": 0.0002504328116567728,
+      "loss": 3.3225,
       "step": 54100
     },
     {
       "epoch": 5.8382749326145555,
-      "grad_norm": 0.7160159349441528,
-      "learning_rate": 0.00025012196438208305,
-      "loss": 3.3357,
+      "grad_norm": 0.7344403862953186,
+      "learning_rate": 0.00025010901241230435,
+      "loss": 3.3354,
       "step": 54150
     },
     {
       "epoch": 5.84366576819407,
-      "grad_norm": 0.744286060333252,
-      "learning_rate": 0.00024979816513761465,
-      "loss": 3.3244,
+      "grad_norm": 0.7250232100486755,
+      "learning_rate": 0.0002497852131678359,
+      "loss": 3.3251,
       "step": 54200
     },
     {
       "epoch": 5.849056603773585,
-      "grad_norm": 0.6898226141929626,
-      "learning_rate": 0.0002494743658931462,
-      "loss": 3.3246,
+      "grad_norm": 0.6815427541732788,
+      "learning_rate": 0.0002494614139233675,
+      "loss": 3.3268,
       "step": 54250
     },
     {
       "epoch": 5.8544474393531,
-      "grad_norm": 0.7597428560256958,
-      "learning_rate": 0.0002491505666486778,
-      "loss": 3.325,
+      "grad_norm": 0.7029092311859131,
+      "learning_rate": 0.00024913761467889906,
+      "loss": 3.3261,
       "step": 54300
     },
     {
       "epoch": 5.859838274932614,
-      "grad_norm": 0.7275387048721313,
-      "learning_rate": 0.00024882676740420936,
-      "loss": 3.3379,
+      "grad_norm": 0.7195873856544495,
+      "learning_rate": 0.0002488138154344306,
+      "loss": 3.3378,
       "step": 54350
     },
     {
       "epoch": 5.8652291105121295,
-      "grad_norm": 0.6952611207962036,
-      "learning_rate": 0.00024850296815974096,
-      "loss": 3.3227,
+      "grad_norm": 0.7248128652572632,
+      "learning_rate": 0.0002484900161899622,
+      "loss": 3.3249,
       "step": 54400
     },
     {
       "epoch": 5.870619946091644,
-      "grad_norm": 0.7258508801460266,
-      "learning_rate": 0.0002481791689152725,
-      "loss": 3.3406,
+      "grad_norm": 0.6670764684677124,
+      "learning_rate": 0.00024816621694549376,
+      "loss": 3.3428,
       "step": 54450
     },
     {
       "epoch": 5.876010781671159,
-      "grad_norm": 0.6898373365402222,
-      "learning_rate": 0.0002478553696708041,
-      "loss": 3.3427,
+      "grad_norm": 0.6498870253562927,
+      "learning_rate": 0.00024784241770102537,
+      "loss": 3.341,
       "step": 54500
     },
     {
       "epoch": 5.881401617250674,
-      "grad_norm": 0.7068774104118347,
-      "learning_rate": 0.0002475315704263356,
-      "loss": 3.3143,
+      "grad_norm": 0.7022345066070557,
+      "learning_rate": 0.0002475186184565569,
+      "loss": 3.3149,
       "step": 54550
     },
     {
       "epoch": 5.886792452830189,
-      "grad_norm": 0.6842253804206848,
-      "learning_rate": 0.0002472077711818672,
-      "loss": 3.3409,
+      "grad_norm": 0.6908873915672302,
+      "learning_rate": 0.00024719481921208847,
+      "loss": 3.3422,
       "step": 54600
     },
     {
       "epoch": 5.892183288409703,
-      "grad_norm": 0.6846425533294678,
-      "learning_rate": 0.00024688397193739877,
-      "loss": 3.3212,
+      "grad_norm": 0.663985013961792,
+      "learning_rate": 0.0002468710199676201,
+      "loss": 3.3228,
       "step": 54650
     },
     {
       "epoch": 5.8975741239892185,
-      "grad_norm": 0.6987577676773071,
-      "learning_rate": 0.0002465601726929304,
-      "loss": 3.3301,
+      "grad_norm": 0.7020365595817566,
+      "learning_rate": 0.0002465472207231516,
+      "loss": 3.3317,
       "step": 54700
     },
     {
       "epoch": 5.902964959568733,
-      "grad_norm": 0.6544787287712097,
-      "learning_rate": 0.0002462363734484619,
-      "loss": 3.338,
+      "grad_norm": 0.6784001588821411,
+      "learning_rate": 0.0002462234214786832,
+      "loss": 3.3385,
       "step": 54750
     },
     {
       "epoch": 5.908355795148248,
-      "grad_norm": 0.7005526423454285,
-      "learning_rate": 0.00024591257420399353,
-      "loss": 3.3302,
+      "grad_norm": 0.7305879592895508,
+      "learning_rate": 0.0002458996222342148,
+      "loss": 3.3328,
       "step": 54800
     },
     {
       "epoch": 5.913746630727763,
-      "grad_norm": 0.7439953684806824,
-      "learning_rate": 0.0002455887749595251,
-      "loss": 3.3389,
+      "grad_norm": 0.7549262046813965,
+      "learning_rate": 0.00024557582298974633,
+      "loss": 3.3406,
       "step": 54850
     },
     {
       "epoch": 5.919137466307277,
-      "grad_norm": 0.714672327041626,
-      "learning_rate": 0.00024526497571505663,
-      "loss": 3.3383,
+      "grad_norm": 0.7167611122131348,
+      "learning_rate": 0.0002452520237452779,
+      "loss": 3.3398,
       "step": 54900
     },
     {
       "epoch": 5.9245283018867925,
-      "grad_norm": 0.7019497156143188,
-      "learning_rate": 0.00024494117647058824,
-      "loss": 3.3411,
+      "grad_norm": 0.7034414410591125,
+      "learning_rate": 0.0002449282245008095,
+      "loss": 3.344,
       "step": 54950
     },
     {
       "epoch": 5.929919137466308,
-      "grad_norm": 0.6823916435241699,
-      "learning_rate": 0.0002446173772261198,
-      "loss": 3.3249,
+      "grad_norm": 0.691204309463501,
+      "learning_rate": 0.00024460442525634104,
+      "loss": 3.3257,
       "step": 55000
     },
     {
       "epoch": 5.929919137466308,
-      "eval_accuracy": 0.3843319928102211,
-      "eval_loss": 3.379180908203125,
-      "eval_runtime": 185.106,
-      "eval_samples_per_second": 97.301,
-      "eval_steps_per_second": 6.083,
+      "eval_accuracy": 0.38400364378324575,
+      "eval_loss": 3.38031268119812,
+      "eval_runtime": 185.1848,
+      "eval_samples_per_second": 97.26,
+      "eval_steps_per_second": 6.08,
       "step": 55000
     },
     {
       "epoch": 5.935309973045822,
-      "grad_norm": 0.7124596834182739,
-      "learning_rate": 0.00024429357798165134,
-      "loss": 3.3525,
+      "grad_norm": 0.6970035433769226,
+      "learning_rate": 0.00024428062601187264,
+      "loss": 3.3526,
       "step": 55050
     },
     {
       "epoch": 5.940700808625337,
-      "grad_norm": 0.706619381904602,
-      "learning_rate": 0.00024396977873718291,
-      "loss": 3.3188,
+      "grad_norm": 0.7048007249832153,
+      "learning_rate": 0.0002439568267674042,
+      "loss": 3.3196,
       "step": 55100
     },
     {
       "epoch": 5.946091644204852,
-      "grad_norm": 0.720799446105957,
-      "learning_rate": 0.0002436459794927145,
-      "loss": 3.3249,
+      "grad_norm": 0.6989700198173523,
+      "learning_rate": 0.00024363302752293574,
+      "loss": 3.3271,
       "step": 55150
     },
     {
       "epoch": 5.951482479784366,
-      "grad_norm": 0.694983720779419,
-      "learning_rate": 0.00024332218024824607,
-      "loss": 3.3204,
+      "grad_norm": 0.714759349822998,
+      "learning_rate": 0.00024330922827846732,
+      "loss": 3.3236,
       "step": 55200
     },
     {
       "epoch": 5.9568733153638815,
-      "grad_norm": 0.6975095272064209,
-      "learning_rate": 0.00024299838100377765,
-      "loss": 3.328,
+      "grad_norm": 0.7349509000778198,
+      "learning_rate": 0.0002429854290339989,
+      "loss": 3.3284,
       "step": 55250
     },
     {
       "epoch": 5.962264150943396,
-      "grad_norm": 0.6743301153182983,
-      "learning_rate": 0.00024267458175930922,
-      "loss": 3.321,
+      "grad_norm": 0.671440839767456,
+      "learning_rate": 0.00024266162978953048,
+      "loss": 3.3217,
       "step": 55300
     },
     {
       "epoch": 5.967654986522911,
-      "grad_norm": 0.7407911419868469,
-      "learning_rate": 0.0002423507825148408,
-      "loss": 3.3238,
+      "grad_norm": 0.7063277363777161,
+      "learning_rate": 0.00024233783054506203,
+      "loss": 3.324,
       "step": 55350
     },
     {
       "epoch": 5.973045822102426,
-      "grad_norm": 0.6877210140228271,
-      "learning_rate": 0.00024202698327037233,
-      "loss": 3.3247,
+      "grad_norm": 0.6923580169677734,
+      "learning_rate": 0.0002420140313005936,
+      "loss": 3.3253,
       "step": 55400
     },
     {
       "epoch": 5.97843665768194,
-      "grad_norm": 0.696134090423584,
-      "learning_rate": 0.0002417031840259039,
-      "loss": 3.3433,
+      "grad_norm": 0.6707276105880737,
+      "learning_rate": 0.00024169023205612518,
+      "loss": 3.3456,
       "step": 55450
     },
     {
       "epoch": 5.9838274932614555,
-      "grad_norm": 0.7565768361091614,
-      "learning_rate": 0.00024137938478143548,
-      "loss": 3.3266,
+      "grad_norm": 0.7710720896720886,
+      "learning_rate": 0.00024136643281165676,
+      "loss": 3.3296,
       "step": 55500
     },
     {
       "epoch": 5.989218328840971,
-      "grad_norm": 0.7190529704093933,
-      "learning_rate": 0.00024105558553696706,
-      "loss": 3.33,
+      "grad_norm": 0.7132817506790161,
+      "learning_rate": 0.00024104263356718834,
+      "loss": 3.3307,
       "step": 55550
     },
     {
       "epoch": 5.994609164420485,
-      "grad_norm": 0.732549786567688,
-      "learning_rate": 0.00024073178629249864,
-      "loss": 3.3474,
+      "grad_norm": 0.7166342735290527,
+      "learning_rate": 0.0002407188343227199,
+      "loss": 3.3496,
       "step": 55600
     },
     {
       "epoch": 6.0,
-      "grad_norm": 1.5415840148925781,
-      "learning_rate": 0.00024040798704803021,
-      "loss": 3.3441,
+      "grad_norm": 1.5364723205566406,
+      "learning_rate": 0.00024039503507825147,
+      "loss": 3.3466,
       "step": 55650
     },
     {
       "epoch": 6.005390835579515,
-      "grad_norm": 0.7237451076507568,
-      "learning_rate": 0.00024008418780356176,
-      "loss": 3.24,
+      "grad_norm": 0.756853461265564,
+      "learning_rate": 0.00024007123583378302,
+      "loss": 3.2414,
       "step": 55700
     },
     {
       "epoch": 6.010781671159029,
-      "grad_norm": 0.7061052918434143,
-      "learning_rate": 0.00023976038855909334,
-      "loss": 3.2419,
+      "grad_norm": 0.6904761791229248,
+      "learning_rate": 0.0002397474365893146,
+      "loss": 3.2437,
       "step": 55750
     },
     {
       "epoch": 6.0161725067385445,
-      "grad_norm": 0.7535544037818909,
-      "learning_rate": 0.00023943658931462492,
-      "loss": 3.234,
+      "grad_norm": 0.7776129245758057,
+      "learning_rate": 0.00023942363734484617,
+      "loss": 3.2365,
       "step": 55800
     },
     {
       "epoch": 6.02156334231806,
-      "grad_norm": 0.7151229977607727,
-      "learning_rate": 0.00023911279007015647,
-      "loss": 3.242,
+      "grad_norm": 0.7028306722640991,
+      "learning_rate": 0.00023909983810037775,
+      "loss": 3.2434,
       "step": 55850
     },
     {
       "epoch": 6.026954177897574,
-      "grad_norm": 0.6773797273635864,
-      "learning_rate": 0.00023878899082568805,
-      "loss": 3.2422,
+      "grad_norm": 0.6557170152664185,
+      "learning_rate": 0.00023877603885590933,
+      "loss": 3.2437,
       "step": 55900
     },
     {
       "epoch": 6.032345013477089,
-      "grad_norm": 0.6992098093032837,
-      "learning_rate": 0.00023846519158121963,
-      "loss": 3.238,
+      "grad_norm": 0.696887731552124,
+      "learning_rate": 0.0002384522396114409,
+      "loss": 3.2402,
       "step": 55950
     },
     {
       "epoch": 6.037735849056604,
-      "grad_norm": 0.6867258548736572,
-      "learning_rate": 0.0002381413923367512,
-      "loss": 3.2353,
+      "grad_norm": 0.6949684619903564,
+      "learning_rate": 0.00023812844036697248,
+      "loss": 3.2365,
       "step": 56000
     },
     {
       "epoch": 6.037735849056604,
-      "eval_accuracy": 0.3839687662062573,
-      "eval_loss": 3.3829169273376465,
-      "eval_runtime": 185.0633,
-      "eval_samples_per_second": 97.323,
-      "eval_steps_per_second": 6.084,
+      "eval_accuracy": 0.3836666025252452,
+      "eval_loss": 3.385128974914551,
+      "eval_runtime": 184.7946,
+      "eval_samples_per_second": 97.465,
+      "eval_steps_per_second": 6.093,
       "step": 56000
     },
     {
       "epoch": 6.0431266846361185,
-      "grad_norm": 0.7631310820579529,
-      "learning_rate": 0.00023781759309228275,
-      "loss": 3.2433,
+      "grad_norm": 0.6826362609863281,
+      "learning_rate": 0.000237804641122504,
+      "loss": 3.246,
       "step": 56050
     },
     {
       "epoch": 6.048517520215634,
-      "grad_norm": 0.7571997046470642,
-      "learning_rate": 0.00023749379384781433,
-      "loss": 3.2398,
+      "grad_norm": 0.7227660417556763,
+      "learning_rate": 0.00023748084187803558,
+      "loss": 3.2431,
       "step": 56100
     },
     {
       "epoch": 6.053908355795148,
-      "grad_norm": 0.7700287699699402,
-      "learning_rate": 0.0002371699946033459,
-      "loss": 3.2567,
+      "grad_norm": 0.7226882576942444,
+      "learning_rate": 0.00023715704263356716,
+      "loss": 3.2582,
       "step": 56150
     },
     {
       "epoch": 6.059299191374663,
-      "grad_norm": 0.7289305329322815,
-      "learning_rate": 0.0002368461953588775,
-      "loss": 3.2385,
+      "grad_norm": 0.7452118396759033,
+      "learning_rate": 0.00023683324338909874,
+      "loss": 3.2401,
       "step": 56200
     },
     {
       "epoch": 6.064690026954178,
-      "grad_norm": 0.7690556049346924,
-      "learning_rate": 0.00023652239611440904,
-      "loss": 3.2467,
+      "grad_norm": 0.7395612001419067,
+      "learning_rate": 0.00023650944414463032,
+      "loss": 3.2499,
       "step": 56250
     },
     {
       "epoch": 6.070080862533692,
-      "grad_norm": 0.6697551608085632,
-      "learning_rate": 0.00023619859686994062,
-      "loss": 3.2544,
+      "grad_norm": 0.7218838334083557,
+      "learning_rate": 0.0002361856449001619,
+      "loss": 3.2576,
       "step": 56300
     },
     {
       "epoch": 6.0754716981132075,
-      "grad_norm": 0.7450080513954163,
-      "learning_rate": 0.00023587479762547217,
-      "loss": 3.2534,
+      "grad_norm": 0.761297345161438,
+      "learning_rate": 0.00023586184565569347,
+      "loss": 3.2537,
       "step": 56350
     },
     {
       "epoch": 6.080862533692723,
-      "grad_norm": 0.7233448624610901,
-      "learning_rate": 0.00023555099838100374,
-      "loss": 3.2729,
+      "grad_norm": 0.7278382778167725,
+      "learning_rate": 0.00023553804641122502,
+      "loss": 3.2738,
       "step": 56400
     },
     {
       "epoch": 6.086253369272237,
-      "grad_norm": 0.6958713531494141,
-      "learning_rate": 0.00023522719913653532,
-      "loss": 3.2581,
+      "grad_norm": 0.6822406053543091,
+      "learning_rate": 0.00023521424716675657,
+      "loss": 3.2623,
       "step": 56450
     },
     {
       "epoch": 6.091644204851752,
-      "grad_norm": 0.7158046364784241,
-      "learning_rate": 0.0002349033998920669,
-      "loss": 3.2399,
+      "grad_norm": 0.7750765681266785,
+      "learning_rate": 0.00023489044792228815,
+      "loss": 3.2439,
       "step": 56500
     },
     {
       "epoch": 6.097035040431267,
-      "grad_norm": 0.6767109036445618,
-      "learning_rate": 0.00023457960064759848,
-      "loss": 3.2716,
+      "grad_norm": 0.6662526726722717,
+      "learning_rate": 0.00023456664867781973,
+      "loss": 3.2745,
       "step": 56550
     },
     {
       "epoch": 6.1024258760107815,
-      "grad_norm": 0.7307221293449402,
-      "learning_rate": 0.00023425580140313005,
-      "loss": 3.2545,
+      "grad_norm": 0.7166032791137695,
+      "learning_rate": 0.0002342428494333513,
+      "loss": 3.256,
       "step": 56600
     },
     {
       "epoch": 6.107816711590297,
-      "grad_norm": 0.7141797542572021,
-      "learning_rate": 0.00023393200215866163,
-      "loss": 3.238,
+      "grad_norm": 0.7623759508132935,
+      "learning_rate": 0.00023391905018888288,
+      "loss": 3.2411,
       "step": 56650
     },
     {
       "epoch": 6.113207547169812,
-      "grad_norm": 0.7216366529464722,
-      "learning_rate": 0.00023360820291419316,
-      "loss": 3.2681,
+      "grad_norm": 0.7492631673812866,
+      "learning_rate": 0.00023359525094441443,
+      "loss": 3.2714,
       "step": 56700
     },
     {
       "epoch": 6.118598382749326,
-      "grad_norm": 0.7594369649887085,
-      "learning_rate": 0.00023328440366972473,
-      "loss": 3.277,
+      "grad_norm": 0.7426467537879944,
+      "learning_rate": 0.000233271451699946,
+      "loss": 3.2786,
       "step": 56750
     },
     {
       "epoch": 6.123989218328841,
-      "grad_norm": 0.7302905321121216,
-      "learning_rate": 0.0002329606044252563,
-      "loss": 3.2676,
+      "grad_norm": 0.6925146579742432,
+      "learning_rate": 0.0002329476524554776,
+      "loss": 3.2701,
       "step": 56800
     },
     {
       "epoch": 6.129380053908355,
-      "grad_norm": 0.734412670135498,
-      "learning_rate": 0.0002326368051807879,
-      "loss": 3.2495,
+      "grad_norm": 0.7421886920928955,
+      "learning_rate": 0.00023263032919589851,
+      "loss": 3.2516,
       "step": 56850
     },
     {
       "epoch": 6.1347708894878705,
-      "grad_norm": 0.7013684511184692,
-      "learning_rate": 0.00023231300593631947,
-      "loss": 3.2643,
+      "grad_norm": 0.6857749223709106,
+      "learning_rate": 0.0002323065299514301,
+      "loss": 3.2697,
       "step": 56900
     },
     {
       "epoch": 6.140161725067386,
-      "grad_norm": 0.7368125319480896,
-      "learning_rate": 0.00023198920669185104,
-      "loss": 3.249,
+      "grad_norm": 0.7283300161361694,
+      "learning_rate": 0.00023198273070696167,
+      "loss": 3.2511,
       "step": 56950
     },
     {
       "epoch": 6.1455525606469,
-      "grad_norm": 0.7295984625816345,
-      "learning_rate": 0.00023166540744738262,
-      "loss": 3.2655,
+      "grad_norm": 0.6971080899238586,
+      "learning_rate": 0.00023165893146249325,
+      "loss": 3.2675,
       "step": 57000
     },
     {
       "epoch": 6.1455525606469,
-      "eval_accuracy": 0.38440663734414965,
-      "eval_loss": 3.3809986114501953,
-      "eval_runtime": 185.0253,
-      "eval_samples_per_second": 97.343,
+      "eval_accuracy": 0.3843410109999097,
+      "eval_loss": 3.3825490474700928,
+      "eval_runtime": 185.0025,
+      "eval_samples_per_second": 97.355,
       "eval_steps_per_second": 6.086,
       "step": 57000
     },
     {
       "epoch": 6.150943396226415,
-      "grad_norm": 0.7095818519592285,
-      "learning_rate": 0.00023134808418780352,
-      "loss": 3.2574,
+      "grad_norm": 0.6910756230354309,
+      "learning_rate": 0.0002313351322180248,
+      "loss": 3.2588,
       "step": 57050
     },
     {
       "epoch": 6.15633423180593,
-      "grad_norm": 0.7285301685333252,
-      "learning_rate": 0.0002310242849433351,
-      "loss": 3.2646,
+      "grad_norm": 0.6956217288970947,
+      "learning_rate": 0.00023101133297355638,
+      "loss": 3.2673,
       "step": 57100
     },
     {
       "epoch": 6.1617250673854445,
-      "grad_norm": 0.6750980019569397,
-      "learning_rate": 0.00023070048569886667,
-      "loss": 3.2782,
+      "grad_norm": 0.6852027773857117,
+      "learning_rate": 0.00023068753372908795,
+      "loss": 3.2824,
       "step": 57150
     },
     {
       "epoch": 6.16711590296496,
-      "grad_norm": 0.7419140934944153,
-      "learning_rate": 0.00023037668645439825,
-      "loss": 3.2553,
+      "grad_norm": 0.7331057190895081,
+      "learning_rate": 0.0002303637344846195,
+      "loss": 3.2579,
       "step": 57200
     },
     {
       "epoch": 6.172506738544475,
-      "grad_norm": 0.6713706851005554,
-      "learning_rate": 0.00023005288720992983,
-      "loss": 3.2533,
+      "grad_norm": 0.6825144290924072,
+      "learning_rate": 0.00023003993524015108,
+      "loss": 3.2576,
       "step": 57250
     },
     {
       "epoch": 6.177897574123989,
-      "grad_norm": 0.7637577056884766,
-      "learning_rate": 0.0002297290879654614,
-      "loss": 3.2685,
+      "grad_norm": 0.746617317199707,
+      "learning_rate": 0.00022971613599568266,
+      "loss": 3.2711,
       "step": 57300
     },
     {
       "epoch": 6.183288409703504,
-      "grad_norm": 0.709025502204895,
-      "learning_rate": 0.00022940528872099299,
-      "loss": 3.256,
+      "grad_norm": 0.6929312944412231,
+      "learning_rate": 0.0002293923367512142,
+      "loss": 3.2592,
       "step": 57350
     },
     {
       "epoch": 6.188679245283019,
-      "grad_norm": 0.7199300527572632,
-      "learning_rate": 0.00022908148947652454,
-      "loss": 3.2631,
+      "grad_norm": 0.7209480404853821,
+      "learning_rate": 0.0002290685375067458,
+      "loss": 3.2663,
       "step": 57400
     },
     {
       "epoch": 6.1940700808625335,
-      "grad_norm": 0.7396861910820007,
-      "learning_rate": 0.0002287576902320561,
-      "loss": 3.2615,
+      "grad_norm": 0.7121549248695374,
+      "learning_rate": 0.00022874473826227736,
+      "loss": 3.2621,
       "step": 57450
     },
     {
       "epoch": 6.199460916442049,
-      "grad_norm": 0.7257159352302551,
-      "learning_rate": 0.00022843389098758766,
-      "loss": 3.2525,
+      "grad_norm": 0.7355096340179443,
+      "learning_rate": 0.00022842093901780894,
+      "loss": 3.256,
       "step": 57500
     },
     {
       "epoch": 6.204851752021563,
-      "grad_norm": 0.7163692712783813,
-      "learning_rate": 0.00022811009174311924,
-      "loss": 3.277,
+      "grad_norm": 0.7216746211051941,
+      "learning_rate": 0.00022809713977334052,
+      "loss": 3.2775,
       "step": 57550
     },
     {
       "epoch": 6.210242587601078,
-      "grad_norm": 0.7706537246704102,
-      "learning_rate": 0.00022778629249865082,
-      "loss": 3.2624,
+      "grad_norm": 0.7493441700935364,
+      "learning_rate": 0.0002277733405288721,
+      "loss": 3.2659,
       "step": 57600
     },
     {
       "epoch": 6.215633423180593,
-      "grad_norm": 0.7195106148719788,
-      "learning_rate": 0.0002274624932541824,
-      "loss": 3.2776,
+      "grad_norm": 0.7113534808158875,
+      "learning_rate": 0.00022744954128440365,
+      "loss": 3.2794,
       "step": 57650
     },
     {
       "epoch": 6.2210242587601075,
-      "grad_norm": 0.7585012912750244,
-      "learning_rate": 0.00022713869400971397,
+      "grad_norm": 0.7359064817428589,
+      "learning_rate": 0.0002271257420399352,
       "loss": 3.2706,
       "step": 57700
     },
     {
       "epoch": 6.226415094339623,
-      "grad_norm": 0.7925708293914795,
-      "learning_rate": 0.00022681489476524553,
-      "loss": 3.261,
+      "grad_norm": 0.7439775466918945,
+      "learning_rate": 0.00022680194279546678,
+      "loss": 3.2631,
       "step": 57750
     },
     {
       "epoch": 6.231805929919138,
-      "grad_norm": 0.7118070125579834,
-      "learning_rate": 0.0002264910955207771,
-      "loss": 3.2546,
+      "grad_norm": 0.7042476534843445,
+      "learning_rate": 0.00022647814355099835,
+      "loss": 3.2573,
       "step": 57800
     },
     {
       "epoch": 6.237196765498652,
-      "grad_norm": 0.7217944264411926,
-      "learning_rate": 0.00022616729627630868,
-      "loss": 3.2693,
+      "grad_norm": 0.7179429531097412,
+      "learning_rate": 0.00022615434430652993,
+      "loss": 3.2712,
       "step": 57850
     },
     {
       "epoch": 6.242587601078167,
-      "grad_norm": 0.7400355935096741,
-      "learning_rate": 0.00022584349703184023,
-      "loss": 3.2768,
+      "grad_norm": 0.7358320355415344,
+      "learning_rate": 0.0002258305450620615,
+      "loss": 3.2811,
       "step": 57900
     },
     {
       "epoch": 6.247978436657682,
-      "grad_norm": 0.7412822842597961,
-      "learning_rate": 0.0002255196977873718,
-      "loss": 3.269,
+      "grad_norm": 0.6951166391372681,
+      "learning_rate": 0.0002255067458175931,
+      "loss": 3.2729,
       "step": 57950
     },
     {
       "epoch": 6.2533692722371965,
-      "grad_norm": 0.694203794002533,
-      "learning_rate": 0.0002251958985429034,
-      "loss": 3.2709,
+      "grad_norm": 0.7272751331329346,
+      "learning_rate": 0.00022518294657312467,
+      "loss": 3.2744,
       "step": 58000
     },
     {
       "epoch": 6.2533692722371965,
-      "eval_accuracy": 0.3845205055705792,
-      "eval_loss": 3.3798089027404785,
-      "eval_runtime": 185.1754,
-      "eval_samples_per_second": 97.265,
-      "eval_steps_per_second": 6.081,
+      "eval_accuracy": 0.3845305016362582,
+      "eval_loss": 3.3816213607788086,
+      "eval_runtime": 185.286,
+      "eval_samples_per_second": 97.207,
+      "eval_steps_per_second": 6.077,
       "step": 58000
     },
     {
       "epoch": 6.258760107816712,
-      "grad_norm": 0.6906482577323914,
-      "learning_rate": 0.00022487209929843494,
-      "loss": 3.2579,
+      "grad_norm": 0.6936237812042236,
+      "learning_rate": 0.0002248591473286562,
+      "loss": 3.2592,
       "step": 58050
     },
     {
       "epoch": 6.264150943396227,
-      "grad_norm": 0.704700767993927,
-      "learning_rate": 0.00022454830005396652,
-      "loss": 3.2713,
+      "grad_norm": 0.7434632182121277,
+      "learning_rate": 0.00022453534808418777,
+      "loss": 3.274,
       "step": 58100
     },
     {
       "epoch": 6.269541778975741,
-      "grad_norm": 0.718530535697937,
-      "learning_rate": 0.0002242245008094981,
-      "loss": 3.2661,
+      "grad_norm": 0.697569727897644,
+      "learning_rate": 0.00022421154883971934,
+      "loss": 3.2678,
       "step": 58150
     },
     {
       "epoch": 6.274932614555256,
-      "grad_norm": 0.7455987334251404,
-      "learning_rate": 0.00022390070156502967,
-      "loss": 3.2877,
+      "grad_norm": 0.710084080696106,
+      "learning_rate": 0.00022388774959525092,
+      "loss": 3.2885,
       "step": 58200
     },
     {
       "epoch": 6.280323450134771,
-      "grad_norm": 0.7250301241874695,
-      "learning_rate": 0.00022357690232056125,
-      "loss": 3.2694,
+      "grad_norm": 0.7134957313537598,
+      "learning_rate": 0.0002235639503507825,
+      "loss": 3.2707,
       "step": 58250
     },
     {
       "epoch": 6.285714285714286,
-      "grad_norm": 0.71706622838974,
-      "learning_rate": 0.0002232531030760928,
-      "loss": 3.2583,
+      "grad_norm": 0.694118320941925,
+      "learning_rate": 0.00022324015110631408,
+      "loss": 3.2621,
       "step": 58300
     },
     {
       "epoch": 6.291105121293801,
-      "grad_norm": 0.7500212788581848,
-      "learning_rate": 0.00022292930383162435,
-      "loss": 3.2811,
+      "grad_norm": 0.7010383009910583,
+      "learning_rate": 0.00022291635186184565,
+      "loss": 3.2829,
       "step": 58350
     },
     {
       "epoch": 6.296495956873315,
-      "grad_norm": 0.7041599750518799,
-      "learning_rate": 0.00022260550458715593,
-      "loss": 3.2633,
+      "grad_norm": 0.6871818900108337,
+      "learning_rate": 0.0002225925526173772,
+      "loss": 3.2675,
       "step": 58400
     },
     {
       "epoch": 6.30188679245283,
-      "grad_norm": 0.6938748955726624,
-      "learning_rate": 0.0002222817053426875,
-      "loss": 3.2907,
+      "grad_norm": 0.7094181776046753,
+      "learning_rate": 0.00022226875337290878,
+      "loss": 3.2923,
       "step": 58450
     },
     {
       "epoch": 6.307277628032345,
-      "grad_norm": 0.7118483185768127,
-      "learning_rate": 0.00022195790609821908,
-      "loss": 3.2895,
+      "grad_norm": 0.697614848613739,
+      "learning_rate": 0.00022194495412844033,
+      "loss": 3.2906,
       "step": 58500
     },
     {
       "epoch": 6.3126684636118595,
-      "grad_norm": 0.7580938935279846,
-      "learning_rate": 0.00022163410685375066,
-      "loss": 3.2791,
+      "grad_norm": 0.7467103004455566,
+      "learning_rate": 0.0002216211548839719,
+      "loss": 3.2814,
       "step": 58550
     },
     {
       "epoch": 6.318059299191375,
-      "grad_norm": 0.7854766845703125,
-      "learning_rate": 0.00022131030760928224,
-      "loss": 3.2838,
+      "grad_norm": 0.7276577353477478,
+      "learning_rate": 0.0002212973556395035,
+      "loss": 3.2854,
       "step": 58600
     },
     {
       "epoch": 6.32345013477089,
-      "grad_norm": 0.7117508053779602,
-      "learning_rate": 0.00022098650836481382,
-      "loss": 3.2798,
+      "grad_norm": 0.7234843969345093,
+      "learning_rate": 0.00022097355639503507,
+      "loss": 3.2813,
       "step": 58650
     },
     {
       "epoch": 6.328840970350404,
-      "grad_norm": 0.8072823286056519,
-      "learning_rate": 0.0002206627091203454,
-      "loss": 3.2804,
+      "grad_norm": 0.7457574009895325,
+      "learning_rate": 0.00022064975715056664,
+      "loss": 3.2817,
       "step": 58700
     },
     {
       "epoch": 6.334231805929919,
-      "grad_norm": 0.7097095847129822,
-      "learning_rate": 0.00022033890987587692,
-      "loss": 3.2692,
+      "grad_norm": 0.7120930552482605,
+      "learning_rate": 0.0002203259579060982,
+      "loss": 3.2705,
       "step": 58750
     },
     {
       "epoch": 6.339622641509434,
-      "grad_norm": 0.7268771529197693,
-      "learning_rate": 0.0002200151106314085,
-      "loss": 3.2774,
+      "grad_norm": 0.7144607305526733,
+      "learning_rate": 0.00022000215866162977,
+      "loss": 3.2781,
       "step": 58800
     },
     {
       "epoch": 6.345013477088949,
-      "grad_norm": 0.7636404037475586,
-      "learning_rate": 0.00021969131138694007,
-      "loss": 3.2894,
+      "grad_norm": 0.6950954794883728,
+      "learning_rate": 0.00021967835941716135,
+      "loss": 3.2913,
       "step": 58850
     },
     {
       "epoch": 6.350404312668464,
-      "grad_norm": 0.7441515326499939,
-      "learning_rate": 0.00021936751214247165,
+      "grad_norm": 0.7282975912094116,
+      "learning_rate": 0.0002193545601726929,
       "loss": 3.2755,
       "step": 58900
     },
     {
       "epoch": 6.355795148247978,
-      "grad_norm": 0.7508131861686707,
-      "learning_rate": 0.00021904371289800323,
-      "loss": 3.2721,
+      "grad_norm": 0.7318394780158997,
+      "learning_rate": 0.00021903076092822448,
+      "loss": 3.276,
       "step": 58950
     },
     {
       "epoch": 6.361185983827493,
-      "grad_norm": 0.7223631739616394,
-      "learning_rate": 0.0002187199136535348,
-      "loss": 3.2884,
+      "grad_norm": 0.7199020385742188,
+      "learning_rate": 0.00021870696168375606,
+      "loss": 3.2922,
       "step": 59000
     },
     {
       "epoch": 6.361185983827493,
-      "eval_accuracy": 0.38532431963463076,
-      "eval_loss": 3.3748483657836914,
-      "eval_runtime": 184.7925,
-      "eval_samples_per_second": 97.466,
-      "eval_steps_per_second": 6.093,
+      "eval_accuracy": 0.3849554430805006,
+      "eval_loss": 3.3758089542388916,
+      "eval_runtime": 184.9696,
+      "eval_samples_per_second": 97.373,
+      "eval_steps_per_second": 6.087,
       "step": 59000
     },
     {
       "epoch": 6.366576819407008,
-      "grad_norm": 0.7511294484138489,
-      "learning_rate": 0.00021839611440906638,
-      "loss": 3.2777,
+      "grad_norm": 0.7290945053100586,
+      "learning_rate": 0.0002183831624392876,
+      "loss": 3.2789,
       "step": 59050
     },
     {
       "epoch": 6.3719676549865225,
-      "grad_norm": 0.7336013317108154,
-      "learning_rate": 0.00021807879114948728,
-      "loss": 3.275,
+      "grad_norm": 0.7788245677947998,
+      "learning_rate": 0.00021805936319481918,
+      "loss": 3.2769,
       "step": 59100
     },
     {
       "epoch": 6.377358490566038,
-      "grad_norm": 0.7369405627250671,
-      "learning_rate": 0.00021775499190501886,
-      "loss": 3.2704,
+      "grad_norm": 0.7099036574363708,
+      "learning_rate": 0.00021773556395035076,
+      "loss": 3.271,
       "step": 59150
     },
     {
       "epoch": 6.382749326145553,
-      "grad_norm": 0.7976130843162537,
-      "learning_rate": 0.00021743119266055044,
-      "loss": 3.2738,
+      "grad_norm": 0.7210767865180969,
+      "learning_rate": 0.00021741176470588234,
+      "loss": 3.2757,
       "step": 59200
     },
     {
       "epoch": 6.388140161725067,
-      "grad_norm": 0.7466988563537598,
-      "learning_rate": 0.000217107393416082,
-      "loss": 3.2758,
+      "grad_norm": 0.7274968028068542,
+      "learning_rate": 0.00021708796546141392,
+      "loss": 3.2776,
       "step": 59250
     },
     {
       "epoch": 6.393530997304582,
-      "grad_norm": 0.7116513848304749,
-      "learning_rate": 0.0002167835941716136,
-      "loss": 3.2786,
+      "grad_norm": 0.6926999688148499,
+      "learning_rate": 0.0002167641662169455,
+      "loss": 3.2813,
       "step": 59300
     },
     {
       "epoch": 6.398921832884097,
-      "grad_norm": 0.7164875864982605,
-      "learning_rate": 0.00021645979492714517,
-      "loss": 3.2747,
+      "grad_norm": 0.6983639001846313,
+      "learning_rate": 0.00021644036697247702,
+      "loss": 3.2765,
       "step": 59350
     },
     {
       "epoch": 6.404312668463612,
-      "grad_norm": 0.7558356523513794,
-      "learning_rate": 0.00021613599568267672,
-      "loss": 3.3011,
+      "grad_norm": 0.7861400246620178,
+      "learning_rate": 0.0002161165677280086,
+      "loss": 3.3026,
       "step": 59400
     },
     {
       "epoch": 6.409703504043127,
-      "grad_norm": 0.7918745875358582,
-      "learning_rate": 0.0002158121964382083,
-      "loss": 3.2843,
+      "grad_norm": 0.7995545864105225,
+      "learning_rate": 0.00021579276848354017,
+      "loss": 3.2864,
       "step": 59450
     },
     {
       "epoch": 6.415094339622642,
-      "grad_norm": 0.7076215744018555,
-      "learning_rate": 0.00021548839719373985,
-      "loss": 3.2718,
+      "grad_norm": 0.7238384485244751,
+      "learning_rate": 0.00021546896923907175,
+      "loss": 3.273,
       "step": 59500
     },
     {
       "epoch": 6.420485175202156,
-      "grad_norm": 0.7596773505210876,
-      "learning_rate": 0.00021516459794927142,
-      "loss": 3.2693,
+      "grad_norm": 0.7718705534934998,
+      "learning_rate": 0.00021514516999460333,
+      "loss": 3.2711,
       "step": 59550
     },
     {
       "epoch": 6.425876010781671,
-      "grad_norm": 0.7385455965995789,
-      "learning_rate": 0.000214840798704803,
-      "loss": 3.293,
+      "grad_norm": 0.7363432049751282,
+      "learning_rate": 0.0002148213707501349,
+      "loss": 3.2944,
       "step": 59600
     },
     {
       "epoch": 6.431266846361186,
-      "grad_norm": 0.7650585174560547,
-      "learning_rate": 0.00021451699946033458,
-      "loss": 3.2822,
+      "grad_norm": 0.7251701354980469,
+      "learning_rate": 0.00021449757150566648,
+      "loss": 3.2848,
       "step": 59650
     },
     {
       "epoch": 6.436657681940701,
-      "grad_norm": 0.7756490111351013,
-      "learning_rate": 0.00021419320021586616,
-      "loss": 3.2918,
+      "grad_norm": 0.754550576210022,
+      "learning_rate": 0.00021417377226119806,
+      "loss": 3.2915,
       "step": 59700
     },
     {
       "epoch": 6.442048517520216,
-      "grad_norm": 0.7733294367790222,
-      "learning_rate": 0.0002138694009713977,
-      "loss": 3.2897,
+      "grad_norm": 0.7444179654121399,
+      "learning_rate": 0.0002138499730167296,
+      "loss": 3.2922,
       "step": 59750
     },
     {
       "epoch": 6.44743935309973,
-      "grad_norm": 0.7549540400505066,
-      "learning_rate": 0.00021354560172692929,
-      "loss": 3.281,
+      "grad_norm": 0.7723394632339478,
+      "learning_rate": 0.00021352617377226116,
+      "loss": 3.2833,
       "step": 59800
     },
     {
       "epoch": 6.452830188679245,
-      "grad_norm": 0.6927609443664551,
-      "learning_rate": 0.00021322180248246086,
-      "loss": 3.2651,
+      "grad_norm": 0.6993033289909363,
+      "learning_rate": 0.00021320237452779274,
+      "loss": 3.2674,
       "step": 59850
     },
     {
       "epoch": 6.45822102425876,
-      "grad_norm": 0.7731172442436218,
-      "learning_rate": 0.00021289800323799241,
-      "loss": 3.2776,
+      "grad_norm": 0.7500173449516296,
+      "learning_rate": 0.00021287857528332432,
+      "loss": 3.2794,
       "step": 59900
     },
     {
       "epoch": 6.463611859838275,
-      "grad_norm": 0.7873961329460144,
-      "learning_rate": 0.000212574203993524,
-      "loss": 3.2673,
+      "grad_norm": 0.7128188610076904,
+      "learning_rate": 0.0002125547760388559,
+      "loss": 3.2688,
       "step": 59950
     },
     {
       "epoch": 6.46900269541779,
-      "grad_norm": 0.7329972982406616,
-      "learning_rate": 0.00021225040474905557,
-      "loss": 3.2774,
+      "grad_norm": 0.6932951211929321,
+      "learning_rate": 0.00021223097679438747,
+      "loss": 3.28,
       "step": 60000
     },
     {
       "epoch": 6.46900269541779,
-      "eval_accuracy": 0.38562365834055967,
-      "eval_loss": 3.3711838722229004,
-      "eval_runtime": 185.2128,
-      "eval_samples_per_second": 97.245,
-      "eval_steps_per_second": 6.079,
+      "eval_accuracy": 0.3855733520535016,
+      "eval_loss": 3.3714606761932373,
+      "eval_runtime": 185.4106,
+      "eval_samples_per_second": 97.141,
+      "eval_steps_per_second": 6.073,
       "step": 60000
     },
     {
       "epoch": 6.474393530997305,
-      "grad_norm": 0.7409731149673462,
-      "learning_rate": 0.00021192660550458712,
-      "loss": 3.2753,
+      "grad_norm": 0.7293769121170044,
+      "learning_rate": 0.00021190717754991905,
+      "loss": 3.2759,
       "step": 60050
     },
     {
       "epoch": 6.479784366576819,
-      "grad_norm": 0.7271058559417725,
-      "learning_rate": 0.0002116028062601187,
-      "loss": 3.2901,
+      "grad_norm": 0.7333296537399292,
+      "learning_rate": 0.0002115833783054506,
+      "loss": 3.2925,
       "step": 60100
     },
     {
       "epoch": 6.485175202156334,
-      "grad_norm": 0.7415794730186462,
-      "learning_rate": 0.00021127900701565028,
-      "loss": 3.2682,
+      "grad_norm": 0.73787522315979,
+      "learning_rate": 0.00021125957906098218,
+      "loss": 3.269,
       "step": 60150
     },
     {
       "epoch": 6.490566037735849,
-      "grad_norm": 0.7169997692108154,
-      "learning_rate": 0.00021095520777118185,
-      "loss": 3.2797,
+      "grad_norm": 0.7034549117088318,
+      "learning_rate": 0.00021093577981651373,
+      "loss": 3.2833,
       "step": 60200
     },
     {
       "epoch": 6.495956873315364,
-      "grad_norm": 0.741539478302002,
-      "learning_rate": 0.00021063140852671343,
-      "loss": 3.2738,
+      "grad_norm": 0.7363021969795227,
+      "learning_rate": 0.0002106119805720453,
+      "loss": 3.2787,
       "step": 60250
     },
     {
       "epoch": 6.501347708894879,
-      "grad_norm": 0.7155291438102722,
-      "learning_rate": 0.000210307609282245,
-      "loss": 3.2714,
+      "grad_norm": 0.7120100259780884,
+      "learning_rate": 0.00021028818132757689,
+      "loss": 3.2729,
       "step": 60300
     },
     {
       "epoch": 6.506738544474393,
-      "grad_norm": 0.7458370327949524,
-      "learning_rate": 0.00020998381003777653,
-      "loss": 3.2564,
+      "grad_norm": 0.7493690252304077,
+      "learning_rate": 0.00020996438208310846,
+      "loss": 3.2609,
       "step": 60350
     },
     {
       "epoch": 6.512129380053908,
-      "grad_norm": 0.7324507832527161,
-      "learning_rate": 0.0002096600107933081,
-      "loss": 3.2867,
+      "grad_norm": 0.7664533257484436,
+      "learning_rate": 0.00020964058283864001,
+      "loss": 3.2888,
       "step": 60400
     },
     {
       "epoch": 6.517520215633423,
-      "grad_norm": 0.7247518301010132,
-      "learning_rate": 0.0002093362115488397,
-      "loss": 3.2764,
+      "grad_norm": 0.7788054347038269,
+      "learning_rate": 0.0002093167835941716,
+      "loss": 3.279,
       "step": 60450
     },
     {
       "epoch": 6.5229110512129385,
-      "grad_norm": 0.7751719355583191,
-      "learning_rate": 0.00020901241230437127,
-      "loss": 3.2722,
+      "grad_norm": 0.7385787963867188,
+      "learning_rate": 0.00020899946033459254,
+      "loss": 3.2747,
       "step": 60500
     },
     {
       "epoch": 6.528301886792453,
-      "grad_norm": 0.7711856961250305,
-      "learning_rate": 0.00020868861305990284,
-      "loss": 3.2942,
+      "grad_norm": 0.7137572765350342,
+      "learning_rate": 0.0002086756610901241,
+      "loss": 3.297,
       "step": 60550
     },
     {
       "epoch": 6.533692722371968,
-      "grad_norm": 0.7641790509223938,
-      "learning_rate": 0.00020836481381543442,
-      "loss": 3.2694,
+      "grad_norm": 0.7212474942207336,
+      "learning_rate": 0.0002083712898003238,
+      "loss": 3.2788,
       "step": 60600
     },
     {
       "epoch": 6.539083557951482,
-      "grad_norm": 0.7420713901519775,
-      "learning_rate": 0.000208041014570966,
-      "loss": 3.2843,
+      "grad_norm": 0.7242739796638489,
+      "learning_rate": 0.00020804749055585537,
+      "loss": 3.2879,
       "step": 60650
     },
     {
       "epoch": 6.544474393530997,
-      "grad_norm": 0.6974769830703735,
-      "learning_rate": 0.00020771721532649758,
-      "loss": 3.2811,
+      "grad_norm": 0.7038812637329102,
+      "learning_rate": 0.0002077236913113869,
+      "loss": 3.2846,
       "step": 60700
     },
     {
       "epoch": 6.549865229110512,
-      "grad_norm": 0.7124184370040894,
-      "learning_rate": 0.00020739341608202915,
-      "loss": 3.2855,
+      "grad_norm": 0.7051358819007874,
+      "learning_rate": 0.00020739989206691847,
+      "loss": 3.2898,
       "step": 60750
     },
     {
       "epoch": 6.555256064690027,
-      "grad_norm": 0.7596436142921448,
-      "learning_rate": 0.00020706961683756068,
-      "loss": 3.2815,
+      "grad_norm": 0.7652502655982971,
+      "learning_rate": 0.00020707609282245005,
+      "loss": 3.2851,
       "step": 60800
     },
     {
       "epoch": 6.560646900269542,
-      "grad_norm": 0.7241122722625732,
-      "learning_rate": 0.00020674581759309225,
-      "loss": 3.2703,
+      "grad_norm": 0.788062334060669,
+      "learning_rate": 0.00020675229357798163,
+      "loss": 3.2743,
       "step": 60850
     },
     {
       "epoch": 6.566037735849057,
-      "grad_norm": 0.7312943935394287,
-      "learning_rate": 0.00020642201834862383,
-      "loss": 3.2659,
+      "grad_norm": 0.73066246509552,
+      "learning_rate": 0.0002064284943335132,
+      "loss": 3.2681,
       "step": 60900
     },
     {
       "epoch": 6.571428571428571,
-      "grad_norm": 0.8218500018119812,
-      "learning_rate": 0.0002060982191041554,
-      "loss": 3.285,
+      "grad_norm": 0.7862104773521423,
+      "learning_rate": 0.00020610469508904478,
+      "loss": 3.2857,
       "step": 60950
     },
     {
       "epoch": 6.576819407008086,
-      "grad_norm": 0.7621256113052368,
-      "learning_rate": 0.000205774419859687,
-      "loss": 3.2727,
+      "grad_norm": 0.7408245801925659,
+      "learning_rate": 0.00020578089584457636,
+      "loss": 3.2752,
       "step": 61000
     },
     {
       "epoch": 6.576819407008086,
-      "eval_accuracy": 0.3862397202144678,
-      "eval_loss": 3.3667685985565186,
-      "eval_runtime": 184.7684,
-      "eval_samples_per_second": 97.479,
-      "eval_steps_per_second": 6.094,
+      "eval_accuracy": 0.38576555901204534,
+      "eval_loss": 3.3688933849334717,
+      "eval_runtime": 184.7982,
+      "eval_samples_per_second": 97.463,
+      "eval_steps_per_second": 6.093,
       "step": 61000
     },
     {
       "epoch": 6.5822102425876015,
-      "grad_norm": 0.7284641861915588,
-      "learning_rate": 0.00020545062061521857,
-      "loss": 3.2699,
+      "grad_norm": 0.6936094164848328,
+      "learning_rate": 0.00020545709660010794,
+      "loss": 3.2732,
       "step": 61050
     },
     {
       "epoch": 6.587601078167116,
-      "grad_norm": 0.7356741428375244,
+      "grad_norm": 0.713800311088562,
       "learning_rate": 0.00020513329735563946,
-      "loss": 3.2767,
+      "loss": 3.2782,
       "step": 61100
     },
     {
       "epoch": 6.592991913746631,
-      "grad_norm": 0.7225967645645142,
+      "grad_norm": 0.7378830909729004,
       "learning_rate": 0.00020480949811117104,
-      "loss": 3.2927,
+      "loss": 3.2957,
       "step": 61150
     },
     {
       "epoch": 6.598382749326145,
-      "grad_norm": 0.7218978404998779,
+      "grad_norm": 0.6994268894195557,
       "learning_rate": 0.00020448569886670262,
-      "loss": 3.2904,
+      "loss": 3.2931,
       "step": 61200
     },
     {
       "epoch": 6.60377358490566,
-      "grad_norm": 0.7737917900085449,
+      "grad_norm": 0.7355708479881287,
       "learning_rate": 0.0002041618996222342,
-      "loss": 3.2809,
+      "loss": 3.2839,
       "step": 61250
     },
     {
       "epoch": 6.609164420485175,
-      "grad_norm": 0.7530986666679382,
+      "grad_norm": 0.7560614347457886,
       "learning_rate": 0.00020383810037776577,
-      "loss": 3.2741,
+      "loss": 3.2765,
       "step": 61300
     },
     {
       "epoch": 6.6145552560646905,
-      "grad_norm": 0.7638270854949951,
+      "grad_norm": 0.7597295045852661,
       "learning_rate": 0.00020351430113329735,
-      "loss": 3.2863,
+      "loss": 3.2885,
       "step": 61350
     },
     {
       "epoch": 6.619946091644205,
-      "grad_norm": 0.7740856409072876,
+      "grad_norm": 0.7206094264984131,
       "learning_rate": 0.00020319050188882893,
-      "loss": 3.2869,
+      "loss": 3.2886,
       "step": 61400
     },
     {
       "epoch": 6.62533692722372,
-      "grad_norm": 0.7981337308883667,
+      "grad_norm": 0.7349038124084473,
       "learning_rate": 0.00020286670264436048,
-      "loss": 3.2831,
+      "loss": 3.2849,
       "step": 61450
     },
     {
       "epoch": 6.630727762803234,
-      "grad_norm": 0.7509191632270813,
+      "grad_norm": 0.7283930778503418,
       "learning_rate": 0.00020254290339989206,
-      "loss": 3.2781,
+      "loss": 3.2816,
       "step": 61500
     },
     {
       "epoch": 6.636118598382749,
-      "grad_norm": 0.7379617691040039,
+      "grad_norm": 0.7200408577919006,
       "learning_rate": 0.0002022191041554236,
-      "loss": 3.2753,
+      "loss": 3.2766,
       "step": 61550
     },
     {
       "epoch": 6.6415094339622645,
-      "grad_norm": 0.7517578601837158,
+      "grad_norm": 0.7172255516052246,
       "learning_rate": 0.00020189530491095519,
-      "loss": 3.3039,
+      "loss": 3.3055,
       "step": 61600
     },
     {
       "epoch": 6.646900269541779,
-      "grad_norm": 0.7389500141143799,
+      "grad_norm": 0.7040255665779114,
       "learning_rate": 0.00020157150566648676,
-      "loss": 3.2594,
+      "loss": 3.261,
       "step": 61650
     },
     {
       "epoch": 6.652291105121294,
-      "grad_norm": 0.7633240222930908,
+      "grad_norm": 0.7185490131378174,
       "learning_rate": 0.00020124770642201834,
-      "loss": 3.2779,
+      "loss": 3.2823,
       "step": 61700
     },
     {
       "epoch": 6.657681940700809,
-      "grad_norm": 0.7727910280227661,
+      "grad_norm": 0.7609604597091675,
       "learning_rate": 0.0002009239071775499,
-      "loss": 3.296,
+      "loss": 3.2977,
       "step": 61750
     },
     {
       "epoch": 6.663072776280323,
-      "grad_norm": 0.7852575778961182,
+      "grad_norm": 0.7795951962471008,
       "learning_rate": 0.00020060010793308147,
-      "loss": 3.2801,
+      "loss": 3.2814,
       "step": 61800
     },
     {
       "epoch": 6.668463611859838,
-      "grad_norm": 0.7530646920204163,
+      "grad_norm": 0.7162206172943115,
       "learning_rate": 0.00020027630868861305,
-      "loss": 3.2923,
+      "loss": 3.2936,
       "step": 61850
     },
     {
       "epoch": 6.6738544474393535,
-      "grad_norm": 0.7305129170417786,
+      "grad_norm": 0.7411020398139954,
       "learning_rate": 0.00019995250944414462,
-      "loss": 3.2805,
+      "loss": 3.283,
       "step": 61900
     },
     {
       "epoch": 6.679245283018868,
-      "grad_norm": 0.7686682343482971,
+      "grad_norm": 0.7376359105110168,
       "learning_rate": 0.00019962871019967617,
-      "loss": 3.2771,
+      "loss": 3.2784,
       "step": 61950
     },
     {
       "epoch": 6.684636118598383,
-      "grad_norm": 0.7675058841705322,
+      "grad_norm": 0.7186854481697083,
       "learning_rate": 0.00019930491095520775,
-      "loss": 3.2761,
+      "loss": 3.2773,
       "step": 62000
     },
     {
       "epoch": 6.684636118598383,
-      "eval_accuracy": 0.3863945505796034,
-      "eval_loss": 3.3637261390686035,
-      "eval_runtime": 184.9692,
-      "eval_samples_per_second": 97.373,
-      "eval_steps_per_second": 6.088,
+      "eval_accuracy": 0.38614823448292795,
+      "eval_loss": 3.36369252204895,
+      "eval_runtime": 185.2042,
+      "eval_samples_per_second": 97.249,
+      "eval_steps_per_second": 6.08,
       "step": 62000
     },
     {
       "epoch": 6.690026954177897,
-      "grad_norm": 0.7356290221214294,
+      "grad_norm": 0.7841340899467468,
       "learning_rate": 0.0001989811117107393,
-      "loss": 3.2717,
+      "loss": 3.2752,
       "step": 62050
     },
     {
       "epoch": 6.695417789757412,
-      "grad_norm": 0.7363114953041077,
+      "grad_norm": 0.7519031167030334,
       "learning_rate": 0.00019865731246627088,
-      "loss": 3.2784,
+      "loss": 3.2814,
       "step": 62100
     },
     {
       "epoch": 6.7008086253369274,
-      "grad_norm": 0.7388237714767456,
+      "grad_norm": 0.7419308423995972,
       "learning_rate": 0.00019833351322180246,
-      "loss": 3.2943,
+      "loss": 3.2982,
       "step": 62150
     },
     {
       "epoch": 6.706199460916442,
-      "grad_norm": 0.7899906039237976,
+      "grad_norm": 0.7742794156074524,
       "learning_rate": 0.00019800971397733404,
-      "loss": 3.3021,
+      "loss": 3.3044,
       "step": 62200
     },
     {
       "epoch": 6.711590296495957,
-      "grad_norm": 0.7731553316116333,
-      "learning_rate": 0.000197692390717755,
-      "loss": 3.2766,
+      "grad_norm": 0.8003810048103333,
+      "learning_rate": 0.00019768591473286561,
+      "loss": 3.2773,
       "step": 62250
     },
     {
       "epoch": 6.716981132075472,
-      "grad_norm": 0.7288447618484497,
-      "learning_rate": 0.00019736859147328654,
-      "loss": 3.2821,
+      "grad_norm": 0.7672263383865356,
+      "learning_rate": 0.0001973621154883972,
+      "loss": 3.2837,
       "step": 62300
     },
     {
       "epoch": 6.722371967654986,
-      "grad_norm": 0.7623395323753357,
-      "learning_rate": 0.00019704479222881812,
-      "loss": 3.2845,
+      "grad_norm": 0.7650006413459778,
+      "learning_rate": 0.00019703831624392877,
+      "loss": 3.2861,
       "step": 62350
     },
     {
       "epoch": 6.727762803234501,
-      "grad_norm": 0.7321327328681946,
-      "learning_rate": 0.00019672099298434967,
-      "loss": 3.2676,
+      "grad_norm": 0.7366001009941101,
+      "learning_rate": 0.0001967145169994603,
+      "loss": 3.2697,
       "step": 62400
     },
     {
       "epoch": 6.7331536388140165,
-      "grad_norm": 0.7705351710319519,
-      "learning_rate": 0.00019639719373988124,
-      "loss": 3.273,
+      "grad_norm": 0.7485739588737488,
+      "learning_rate": 0.00019639071775499187,
+      "loss": 3.2756,
       "step": 62450
     },
     {
       "epoch": 6.738544474393531,
-      "grad_norm": 0.80072021484375,
-      "learning_rate": 0.00019607339449541282,
-      "loss": 3.2825,
+      "grad_norm": 0.9604521989822388,
+      "learning_rate": 0.00019606691851052345,
+      "loss": 3.2846,
       "step": 62500
     },
     {
       "epoch": 6.743935309973046,
-      "grad_norm": 0.7516555190086365,
-      "learning_rate": 0.0001957495952509444,
-      "loss": 3.2804,
+      "grad_norm": 0.8006502985954285,
+      "learning_rate": 0.00019574311926605503,
+      "loss": 3.2832,
       "step": 62550
     },
     {
       "epoch": 6.74932614555256,
-      "grad_norm": 0.7890444993972778,
-      "learning_rate": 0.00019542579600647598,
-      "loss": 3.2755,
+      "grad_norm": 0.7882602214813232,
+      "learning_rate": 0.0001954193200215866,
+      "loss": 3.2776,
       "step": 62600
     },
     {
       "epoch": 6.754716981132075,
-      "grad_norm": 0.8166410326957703,
-      "learning_rate": 0.00019510199676200756,
-      "loss": 3.2874,
+      "grad_norm": 0.7486486434936523,
+      "learning_rate": 0.00019509552077711818,
+      "loss": 3.2898,
       "step": 62650
     },
     {
       "epoch": 6.7601078167115904,
-      "grad_norm": 0.7606351971626282,
-      "learning_rate": 0.00019477819751753913,
-      "loss": 3.2774,
+      "grad_norm": 0.7505764961242676,
+      "learning_rate": 0.00019477172153264976,
+      "loss": 3.2781,
       "step": 62700
     },
     {
       "epoch": 6.765498652291106,
-      "grad_norm": 0.7485134601593018,
-      "learning_rate": 0.00019445439827307066,
-      "loss": 3.2952,
+      "grad_norm": 0.7206476330757141,
+      "learning_rate": 0.00019444792228818134,
+      "loss": 3.2958,
       "step": 62750
     },
     {
       "epoch": 6.77088948787062,
-      "grad_norm": 0.7248587012290955,
-      "learning_rate": 0.00019413059902860223,
-      "loss": 3.2912,
+      "grad_norm": 0.7169566750526428,
+      "learning_rate": 0.00019412412304371286,
+      "loss": 3.2926,
       "step": 62800
     },
     {
       "epoch": 6.776280323450135,
-      "grad_norm": 0.7899085879325867,
-      "learning_rate": 0.0001938067997841338,
-      "loss": 3.2829,
+      "grad_norm": 0.8056523203849792,
+      "learning_rate": 0.00019380032379924444,
+      "loss": 3.2864,
       "step": 62850
     },
     {
       "epoch": 6.781671159029649,
-      "grad_norm": 0.7482374310493469,
-      "learning_rate": 0.0001934830005396654,
-      "loss": 3.2766,
+      "grad_norm": 0.7459729313850403,
+      "learning_rate": 0.00019347652455477602,
+      "loss": 3.279,
       "step": 62900
     },
     {
       "epoch": 6.787061994609164,
-      "grad_norm": 0.757675290107727,
-      "learning_rate": 0.00019315920129519697,
-      "loss": 3.2787,
+      "grad_norm": 0.7194493412971497,
+      "learning_rate": 0.0001931527253103076,
+      "loss": 3.2804,
       "step": 62950
     },
     {
       "epoch": 6.7924528301886795,
-      "grad_norm": 0.7783774137496948,
-      "learning_rate": 0.00019283540205072854,
-      "loss": 3.2838,
+      "grad_norm": 0.7365685701370239,
+      "learning_rate": 0.00019282892606583917,
+      "loss": 3.2849,
       "step": 63000
     },
     {
       "epoch": 6.7924528301886795,
-      "eval_accuracy": 0.38692890548187503,
-      "eval_loss": 3.3590385913848877,
-      "eval_runtime": 185.3079,
-      "eval_samples_per_second": 97.195,
-      "eval_steps_per_second": 6.076,
+      "eval_accuracy": 0.38663358693279526,
+      "eval_loss": 3.360600709915161,
+      "eval_runtime": 185.3706,
+      "eval_samples_per_second": 97.162,
+      "eval_steps_per_second": 6.074,
       "step": 63000
     },
     {
       "epoch": 6.797843665768194,
-      "grad_norm": 0.7797094583511353,
-      "learning_rate": 0.00019251160280626012,
-      "loss": 3.2693,
+      "grad_norm": 0.7655932903289795,
+      "learning_rate": 0.00019250512682137075,
+      "loss": 3.2718,
       "step": 63050
     },
     {
       "epoch": 6.803234501347709,
-      "grad_norm": 0.8045250773429871,
-      "learning_rate": 0.0001921878035617917,
-      "loss": 3.2952,
+      "grad_norm": 0.7850986123085022,
+      "learning_rate": 0.0001921813275769023,
+      "loss": 3.2966,
       "step": 63100
     },
     {
       "epoch": 6.808625336927224,
-      "grad_norm": 0.7782381176948547,
-      "learning_rate": 0.00019186400431732322,
-      "loss": 3.2823,
+      "grad_norm": 0.761485755443573,
+      "learning_rate": 0.00019185752833243388,
+      "loss": 3.2845,
       "step": 63150
     },
     {
       "epoch": 6.814016172506738,
-      "grad_norm": 0.7246415019035339,
-      "learning_rate": 0.0001915402050728548,
-      "loss": 3.2807,
+      "grad_norm": 0.7269786596298218,
+      "learning_rate": 0.00019153372908796545,
+      "loss": 3.2811,
       "step": 63200
     },
     {
       "epoch": 6.819407008086253,
-      "grad_norm": 0.7466952204704285,
-      "learning_rate": 0.00019121640582838638,
-      "loss": 3.2889,
+      "grad_norm": 0.7788788080215454,
+      "learning_rate": 0.000191209929843497,
+      "loss": 3.2914,
       "step": 63250
     },
     {
       "epoch": 6.824797843665769,
-      "grad_norm": 0.7936302423477173,
-      "learning_rate": 0.00019089260658391796,
-      "loss": 3.2867,
+      "grad_norm": 0.7736331820487976,
+      "learning_rate": 0.00019088613059902858,
+      "loss": 3.2903,
       "step": 63300
     },
     {
       "epoch": 6.830188679245283,
-      "grad_norm": 0.7542206048965454,
-      "learning_rate": 0.00019056880733944953,
-      "loss": 3.2796,
+      "grad_norm": 0.7407358884811401,
+      "learning_rate": 0.00019056233135456016,
+      "loss": 3.281,
       "step": 63350
     },
     {
       "epoch": 6.835579514824798,
-      "grad_norm": 0.767737865447998,
-      "learning_rate": 0.0001902450080949811,
-      "loss": 3.2729,
+      "grad_norm": 0.7694665193557739,
+      "learning_rate": 0.0001902385321100917,
+      "loss": 3.2754,
       "step": 63400
     },
     {
       "epoch": 6.840970350404312,
-      "grad_norm": 0.7150385975837708,
-      "learning_rate": 0.00018992120885051266,
-      "loss": 3.2794,
+      "grad_norm": 0.746585488319397,
+      "learning_rate": 0.0001899147328656233,
+      "loss": 3.2803,
       "step": 63450
     },
     {
       "epoch": 6.846361185983827,
-      "grad_norm": 0.7381755113601685,
-      "learning_rate": 0.00018959740960604424,
-      "loss": 3.2799,
+      "grad_norm": 0.7027455568313599,
+      "learning_rate": 0.00018959093362115487,
+      "loss": 3.2809,
       "step": 63500
     },
     {
       "epoch": 6.8517520215633425,
-      "grad_norm": 0.7633397579193115,
-      "learning_rate": 0.00018927361036157582,
-      "loss": 3.2966,
+      "grad_norm": 0.723949134349823,
+      "learning_rate": 0.00018926713437668644,
+      "loss": 3.2992,
       "step": 63550
     },
     {
       "epoch": 6.857142857142857,
-      "grad_norm": 0.8028882741928101,
-      "learning_rate": 0.00018894981111710737,
-      "loss": 3.2818,
+      "grad_norm": 0.8229068517684937,
+      "learning_rate": 0.00018894333513221802,
+      "loss": 3.2837,
       "step": 63600
     },
     {
       "epoch": 6.862533692722372,
-      "grad_norm": 0.7695461511611938,
-      "learning_rate": 0.00018862601187263895,
-      "loss": 3.285,
+      "grad_norm": 0.7902956604957581,
+      "learning_rate": 0.00018861953588774957,
+      "loss": 3.2873,
       "step": 63650
     },
     {
       "epoch": 6.867924528301887,
-      "grad_norm": 0.7666114568710327,
-      "learning_rate": 0.00018830221262817052,
-      "loss": 3.2613,
+      "grad_norm": 0.800199031829834,
+      "learning_rate": 0.00018829573664328115,
+      "loss": 3.2631,
       "step": 63700
     },
     {
       "epoch": 6.873315363881401,
-      "grad_norm": 0.7522419691085815,
-      "learning_rate": 0.00018797841338370207,
-      "loss": 3.2706,
+      "grad_norm": 0.7298393249511719,
+      "learning_rate": 0.0001879719373988127,
+      "loss": 3.2751,
       "step": 63750
     },
     {
       "epoch": 6.878706199460916,
-      "grad_norm": 0.7789742946624756,
-      "learning_rate": 0.00018765461413923365,
-      "loss": 3.2844,
+      "grad_norm": 0.7546523213386536,
+      "learning_rate": 0.00018764813815434428,
+      "loss": 3.2867,
       "step": 63800
     },
     {
       "epoch": 6.884097035040432,
-      "grad_norm": 0.735979437828064,
-      "learning_rate": 0.00018733081489476523,
-      "loss": 3.2571,
+      "grad_norm": 0.7286291718482971,
+      "learning_rate": 0.00018732433890987586,
+      "loss": 3.2573,
       "step": 63850
     },
     {
       "epoch": 6.889487870619946,
-      "grad_norm": 0.7391952872276306,
-      "learning_rate": 0.0001870070156502968,
-      "loss": 3.2718,
+      "grad_norm": 0.7267189025878906,
+      "learning_rate": 0.00018700053966540743,
+      "loss": 3.2745,
       "step": 63900
     },
     {
       "epoch": 6.894878706199461,
-      "grad_norm": 0.784604012966156,
-      "learning_rate": 0.00018668321640582838,
-      "loss": 3.2887,
+      "grad_norm": 0.7725000977516174,
+      "learning_rate": 0.000186676740420939,
+      "loss": 3.2904,
       "step": 63950
     },
     {
       "epoch": 6.900269541778976,
-      "grad_norm": 0.7470083832740784,
-      "learning_rate": 0.00018635941716135994,
-      "loss": 3.2991,
+      "grad_norm": 0.7633524537086487,
+      "learning_rate": 0.0001863529411764706,
+      "loss": 3.3004,
       "step": 64000
     },
     {
       "epoch": 6.900269541778976,
-      "eval_accuracy": 0.3873211424068853,
-      "eval_loss": 3.353762626647949,
-      "eval_runtime": 184.9046,
-      "eval_samples_per_second": 97.407,
-      "eval_steps_per_second": 6.09,
+      "eval_accuracy": 0.3869971394954225,
+      "eval_loss": 3.354445219039917,
+      "eval_runtime": 184.9666,
+      "eval_samples_per_second": 97.374,
+      "eval_steps_per_second": 6.088,
       "step": 64000
     },
     {
       "epoch": 6.90566037735849,
-      "grad_norm": 0.7583284378051758,
-      "learning_rate": 0.00018603561791689149,
-      "loss": 3.2804,
+      "grad_norm": 0.7619683146476746,
+      "learning_rate": 0.00018602914193200217,
+      "loss": 3.2826,
       "step": 64050
     },
     {
       "epoch": 6.9110512129380055,
-      "grad_norm": 0.7391751408576965,
-      "learning_rate": 0.00018571181867242306,
-      "loss": 3.2795,
+      "grad_norm": 0.7622038125991821,
+      "learning_rate": 0.0001857053426875337,
+      "loss": 3.28,
       "step": 64100
     },
     {
       "epoch": 6.916442048517521,
-      "grad_norm": 0.7356597781181335,
-      "learning_rate": 0.00018538801942795464,
-      "loss": 3.28,
+      "grad_norm": 0.7104603052139282,
+      "learning_rate": 0.00018538154344306527,
+      "loss": 3.2808,
       "step": 64150
     },
     {
       "epoch": 6.921832884097035,
-      "grad_norm": 0.7817914485931396,
-      "learning_rate": 0.00018506422018348622,
-      "loss": 3.305,
+      "grad_norm": 0.7579168081283569,
+      "learning_rate": 0.00018505774419859684,
+      "loss": 3.3057,
       "step": 64200
     },
     {
       "epoch": 6.92722371967655,
-      "grad_norm": 0.7687154412269592,
-      "learning_rate": 0.0001847404209390178,
-      "loss": 3.2742,
+      "grad_norm": 0.8231496214866638,
+      "learning_rate": 0.00018473394495412842,
+      "loss": 3.2746,
       "step": 64250
     },
     {
       "epoch": 6.932614555256064,
-      "grad_norm": 0.779971182346344,
-      "learning_rate": 0.00018441662169454937,
-      "loss": 3.2807,
+      "grad_norm": 0.7839766144752502,
+      "learning_rate": 0.00018441014570966,
+      "loss": 3.2825,
       "step": 64300
     },
     {
       "epoch": 6.938005390835579,
-      "grad_norm": 0.7690040469169617,
-      "learning_rate": 0.00018409282245008095,
-      "loss": 3.2714,
+      "grad_norm": 0.7433233857154846,
+      "learning_rate": 0.00018408634646519158,
+      "loss": 3.2741,
       "step": 64350
     },
     {
       "epoch": 6.943396226415095,
-      "grad_norm": 0.7556252479553223,
-      "learning_rate": 0.00018376902320561253,
-      "loss": 3.2809,
+      "grad_norm": 0.7666583061218262,
+      "learning_rate": 0.00018376254722072316,
+      "loss": 3.2823,
       "step": 64400
     },
     {
       "epoch": 6.948787061994609,
-      "grad_norm": 0.7381365299224854,
-      "learning_rate": 0.00018344522396114405,
-      "loss": 3.2561,
+      "grad_norm": 0.7528730630874634,
+      "learning_rate": 0.0001834387479762547,
+      "loss": 3.2574,
       "step": 64450
     },
     {
       "epoch": 6.954177897574124,
-      "grad_norm": 0.7619420289993286,
-      "learning_rate": 0.00018312142471667563,
-      "loss": 3.2978,
+      "grad_norm": 0.7804055213928223,
+      "learning_rate": 0.00018311494873178628,
+      "loss": 3.3006,
       "step": 64500
     },
     {
       "epoch": 6.959568733153639,
-      "grad_norm": 0.7599936127662659,
-      "learning_rate": 0.0001827976254722072,
-      "loss": 3.2935,
+      "grad_norm": 0.7621782422065735,
+      "learning_rate": 0.00018279114948731783,
+      "loss": 3.2945,
       "step": 64550
     },
     {
       "epoch": 6.964959568733153,
-      "grad_norm": 0.7568638324737549,
-      "learning_rate": 0.00018247382622773879,
-      "loss": 3.2925,
+      "grad_norm": 0.7308817505836487,
+      "learning_rate": 0.0001824673502428494,
+      "loss": 3.2963,
       "step": 64600
     },
     {
       "epoch": 6.9703504043126685,
-      "grad_norm": 0.8307446241378784,
-      "learning_rate": 0.00018215002698327036,
-      "loss": 3.3057,
+      "grad_norm": 0.795536994934082,
+      "learning_rate": 0.000182143550998381,
+      "loss": 3.3076,
       "step": 64650
     },
     {
       "epoch": 6.975741239892184,
-      "grad_norm": 0.757452666759491,
-      "learning_rate": 0.00018182622773880194,
-      "loss": 3.2963,
+      "grad_norm": 0.7348906397819519,
+      "learning_rate": 0.00018181975175391257,
+      "loss": 3.2982,
       "step": 64700
     },
     {
       "epoch": 6.981132075471698,
-      "grad_norm": 0.7277268767356873,
-      "learning_rate": 0.00018150242849433352,
-      "loss": 3.2846,
+      "grad_norm": 0.7965940237045288,
+      "learning_rate": 0.00018149595250944414,
+      "loss": 3.2848,
       "step": 64750
     },
     {
       "epoch": 6.986522911051213,
-      "grad_norm": 0.7873193621635437,
-      "learning_rate": 0.00018117862924986507,
-      "loss": 3.2625,
+      "grad_norm": 0.752318799495697,
+      "learning_rate": 0.0001811721532649757,
+      "loss": 3.2633,
       "step": 64800
     },
     {
       "epoch": 6.991913746630727,
-      "grad_norm": 0.7263054847717285,
-      "learning_rate": 0.00018085483000539662,
-      "loss": 3.2735,
+      "grad_norm": 0.8075865507125854,
+      "learning_rate": 0.00018084835402050727,
+      "loss": 3.276,
       "step": 64850
     },
     {
       "epoch": 6.997304582210242,
-      "grad_norm": 0.7373244166374207,
-      "learning_rate": 0.0001805310307609282,
-      "loss": 3.2756,
+      "grad_norm": 0.7498311996459961,
+      "learning_rate": 0.00018052455477603885,
+      "loss": 3.278,
       "step": 64900
     },
     {
       "epoch": 7.002695417789758,
-      "grad_norm": 0.7862066030502319,
-      "learning_rate": 0.00018020723151645978,
-      "loss": 3.2283,
+      "grad_norm": 0.7839092016220093,
+      "learning_rate": 0.0001802007555315704,
+      "loss": 3.2299,
       "step": 64950
     },
     {
       "epoch": 7.008086253369272,
-      "grad_norm": 0.7903035879135132,
-      "learning_rate": 0.00017988343227199135,
-      "loss": 3.2075,
+      "grad_norm": 0.816551923751831,
+      "learning_rate": 0.00017987695628710198,
+      "loss": 3.2095,
       "step": 65000
     },
     {
       "epoch": 7.008086253369272,
-      "eval_accuracy": 0.3872658380869877,
-      "eval_loss": 3.3579013347625732,
-      "eval_runtime": 184.825,
-      "eval_samples_per_second": 97.449,
-      "eval_steps_per_second": 6.092,
+      "eval_accuracy": 0.3868770780543875,
+      "eval_loss": 3.35855770111084,
+      "eval_runtime": 185.2641,
+      "eval_samples_per_second": 97.218,
+      "eval_steps_per_second": 6.078,
       "step": 65000
     },
     {
       "epoch": 7.013477088948787,
-      "grad_norm": 0.7548803687095642,
-      "learning_rate": 0.00017955963302752293,
-      "loss": 3.1932,
+      "grad_norm": 0.7568230032920837,
+      "learning_rate": 0.00017955315704263356,
+      "loss": 3.1947,
       "step": 65050
     },
     {
       "epoch": 7.018867924528302,
-      "grad_norm": 0.8561325669288635,
-      "learning_rate": 0.00017923583378305448,
-      "loss": 3.2053,
+      "grad_norm": 0.8540091514587402,
+      "learning_rate": 0.0001792293577981651,
+      "loss": 3.2084,
       "step": 65100
     },
     {
       "epoch": 7.024258760107816,
-      "grad_norm": 0.7893680334091187,
-      "learning_rate": 0.00017891203453858606,
-      "loss": 3.2151,
+      "grad_norm": 0.7811715006828308,
+      "learning_rate": 0.00017890555855369668,
+      "loss": 3.2189,
       "step": 65150
     },
     {
       "epoch": 7.0296495956873315,
-      "grad_norm": 0.7915575504302979,
-      "learning_rate": 0.00017858823529411764,
-      "loss": 3.1957,
+      "grad_norm": 0.7806506752967834,
+      "learning_rate": 0.00017858175930922826,
+      "loss": 3.1967,
       "step": 65200
     },
     {
       "epoch": 7.035040431266847,
-      "grad_norm": 0.7586877346038818,
-      "learning_rate": 0.00017826443604964921,
-      "loss": 3.1878,
+      "grad_norm": 0.7606923580169678,
+      "learning_rate": 0.00017825796006475984,
+      "loss": 3.1901,
       "step": 65250
     },
     {
       "epoch": 7.040431266846361,
-      "grad_norm": 0.7624568939208984,
-      "learning_rate": 0.00017794063680518077,
-      "loss": 3.2072,
+      "grad_norm": 0.7610327005386353,
+      "learning_rate": 0.00017793416082029142,
+      "loss": 3.2096,
       "step": 65300
     },
     {
       "epoch": 7.045822102425876,
-      "grad_norm": 0.7240259647369385,
-      "learning_rate": 0.00017761683756071234,
-      "loss": 3.1874,
+      "grad_norm": 0.7111101746559143,
+      "learning_rate": 0.000177610361575823,
+      "loss": 3.1885,
       "step": 65350
     },
     {
       "epoch": 7.051212938005391,
-      "grad_norm": 0.7831783294677734,
-      "learning_rate": 0.00017729303831624392,
-      "loss": 3.2062,
+      "grad_norm": 0.8230857253074646,
+      "learning_rate": 0.00017728656233135452,
+      "loss": 3.2079,
       "step": 65400
     },
     {
       "epoch": 7.056603773584905,
-      "grad_norm": 0.7812415957450867,
-      "learning_rate": 0.00017696923907177547,
-      "loss": 3.2053,
+      "grad_norm": 0.758479654788971,
+      "learning_rate": 0.0001769627630868861,
+      "loss": 3.2078,
       "step": 65450
     },
     {
       "epoch": 7.061994609164421,
-      "grad_norm": 0.7808825373649597,
-      "learning_rate": 0.00017664543982730705,
-      "loss": 3.2024,
+      "grad_norm": 0.7583780288696289,
+      "learning_rate": 0.00017663896384241767,
+      "loss": 3.2052,
       "step": 65500
     },
     {
       "epoch": 7.067385444743936,
-      "grad_norm": 0.7911673784255981,
-      "learning_rate": 0.00017632164058283863,
-      "loss": 3.2138,
+      "grad_norm": 0.8004472851753235,
+      "learning_rate": 0.00017631516459794925,
+      "loss": 3.2172,
       "step": 65550
     },
     {
       "epoch": 7.07277628032345,
-      "grad_norm": 0.7514714598655701,
-      "learning_rate": 0.0001759978413383702,
-      "loss": 3.2121,
+      "grad_norm": 0.7332755923271179,
+      "learning_rate": 0.00017599136535348083,
+      "loss": 3.2155,
       "step": 65600
     },
     {
       "epoch": 7.078167115902965,
-      "grad_norm": 0.8336493968963623,
-      "learning_rate": 0.00017567404209390178,
-      "loss": 3.1991,
+      "grad_norm": 0.7847483158111572,
+      "learning_rate": 0.0001756675661090124,
+      "loss": 3.2015,
       "step": 65650
     },
     {
       "epoch": 7.083557951482479,
-      "grad_norm": 0.7754457592964172,
-      "learning_rate": 0.00017535024284943333,
-      "loss": 3.211,
+      "grad_norm": 0.7899592518806458,
+      "learning_rate": 0.00017534376686454398,
+      "loss": 3.2144,
       "step": 65700
     },
     {
       "epoch": 7.0889487870619945,
-      "grad_norm": 0.7741825580596924,
-      "learning_rate": 0.00017502644360496488,
-      "loss": 3.2039,
+      "grad_norm": 0.7482497096061707,
+      "learning_rate": 0.00017501996762007556,
+      "loss": 3.2072,
       "step": 65750
     },
     {
       "epoch": 7.09433962264151,
-      "grad_norm": 0.7761595249176025,
-      "learning_rate": 0.00017470264436049646,
-      "loss": 3.2095,
+      "grad_norm": 0.8108317852020264,
+      "learning_rate": 0.00017469616837560709,
+      "loss": 3.2129,
       "step": 65800
     },
     {
       "epoch": 7.099730458221024,
-      "grad_norm": 0.781187117099762,
-      "learning_rate": 0.00017437884511602804,
-      "loss": 3.1897,
+      "grad_norm": 0.7613179087638855,
+      "learning_rate": 0.00017437236913113866,
+      "loss": 3.1923,
       "step": 65850
     },
     {
       "epoch": 7.105121293800539,
-      "grad_norm": 0.7472749948501587,
-      "learning_rate": 0.00017405504587155962,
-      "loss": 3.2028,
+      "grad_norm": 0.7778836488723755,
+      "learning_rate": 0.00017404856988667024,
+      "loss": 3.2073,
       "step": 65900
     },
     {
       "epoch": 7.110512129380054,
-      "grad_norm": 0.7998760342597961,
-      "learning_rate": 0.0001737312466270912,
-      "loss": 3.1907,
+      "grad_norm": 0.8139280676841736,
+      "learning_rate": 0.00017372477064220182,
+      "loss": 3.1943,
       "step": 65950
     },
     {
       "epoch": 7.115902964959568,
-      "grad_norm": 0.7840926051139832,
-      "learning_rate": 0.00017340744738262277,
-      "loss": 3.2182,
+      "grad_norm": 0.8360422849655151,
+      "learning_rate": 0.0001734009713977334,
+      "loss": 3.22,
       "step": 66000
     },
     {
       "epoch": 7.115902964959568,
-      "eval_accuracy": 0.3876529683262707,
-      "eval_loss": 3.3591625690460205,
-      "eval_runtime": 185.3265,
-      "eval_samples_per_second": 97.185,
-      "eval_steps_per_second": 6.076,
+      "eval_accuracy": 0.38747086608629366,
+      "eval_loss": 3.3592610359191895,
+      "eval_runtime": 184.7874,
+      "eval_samples_per_second": 97.469,
+      "eval_steps_per_second": 6.093,
       "step": 66000
     },
     {
       "epoch": 7.121293800539084,
-      "grad_norm": 0.7974547147750854,
-      "learning_rate": 0.00017308364813815435,
-      "loss": 3.2075,
+      "grad_norm": 0.756746768951416,
+      "learning_rate": 0.00017307717215326497,
+      "loss": 3.2077,
       "step": 66050
     },
     {
       "epoch": 7.126684636118599,
-      "grad_norm": 0.7505779266357422,
-      "learning_rate": 0.00017275984889368593,
-      "loss": 3.2087,
+      "grad_norm": 0.7934142351150513,
+      "learning_rate": 0.00017275337290879655,
+      "loss": 3.2103,
       "step": 66100
     },
     {
       "epoch": 7.132075471698113,
-      "grad_norm": 0.7844457030296326,
-      "learning_rate": 0.00017243604964921745,
-      "loss": 3.2189,
+      "grad_norm": 0.8033551573753357,
+      "learning_rate": 0.0001724295736643281,
+      "loss": 3.2184,
       "step": 66150
     },
     {
       "epoch": 7.137466307277628,
-      "grad_norm": 0.7378308773040771,
-      "learning_rate": 0.00017211225040474903,
-      "loss": 3.2015,
+      "grad_norm": 0.7241845726966858,
+      "learning_rate": 0.00017210577441985968,
+      "loss": 3.2062,
       "step": 66200
     },
     {
       "epoch": 7.142857142857143,
-      "grad_norm": 0.7681424617767334,
-      "learning_rate": 0.00017179492714516998,
-      "loss": 3.2217,
+      "grad_norm": 0.7614899277687073,
+      "learning_rate": 0.00017178197517539123,
+      "loss": 3.2244,
       "step": 66250
     },
     {
       "epoch": 7.1482479784366575,
-      "grad_norm": 0.7983142137527466,
-      "learning_rate": 0.00017147112790070156,
-      "loss": 3.2042,
+      "grad_norm": 0.7865104675292969,
+      "learning_rate": 0.0001714581759309228,
+      "loss": 3.2087,
       "step": 66300
     },
     {
       "epoch": 7.153638814016173,
-      "grad_norm": 0.8198614716529846,
-      "learning_rate": 0.00017114732865623313,
-      "loss": 3.2091,
+      "grad_norm": 0.7773383855819702,
+      "learning_rate": 0.00017113437668645439,
+      "loss": 3.213,
       "step": 66350
     },
     {
       "epoch": 7.159029649595688,
-      "grad_norm": 0.794266402721405,
-      "learning_rate": 0.0001708235294117647,
-      "loss": 3.2117,
+      "grad_norm": 0.7900416254997253,
+      "learning_rate": 0.00017081057744198596,
+      "loss": 3.2149,
       "step": 66400
     },
     {
       "epoch": 7.164420485175202,
-      "grad_norm": 0.7538433074951172,
-      "learning_rate": 0.0001704997301672963,
-      "loss": 3.2196,
+      "grad_norm": 0.7785615921020508,
+      "learning_rate": 0.00017048677819751751,
+      "loss": 3.2208,
       "step": 66450
     },
     {
       "epoch": 7.169811320754717,
-      "grad_norm": 0.7691895365715027,
-      "learning_rate": 0.00017017593092282781,
-      "loss": 3.2085,
+      "grad_norm": 0.7936018705368042,
+      "learning_rate": 0.0001701629789530491,
+      "loss": 3.2102,
       "step": 66500
     },
     {
       "epoch": 7.175202156334231,
-      "grad_norm": 0.8115940093994141,
-      "learning_rate": 0.0001698521316783594,
-      "loss": 3.2232,
+      "grad_norm": 0.8073174357414246,
+      "learning_rate": 0.00016983917970858067,
+      "loss": 3.2258,
       "step": 66550
     },
     {
       "epoch": 7.180592991913747,
-      "grad_norm": 0.7686892151832581,
-      "learning_rate": 0.00016952833243389097,
-      "loss": 3.2104,
+      "grad_norm": 0.7533419132232666,
+      "learning_rate": 0.00016951538046411225,
+      "loss": 3.2115,
       "step": 66600
     },
     {
       "epoch": 7.185983827493262,
-      "grad_norm": 0.7379952669143677,
-      "learning_rate": 0.00016920453318942255,
-      "loss": 3.2245,
+      "grad_norm": 0.7639559507369995,
+      "learning_rate": 0.0001691915812196438,
+      "loss": 3.2269,
       "step": 66650
     },
     {
       "epoch": 7.191374663072776,
-      "grad_norm": 0.8049662709236145,
-      "learning_rate": 0.00016888073394495412,
-      "loss": 3.2166,
+      "grad_norm": 0.7794451117515564,
+      "learning_rate": 0.00016886778197517538,
+      "loss": 3.219,
       "step": 66700
     },
     {
       "epoch": 7.196765498652291,
-      "grad_norm": 0.7826871275901794,
-      "learning_rate": 0.0001685569347004857,
-      "loss": 3.2239,
+      "grad_norm": 0.7711024284362793,
+      "learning_rate": 0.00016854398273070693,
+      "loss": 3.2267,
       "step": 66750
     },
     {
       "epoch": 7.202156334231806,
-      "grad_norm": 0.7626354694366455,
-      "learning_rate": 0.00016823313545601725,
-      "loss": 3.1991,
+      "grad_norm": 0.7972269654273987,
+      "learning_rate": 0.0001682201834862385,
+      "loss": 3.2023,
       "step": 66800
     },
     {
       "epoch": 7.2075471698113205,
-      "grad_norm": 0.8057540059089661,
-      "learning_rate": 0.00016790933621154883,
-      "loss": 3.2137,
+      "grad_norm": 0.7715423703193665,
+      "learning_rate": 0.00016789638424177008,
+      "loss": 3.2181,
       "step": 66850
     },
     {
       "epoch": 7.212938005390836,
-      "grad_norm": 0.7812713980674744,
-      "learning_rate": 0.00016758553696708038,
-      "loss": 3.2337,
+      "grad_norm": 0.7480494976043701,
+      "learning_rate": 0.00016757258499730166,
+      "loss": 3.2358,
       "step": 66900
     },
     {
       "epoch": 7.218328840970351,
-      "grad_norm": 0.7958628535270691,
-      "learning_rate": 0.00016726173772261196,
-      "loss": 3.2238,
+      "grad_norm": 0.7672910690307617,
+      "learning_rate": 0.00016724878575283324,
+      "loss": 3.2271,
       "step": 66950
     },
     {
       "epoch": 7.223719676549865,
-      "grad_norm": 0.776133120059967,
-      "learning_rate": 0.00016693793847814354,
-      "loss": 3.2277,
+      "grad_norm": 0.7870936989784241,
+      "learning_rate": 0.00016692498650836481,
+      "loss": 3.23,
       "step": 67000
     },
     {
       "epoch": 7.223719676549865,
-      "eval_accuracy": 0.3875116109192241,
-      "eval_loss": 3.3559296131134033,
-      "eval_runtime": 184.9003,
-      "eval_samples_per_second": 97.409,
-      "eval_steps_per_second": 6.09,
+      "eval_accuracy": 0.38741023777489314,
+      "eval_loss": 3.3569626808166504,
+      "eval_runtime": 185.0967,
+      "eval_samples_per_second": 97.306,
+      "eval_steps_per_second": 6.083,
       "step": 67000
     },
     {
       "epoch": 7.22911051212938,
-      "grad_norm": 0.7707949876785278,
-      "learning_rate": 0.00016661413923367511,
-      "loss": 3.2224,
+      "grad_norm": 0.7603834271430969,
+      "learning_rate": 0.0001666011872638964,
+      "loss": 3.2234,
       "step": 67050
     },
     {
       "epoch": 7.234501347708895,
-      "grad_norm": 0.8489854335784912,
-      "learning_rate": 0.00016629033998920666,
-      "loss": 3.2264,
+      "grad_norm": 0.7955418825149536,
+      "learning_rate": 0.00016627738801942792,
+      "loss": 3.2286,
       "step": 67100
     },
     {
       "epoch": 7.2398921832884096,
-      "grad_norm": 0.7626428604125977,
-      "learning_rate": 0.00016596654074473824,
-      "loss": 3.2253,
+      "grad_norm": 0.7377505302429199,
+      "learning_rate": 0.0001659535887749595,
+      "loss": 3.2262,
       "step": 67150
     },
     {
       "epoch": 7.245283018867925,
-      "grad_norm": 0.7870113253593445,
-      "learning_rate": 0.00016564274150026982,
-      "loss": 3.2146,
+      "grad_norm": 0.7720267176628113,
+      "learning_rate": 0.00016562978953049107,
+      "loss": 3.2166,
       "step": 67200
     },
     {
       "epoch": 7.250673854447439,
-      "grad_norm": 1.6695157289505005,
-      "learning_rate": 0.0001653189422558014,
-      "loss": 3.2152,
+      "grad_norm": 0.8122389316558838,
+      "learning_rate": 0.00016530599028602265,
+      "loss": 3.2182,
       "step": 67250
     },
     {
       "epoch": 7.256064690026954,
-      "grad_norm": 0.7939939498901367,
-      "learning_rate": 0.00016499514301133298,
-      "loss": 3.2307,
+      "grad_norm": 0.8003917336463928,
+      "learning_rate": 0.00016498219104155423,
+      "loss": 3.2336,
       "step": 67300
     },
     {
       "epoch": 7.261455525606469,
-      "grad_norm": 0.79146808385849,
-      "learning_rate": 0.00016467134376686453,
-      "loss": 3.219,
+      "grad_norm": 0.7706378698348999,
+      "learning_rate": 0.0001646583917970858,
+      "loss": 3.2216,
       "step": 67350
     },
     {
       "epoch": 7.2668463611859835,
-      "grad_norm": 0.789994478225708,
-      "learning_rate": 0.0001643475445223961,
-      "loss": 3.2288,
+      "grad_norm": 0.7822121381759644,
+      "learning_rate": 0.00016433459255261738,
+      "loss": 3.2279,
       "step": 67400
     },
     {
       "epoch": 7.272237196765499,
-      "grad_norm": 0.7432640194892883,
-      "learning_rate": 0.00016402374527792765,
-      "loss": 3.2366,
+      "grad_norm": 0.7819550633430481,
+      "learning_rate": 0.00016401079330814896,
+      "loss": 3.2403,
       "step": 67450
     },
     {
       "epoch": 7.277628032345014,
-      "grad_norm": 0.7772762775421143,
-      "learning_rate": 0.00016369994603345923,
-      "loss": 3.232,
+      "grad_norm": 0.7725815176963806,
+      "learning_rate": 0.00016368699406368048,
+      "loss": 3.2323,
       "step": 67500
     },
     {
       "epoch": 7.283018867924528,
-      "grad_norm": 0.8248816132545471,
-      "learning_rate": 0.0001633761467889908,
-      "loss": 3.2216,
+      "grad_norm": 0.8235310316085815,
+      "learning_rate": 0.00016336319481921206,
+      "loss": 3.2266,
       "step": 67550
     },
     {
       "epoch": 7.288409703504043,
-      "grad_norm": 0.7746945023536682,
-      "learning_rate": 0.0001630523475445224,
-      "loss": 3.245,
+      "grad_norm": 0.77774578332901,
+      "learning_rate": 0.00016303939557474364,
+      "loss": 3.2468,
       "step": 67600
     },
     {
       "epoch": 7.293800539083558,
-      "grad_norm": 0.8768308758735657,
-      "learning_rate": 0.00016272854830005396,
-      "loss": 3.2276,
+      "grad_norm": 0.8502614498138428,
+      "learning_rate": 0.00016271559633027522,
+      "loss": 3.2285,
       "step": 67650
     },
     {
       "epoch": 7.2991913746630726,
-      "grad_norm": 0.7637453675270081,
-      "learning_rate": 0.00016240474905558554,
-      "loss": 3.2252,
+      "grad_norm": 0.7677310705184937,
+      "learning_rate": 0.0001623917970858068,
+      "loss": 3.2293,
       "step": 67700
     },
     {
       "epoch": 7.304582210242588,
-      "grad_norm": 0.8112920522689819,
-      "learning_rate": 0.00016208094981111707,
-      "loss": 3.2282,
+      "grad_norm": 0.7754825353622437,
+      "learning_rate": 0.00016206799784133837,
+      "loss": 3.2326,
       "step": 67750
     },
     {
       "epoch": 7.309973045822103,
-      "grad_norm": 0.7835080027580261,
-      "learning_rate": 0.00016175715056664864,
-      "loss": 3.2117,
+      "grad_norm": 0.7810065150260925,
+      "learning_rate": 0.00016174419859686992,
+      "loss": 3.2142,
       "step": 67800
     },
     {
       "epoch": 7.315363881401617,
-      "grad_norm": 0.7686965465545654,
-      "learning_rate": 0.00016143335132218022,
-      "loss": 3.2489,
+      "grad_norm": 0.8359429836273193,
+      "learning_rate": 0.0001614203993524015,
+      "loss": 3.2522,
       "step": 67850
     },
     {
       "epoch": 7.320754716981132,
-      "grad_norm": 0.7612333297729492,
-      "learning_rate": 0.0001611095520777118,
-      "loss": 3.2279,
+      "grad_norm": 0.7552247643470764,
+      "learning_rate": 0.00016109660010793308,
+      "loss": 3.2309,
       "step": 67900
     },
     {
       "epoch": 7.3261455525606465,
-      "grad_norm": 0.8523034453392029,
-      "learning_rate": 0.00016078575283324338,
-      "loss": 3.2315,
+      "grad_norm": 0.813887894153595,
+      "learning_rate": 0.00016077280086346463,
+      "loss": 3.2337,
       "step": 67950
     },
     {
       "epoch": 7.331536388140162,
-      "grad_norm": 0.798283576965332,
-      "learning_rate": 0.00016046195358877495,
-      "loss": 3.21,
+      "grad_norm": 0.8046641945838928,
+      "learning_rate": 0.0001604490016189962,
+      "loss": 3.2111,
       "step": 68000
     },
     {
       "epoch": 7.331536388140162,
-      "eval_accuracy": 0.3883520410064691,
-      "eval_loss": 3.3501393795013428,
-      "eval_runtime": 184.9896,
-      "eval_samples_per_second": 97.362,
-      "eval_steps_per_second": 6.087,
+      "eval_accuracy": 0.38813169294998134,
+      "eval_loss": 3.3522262573242188,
+      "eval_runtime": 185.0686,
+      "eval_samples_per_second": 97.321,
+      "eval_steps_per_second": 6.084,
       "step": 68000
     },
     {
       "epoch": 7.336927223719677,
-      "grad_norm": 0.7581954598426819,
-      "learning_rate": 0.00016013815434430653,
-      "loss": 3.2302,
+      "grad_norm": 0.7817372679710388,
+      "learning_rate": 0.00016012520237452778,
+      "loss": 3.2321,
       "step": 68050
     },
     {
       "epoch": 7.342318059299191,
-      "grad_norm": 0.777908205986023,
-      "learning_rate": 0.0001598143550998381,
-      "loss": 3.235,
+      "grad_norm": 0.7686135768890381,
+      "learning_rate": 0.00015980140313005933,
+      "loss": 3.24,
       "step": 68100
     },
     {
       "epoch": 7.347708894878706,
-      "grad_norm": 0.7884739637374878,
-      "learning_rate": 0.00015949055585536966,
-      "loss": 3.2196,
+      "grad_norm": 0.7931602001190186,
+      "learning_rate": 0.0001594776038855909,
+      "loss": 3.2219,
       "step": 68150
     },
     {
       "epoch": 7.353099730458221,
-      "grad_norm": 0.8096325993537903,
-      "learning_rate": 0.0001591667566109012,
-      "loss": 3.2419,
+      "grad_norm": 0.8089348077774048,
+      "learning_rate": 0.0001591538046411225,
+      "loss": 3.244,
       "step": 68200
     },
     {
       "epoch": 7.3584905660377355,
-      "grad_norm": 0.7861303091049194,
-      "learning_rate": 0.00015884943335132216,
-      "loss": 3.2444,
+      "grad_norm": 0.7867382168769836,
+      "learning_rate": 0.00015883000539665407,
+      "loss": 3.2467,
       "step": 68250
     },
     {
       "epoch": 7.363881401617251,
-      "grad_norm": 0.8135442733764648,
-      "learning_rate": 0.00015852563410685374,
-      "loss": 3.2322,
+      "grad_norm": 0.7838876247406006,
+      "learning_rate": 0.00015850620615218564,
+      "loss": 3.2329,
       "step": 68300
     },
     {
       "epoch": 7.369272237196766,
-      "grad_norm": 0.7923033833503723,
-      "learning_rate": 0.00015820183486238532,
-      "loss": 3.2313,
+      "grad_norm": 0.8173036575317383,
+      "learning_rate": 0.0001581824069077172,
+      "loss": 3.2348,
       "step": 68350
     },
     {
       "epoch": 7.37466307277628,
-      "grad_norm": 0.8019893169403076,
-      "learning_rate": 0.0001578780356179169,
-      "loss": 3.2225,
+      "grad_norm": 0.7778081893920898,
+      "learning_rate": 0.00015785860766324877,
+      "loss": 3.2254,
       "step": 68400
     },
     {
       "epoch": 7.380053908355795,
-      "grad_norm": 0.8459454774856567,
-      "learning_rate": 0.00015755423637344847,
-      "loss": 3.2065,
+      "grad_norm": 0.8123937845230103,
+      "learning_rate": 0.00015753480841878032,
+      "loss": 3.2109,
       "step": 68450
     },
     {
       "epoch": 7.38544474393531,
-      "grad_norm": 0.7834682464599609,
-      "learning_rate": 0.00015723043712898,
-      "loss": 3.2141,
+      "grad_norm": 0.7795190811157227,
+      "learning_rate": 0.0001572110091743119,
+      "loss": 3.214,
       "step": 68500
     },
     {
       "epoch": 7.390835579514825,
-      "grad_norm": 0.7821386456489563,
-      "learning_rate": 0.00015690663788451157,
-      "loss": 3.2233,
+      "grad_norm": 0.7896884083747864,
+      "learning_rate": 0.00015688720992984348,
+      "loss": 3.2268,
       "step": 68550
     },
     {
       "epoch": 7.39622641509434,
-      "grad_norm": 0.7885687351226807,
-      "learning_rate": 0.00015658283864004315,
-      "loss": 3.2407,
+      "grad_norm": 0.7429280877113342,
+      "learning_rate": 0.00015656341068537506,
+      "loss": 3.2429,
       "step": 68600
     },
     {
       "epoch": 7.401617250673855,
-      "grad_norm": 0.8140493035316467,
-      "learning_rate": 0.00015625903939557473,
-      "loss": 3.2106,
+      "grad_norm": 0.8340896368026733,
+      "learning_rate": 0.00015623961144090663,
+      "loss": 3.2127,
       "step": 68650
     },
     {
       "epoch": 7.407008086253369,
-      "grad_norm": 0.8063366413116455,
-      "learning_rate": 0.0001559352401511063,
-      "loss": 3.2162,
+      "grad_norm": 0.8026847839355469,
+      "learning_rate": 0.0001559158121964382,
+      "loss": 3.2176,
       "step": 68700
     },
     {
       "epoch": 7.412398921832884,
-      "grad_norm": 0.8016238212585449,
-      "learning_rate": 0.00015561144090663788,
-      "loss": 3.22,
+      "grad_norm": 0.757952094078064,
+      "learning_rate": 0.0001555920129519698,
+      "loss": 3.2232,
       "step": 68750
     },
     {
       "epoch": 7.4177897574123985,
-      "grad_norm": 0.7843878865242004,
-      "learning_rate": 0.00015528764166216944,
-      "loss": 3.2307,
+      "grad_norm": 0.8089631199836731,
+      "learning_rate": 0.0001552682137075013,
+      "loss": 3.2333,
       "step": 68800
     },
     {
       "epoch": 7.423180592991914,
-      "grad_norm": 0.768501341342926,
-      "learning_rate": 0.000154963842417701,
-      "loss": 3.2143,
+      "grad_norm": 0.7726781964302063,
+      "learning_rate": 0.0001549444144630329,
+      "loss": 3.2168,
       "step": 68850
     },
     {
       "epoch": 7.428571428571429,
-      "grad_norm": 0.7922945618629456,
-      "learning_rate": 0.0001546400431732326,
-      "loss": 3.2301,
+      "grad_norm": 0.7755560874938965,
+      "learning_rate": 0.00015462061521856447,
+      "loss": 3.234,
       "step": 68900
     },
     {
       "epoch": 7.433962264150943,
-      "grad_norm": 0.8232368230819702,
-      "learning_rate": 0.00015431624392876414,
-      "loss": 3.222,
+      "grad_norm": 0.8190703392028809,
+      "learning_rate": 0.00015429681597409605,
+      "loss": 3.2245,
       "step": 68950
     },
     {
       "epoch": 7.439353099730458,
-      "grad_norm": 0.7619295120239258,
-      "learning_rate": 0.00015399244468429572,
-      "loss": 3.2443,
+      "grad_norm": 0.7696461081504822,
+      "learning_rate": 0.00015397301672962762,
+      "loss": 3.2467,
       "step": 69000
     },
     {
       "epoch": 7.439353099730458,
-      "eval_accuracy": 0.3885381634032953,
-      "eval_loss": 3.347168445587158,
-      "eval_runtime": 184.9909,
-      "eval_samples_per_second": 97.362,
-      "eval_steps_per_second": 6.087,
+      "eval_accuracy": 0.38801489109558074,
+      "eval_loss": 3.3500401973724365,
+      "eval_runtime": 185.1521,
+      "eval_samples_per_second": 97.277,
+      "eval_steps_per_second": 6.081,
       "step": 69000
     },
     {
       "epoch": 7.444743935309973,
-      "grad_norm": 0.8314117193222046,
-      "learning_rate": 0.0001536686454398273,
-      "loss": 3.2238,
+      "grad_norm": 0.8240043520927429,
+      "learning_rate": 0.0001536492174851592,
+      "loss": 3.2245,
       "step": 69050
     },
     {
       "epoch": 7.450134770889488,
-      "grad_norm": 0.7766390442848206,
-      "learning_rate": 0.00015334484619535887,
-      "loss": 3.216,
+      "grad_norm": 0.798420786857605,
+      "learning_rate": 0.00015332541824069078,
+      "loss": 3.219,
       "step": 69100
     },
     {
       "epoch": 7.455525606469003,
-      "grad_norm": 0.8217452168464661,
-      "learning_rate": 0.00015302104695089042,
-      "loss": 3.2008,
+      "grad_norm": 0.7799826264381409,
+      "learning_rate": 0.00015300161899622233,
+      "loss": 3.203,
       "step": 69150
     },
     {
       "epoch": 7.460916442048518,
-      "grad_norm": 0.8428397178649902,
-      "learning_rate": 0.000152697247706422,
-      "loss": 3.2424,
+      "grad_norm": 0.8098189830780029,
+      "learning_rate": 0.00015267781975175388,
+      "loss": 3.2447,
       "step": 69200
     },
     {
       "epoch": 7.466307277628032,
-      "grad_norm": 0.7982749342918396,
-      "learning_rate": 0.00015237344846195358,
-      "loss": 3.2283,
+      "grad_norm": 0.7774802446365356,
+      "learning_rate": 0.00015235402050728546,
+      "loss": 3.2306,
       "step": 69250
     },
     {
       "epoch": 7.471698113207547,
-      "grad_norm": 0.7763500809669495,
-      "learning_rate": 0.00015204964921748516,
-      "loss": 3.2182,
+      "grad_norm": 0.8291066884994507,
+      "learning_rate": 0.00015203022126281703,
+      "loss": 3.2186,
       "step": 69300
     },
     {
       "epoch": 7.4770889487870615,
-      "grad_norm": 0.7948207855224609,
-      "learning_rate": 0.0001517258499730167,
-      "loss": 3.2375,
+      "grad_norm": 0.7912863492965698,
+      "learning_rate": 0.0001517064220183486,
+      "loss": 3.2399,
       "step": 69350
     },
     {
       "epoch": 7.482479784366577,
-      "grad_norm": 0.7881017923355103,
-      "learning_rate": 0.00015140205072854829,
-      "loss": 3.2398,
+      "grad_norm": 0.7798894643783569,
+      "learning_rate": 0.0001513826227738802,
+      "loss": 3.242,
       "step": 69400
     },
     {
       "epoch": 7.487870619946092,
-      "grad_norm": 0.7627866864204407,
-      "learning_rate": 0.00015107825148407984,
-      "loss": 3.2324,
+      "grad_norm": 0.7952271699905396,
+      "learning_rate": 0.00015105882352941177,
+      "loss": 3.235,
       "step": 69450
     },
     {
       "epoch": 7.493261455525606,
-      "grad_norm": 0.8260999917984009,
-      "learning_rate": 0.00015075445223961141,
-      "loss": 3.2223,
+      "grad_norm": 0.8175017833709717,
+      "learning_rate": 0.00015073502428494332,
+      "loss": 3.2245,
       "step": 69500
     },
     {
       "epoch": 7.498652291105121,
-      "grad_norm": 0.792458176612854,
-      "learning_rate": 0.000150430652995143,
-      "loss": 3.2263,
+      "grad_norm": 0.8068032264709473,
+      "learning_rate": 0.0001504112250404749,
+      "loss": 3.2297,
       "step": 69550
     },
     {
       "epoch": 7.504043126684636,
-      "grad_norm": 0.8109192252159119,
-      "learning_rate": 0.00015010685375067457,
-      "loss": 3.2224,
+      "grad_norm": 0.8073570132255554,
+      "learning_rate": 0.00015008742579600647,
+      "loss": 3.2234,
       "step": 69600
     },
     {
       "epoch": 7.509433962264151,
-      "grad_norm": 0.7631048560142517,
-      "learning_rate": 0.00014978305450620615,
-      "loss": 3.2264,
+      "grad_norm": 0.7827601432800293,
+      "learning_rate": 0.00014976362655153802,
+      "loss": 3.2288,
       "step": 69650
     },
     {
       "epoch": 7.514824797843666,
-      "grad_norm": 0.8018741607666016,
-      "learning_rate": 0.00014946573124662707,
-      "loss": 3.2373,
+      "grad_norm": 0.7789676189422607,
+      "learning_rate": 0.0001494398273070696,
+      "loss": 3.2391,
       "step": 69700
     },
     {
       "epoch": 7.520215633423181,
-      "grad_norm": 0.8057368397712708,
-      "learning_rate": 0.00014914193200215865,
-      "loss": 3.2268,
+      "grad_norm": 0.7658881545066833,
+      "learning_rate": 0.00014911602806260118,
+      "loss": 3.2284,
       "step": 69750
     },
     {
       "epoch": 7.525606469002695,
-      "grad_norm": 0.7862183451652527,
-      "learning_rate": 0.0001488181327576902,
-      "loss": 3.2417,
+      "grad_norm": 0.7733731269836426,
+      "learning_rate": 0.00014879222881813273,
+      "loss": 3.2451,
       "step": 69800
     },
     {
       "epoch": 7.53099730458221,
-      "grad_norm": 0.7762759923934937,
-      "learning_rate": 0.00014849433351322178,
-      "loss": 3.2157,
+      "grad_norm": 0.8035315275192261,
+      "learning_rate": 0.0001484684295736643,
+      "loss": 3.2174,
       "step": 69850
     },
     {
       "epoch": 7.536388140161725,
-      "grad_norm": 0.7878009676933289,
-      "learning_rate": 0.00014817053426875336,
-      "loss": 3.2251,
+      "grad_norm": 0.7829474210739136,
+      "learning_rate": 0.00014814463032919589,
+      "loss": 3.2271,
       "step": 69900
     },
     {
       "epoch": 7.54177897574124,
-      "grad_norm": 0.8166378140449524,
-      "learning_rate": 0.00014784673502428493,
-      "loss": 3.2327,
+      "grad_norm": 0.8561943173408508,
+      "learning_rate": 0.00014782083108472744,
+      "loss": 3.2351,
       "step": 69950
     },
     {
       "epoch": 7.547169811320755,
-      "grad_norm": 0.8306125402450562,
-      "learning_rate": 0.0001475229357798165,
-      "loss": 3.2315,
+      "grad_norm": 0.8462625741958618,
+      "learning_rate": 0.00014749703184025901,
+      "loss": 3.2316,
       "step": 70000
     },
     {
       "epoch": 7.547169811320755,
-      "eval_accuracy": 0.3888810719172378,
-      "eval_loss": 3.3454511165618896,
-      "eval_runtime": 185.1045,
-      "eval_samples_per_second": 97.302,
-      "eval_steps_per_second": 6.083,
+      "eval_accuracy": 0.38856282760882915,
+      "eval_loss": 3.34738826751709,
+      "eval_runtime": 185.0786,
+      "eval_samples_per_second": 97.315,
+      "eval_steps_per_second": 6.084,
       "step": 70000
     },
     {
       "epoch": 7.55256064690027,
-      "grad_norm": 0.8042418360710144,
-      "learning_rate": 0.00014719913653534806,
-      "loss": 3.2114,
+      "grad_norm": 0.788134753704071,
+      "learning_rate": 0.0001471732325957906,
+      "loss": 3.2129,
       "step": 70050
     },
     {
       "epoch": 7.557951482479784,
-      "grad_norm": 0.7494966983795166,
-      "learning_rate": 0.00014687533729087964,
-      "loss": 3.2326,
+      "grad_norm": 0.7644175887107849,
+      "learning_rate": 0.00014684943335132217,
+      "loss": 3.2347,
       "step": 70100
     },
     {
       "epoch": 7.563342318059299,
-      "grad_norm": 0.7927791476249695,
-      "learning_rate": 0.00014655153804641122,
-      "loss": 3.2304,
+      "grad_norm": 0.8131969571113586,
+      "learning_rate": 0.00014652563410685375,
+      "loss": 3.2308,
       "step": 70150
     },
     {
       "epoch": 7.568733153638814,
-      "grad_norm": 0.7527803182601929,
-      "learning_rate": 0.0001462277388019428,
-      "loss": 3.2309,
+      "grad_norm": 0.7657596468925476,
+      "learning_rate": 0.0001462018348623853,
+      "loss": 3.2316,
       "step": 70200
     },
     {
       "epoch": 7.574123989218329,
-      "grad_norm": 0.7601218819618225,
-      "learning_rate": 0.00014590393955747435,
-      "loss": 3.226,
+      "grad_norm": 0.7949525713920593,
+      "learning_rate": 0.00014587803561791687,
+      "loss": 3.2284,
       "step": 70250
     },
     {
       "epoch": 7.579514824797844,
-      "grad_norm": 0.7645830512046814,
-      "learning_rate": 0.00014558014031300592,
-      "loss": 3.2393,
+      "grad_norm": 0.7503776550292969,
+      "learning_rate": 0.00014555423637344845,
+      "loss": 3.2428,
       "step": 70300
     },
     {
       "epoch": 7.584905660377358,
-      "grad_norm": 0.8330966830253601,
-      "learning_rate": 0.0001452563410685375,
-      "loss": 3.2108,
+      "grad_norm": 0.7979116439819336,
+      "learning_rate": 0.00014523043712898003,
+      "loss": 3.2158,
       "step": 70350
     },
     {
       "epoch": 7.590296495956873,
-      "grad_norm": 0.7954941391944885,
-      "learning_rate": 0.00014493254182406908,
-      "loss": 3.2179,
+      "grad_norm": 0.7930637001991272,
+      "learning_rate": 0.00014490663788451158,
+      "loss": 3.2204,
       "step": 70400
     },
     {
       "epoch": 7.595687331536388,
-      "grad_norm": 0.7679847478866577,
-      "learning_rate": 0.00014460874257960063,
-      "loss": 3.2248,
+      "grad_norm": 0.7918315529823303,
+      "learning_rate": 0.00014458283864004316,
+      "loss": 3.2266,
       "step": 70450
     },
     {
       "epoch": 7.601078167115903,
-      "grad_norm": 0.7807341814041138,
-      "learning_rate": 0.0001442849433351322,
-      "loss": 3.2387,
+      "grad_norm": 0.746759831905365,
+      "learning_rate": 0.00014425903939557474,
+      "loss": 3.2417,
       "step": 70500
     },
     {
       "epoch": 7.606469002695418,
-      "grad_norm": 0.8021829724311829,
-      "learning_rate": 0.00014396114409066378,
-      "loss": 3.2244,
+      "grad_norm": 0.8025341033935547,
+      "learning_rate": 0.00014393524015110631,
+      "loss": 3.2262,
       "step": 70550
     },
     {
       "epoch": 7.611859838274933,
-      "grad_norm": 0.8153308033943176,
-      "learning_rate": 0.00014363734484619536,
-      "loss": 3.251,
+      "grad_norm": 0.7924248576164246,
+      "learning_rate": 0.00014361791689152724,
+      "loss": 3.2529,
       "step": 70600
     },
     {
       "epoch": 7.617250673854447,
-      "grad_norm": 0.8160524964332581,
-      "learning_rate": 0.0001433135456017269,
-      "loss": 3.2256,
+      "grad_norm": 0.810879647731781,
+      "learning_rate": 0.00014329411764705882,
+      "loss": 3.2276,
       "step": 70650
     },
     {
       "epoch": 7.622641509433962,
-      "grad_norm": 0.8022107481956482,
-      "learning_rate": 0.0001429897463572585,
-      "loss": 3.2283,
+      "grad_norm": 0.8111072182655334,
+      "learning_rate": 0.00014297679438747974,
+      "loss": 3.2287,
       "step": 70700
     },
     {
       "epoch": 7.628032345013477,
-      "grad_norm": 0.8894932866096497,
-      "learning_rate": 0.00014266594711279007,
-      "loss": 3.2202,
+      "grad_norm": 0.8351150155067444,
+      "learning_rate": 0.00014265299514301132,
+      "loss": 3.2233,
       "step": 70750
     },
     {
       "epoch": 7.633423180592992,
-      "grad_norm": 0.7843272686004639,
-      "learning_rate": 0.00014234214786832165,
-      "loss": 3.2222,
+      "grad_norm": 0.7544995546340942,
+      "learning_rate": 0.0001423291958985429,
+      "loss": 3.2255,
       "step": 70800
     },
     {
       "epoch": 7.638814016172507,
-      "grad_norm": 0.8247528076171875,
-      "learning_rate": 0.0001420183486238532,
-      "loss": 3.2281,
+      "grad_norm": 0.8261779546737671,
+      "learning_rate": 0.00014200539665407445,
+      "loss": 3.231,
       "step": 70850
     },
     {
       "epoch": 7.644204851752022,
-      "grad_norm": 0.7786036133766174,
-      "learning_rate": 0.00014169454937938477,
-      "loss": 3.2223,
+      "grad_norm": 0.7995532751083374,
+      "learning_rate": 0.00014168159740960602,
+      "loss": 3.2283,
       "step": 70900
     },
     {
       "epoch": 7.649595687331536,
-      "grad_norm": 0.8587459921836853,
-      "learning_rate": 0.00014137075013491635,
-      "loss": 3.2313,
+      "grad_norm": 0.828923225402832,
+      "learning_rate": 0.0001413577981651376,
+      "loss": 3.2328,
       "step": 70950
     },
     {
       "epoch": 7.654986522911051,
-      "grad_norm": 0.8270299434661865,
-      "learning_rate": 0.0001410469508904479,
-      "loss": 3.2304,
+      "grad_norm": 0.8156834244728088,
+      "learning_rate": 0.00014103399892066918,
+      "loss": 3.2337,
       "step": 71000
     },
     {
       "epoch": 7.654986522911051,
-      "eval_accuracy": 0.38917465202011253,
-      "eval_loss": 3.3405866622924805,
-      "eval_runtime": 184.8909,
-      "eval_samples_per_second": 97.414,
-      "eval_steps_per_second": 6.09,
+      "eval_accuracy": 0.389215940117482,
+      "eval_loss": 3.340672254562378,
+      "eval_runtime": 184.7558,
+      "eval_samples_per_second": 97.485,
+      "eval_steps_per_second": 6.095,
       "step": 71000
     },
     {
       "epoch": 7.660377358490566,
-      "grad_norm": 0.8775826692581177,
-      "learning_rate": 0.00014072315164597948,
-      "loss": 3.2339,
+      "grad_norm": 0.820132315158844,
+      "learning_rate": 0.00014071019967620073,
+      "loss": 3.2363,
       "step": 71050
     },
     {
       "epoch": 7.665768194070081,
-      "grad_norm": 0.7978357076644897,
-      "learning_rate": 0.00014039935240151106,
-      "loss": 3.2172,
+      "grad_norm": 0.7700738310813904,
+      "learning_rate": 0.0001403864004317323,
+      "loss": 3.2192,
       "step": 71100
     },
     {
       "epoch": 7.671159029649596,
-      "grad_norm": 0.7993744015693665,
-      "learning_rate": 0.0001400755531570426,
-      "loss": 3.2088,
+      "grad_norm": 0.8299509286880493,
+      "learning_rate": 0.00014006260118726389,
+      "loss": 3.213,
       "step": 71150
     },
     {
       "epoch": 7.67654986522911,
-      "grad_norm": 0.7915191054344177,
-      "learning_rate": 0.00013975175391257419,
-      "loss": 3.2163,
+      "grad_norm": 0.7950608134269714,
+      "learning_rate": 0.00013973880194279546,
+      "loss": 3.2193,
       "step": 71200
     },
     {
       "epoch": 7.681940700808625,
-      "grad_norm": 0.8002047538757324,
-      "learning_rate": 0.00013942795466810576,
-      "loss": 3.2241,
+      "grad_norm": 0.7542383670806885,
+      "learning_rate": 0.00013941500269832704,
+      "loss": 3.2268,
       "step": 71250
     },
     {
       "epoch": 7.6873315363881405,
-      "grad_norm": 0.7947512269020081,
-      "learning_rate": 0.00013910415542363731,
-      "loss": 3.2479,
+      "grad_norm": 0.7638273239135742,
+      "learning_rate": 0.0001390912034538586,
+      "loss": 3.2504,
       "step": 71300
     },
     {
       "epoch": 7.692722371967655,
-      "grad_norm": 0.7719897627830505,
-      "learning_rate": 0.0001387803561791689,
-      "loss": 3.2372,
+      "grad_norm": 0.7596293687820435,
+      "learning_rate": 0.00013876740420939017,
+      "loss": 3.2385,
       "step": 71350
     },
     {
       "epoch": 7.69811320754717,
-      "grad_norm": 0.784153938293457,
-      "learning_rate": 0.00013845655693470047,
-      "loss": 3.2322,
+      "grad_norm": 0.7992427945137024,
+      "learning_rate": 0.00013844360496492175,
+      "loss": 3.234,
       "step": 71400
     },
     {
       "epoch": 7.703504043126685,
-      "grad_norm": 0.8173136711120605,
-      "learning_rate": 0.00013813275769023205,
-      "loss": 3.2203,
+      "grad_norm": 0.8380318284034729,
+      "learning_rate": 0.00013811980572045333,
+      "loss": 3.2229,
       "step": 71450
     },
     {
       "epoch": 7.708894878706199,
-      "grad_norm": 0.830466628074646,
-      "learning_rate": 0.0001378089584457636,
-      "loss": 3.2244,
+      "grad_norm": 0.8700864911079407,
+      "learning_rate": 0.00013779600647598488,
+      "loss": 3.2266,
       "step": 71500
     },
     {
       "epoch": 7.714285714285714,
-      "grad_norm": 0.7629924416542053,
-      "learning_rate": 0.00013748515920129517,
-      "loss": 3.2486,
+      "grad_norm": 0.7592607140541077,
+      "learning_rate": 0.00013747220723151645,
+      "loss": 3.2511,
       "step": 71550
     },
     {
       "epoch": 7.719676549865229,
-      "grad_norm": 0.7888294458389282,
-      "learning_rate": 0.00013716135995682675,
-      "loss": 3.2378,
+      "grad_norm": 0.7742652893066406,
+      "learning_rate": 0.00013714840798704803,
+      "loss": 3.2406,
       "step": 71600
     },
     {
       "epoch": 7.725067385444744,
-      "grad_norm": 0.7922984957695007,
-      "learning_rate": 0.00013683756071235833,
-      "loss": 3.2447,
+      "grad_norm": 0.8119229078292847,
+      "learning_rate": 0.00013682460874257958,
+      "loss": 3.2462,
       "step": 71650
     },
     {
       "epoch": 7.730458221024259,
-      "grad_norm": 0.8440670967102051,
-      "learning_rate": 0.0001365137614678899,
-      "loss": 3.2428,
+      "grad_norm": 0.8372693657875061,
+      "learning_rate": 0.00013650080949811116,
+      "loss": 3.2452,
       "step": 71700
     },
     {
       "epoch": 7.735849056603773,
-      "grad_norm": 0.7881962656974792,
-      "learning_rate": 0.00013618996222342146,
-      "loss": 3.245,
+      "grad_norm": 0.810852587223053,
+      "learning_rate": 0.00013617701025364274,
+      "loss": 3.2478,
       "step": 71750
     },
     {
       "epoch": 7.741239892183288,
-      "grad_norm": 0.824066698551178,
-      "learning_rate": 0.00013586616297895304,
-      "loss": 3.2387,
+      "grad_norm": 0.8200960755348206,
+      "learning_rate": 0.0001358532110091743,
+      "loss": 3.2403,
       "step": 71800
     },
     {
       "epoch": 7.7466307277628035,
-      "grad_norm": 0.8074589967727661,
-      "learning_rate": 0.00013554236373448461,
-      "loss": 3.2506,
+      "grad_norm": 0.8275381922721863,
+      "learning_rate": 0.00013552941176470587,
+      "loss": 3.253,
       "step": 71850
     },
     {
       "epoch": 7.752021563342318,
-      "grad_norm": 0.8384812474250793,
-      "learning_rate": 0.0001352185644900162,
-      "loss": 3.2281,
+      "grad_norm": 0.7997806072235107,
+      "learning_rate": 0.00013520561252023744,
+      "loss": 3.2331,
       "step": 71900
     },
     {
       "epoch": 7.757412398921833,
-      "grad_norm": 0.8503368496894836,
-      "learning_rate": 0.00013489476524554774,
-      "loss": 3.2573,
+      "grad_norm": 0.8084086179733276,
+      "learning_rate": 0.00013488181327576902,
+      "loss": 3.2599,
       "step": 71950
     },
     {
       "epoch": 7.762803234501348,
-      "grad_norm": 0.7815788984298706,
-      "learning_rate": 0.00013457096600107932,
-      "loss": 3.227,
+      "grad_norm": 0.8054947853088379,
+      "learning_rate": 0.00013455801403130057,
+      "loss": 3.2284,
       "step": 72000
     },
     {
       "epoch": 7.762803234501348,
-      "eval_accuracy": 0.38971911164095085,
-      "eval_loss": 3.3369975090026855,
-      "eval_runtime": 185.05,
-      "eval_samples_per_second": 97.33,
-      "eval_steps_per_second": 6.085,
+      "eval_accuracy": 0.38960480880297005,
+      "eval_loss": 3.338857412338257,
+      "eval_runtime": 185.2016,
+      "eval_samples_per_second": 97.251,
+      "eval_steps_per_second": 6.08,
       "step": 72000
     },
     {
       "epoch": 7.768194070080862,
-      "grad_norm": 0.7871066331863403,
-      "learning_rate": 0.0001342471667566109,
-      "loss": 3.2261,
+      "grad_norm": 0.7348754405975342,
+      "learning_rate": 0.00013423421478683215,
+      "loss": 3.2291,
       "step": 72050
     },
     {
       "epoch": 7.773584905660377,
-      "grad_norm": 0.8432220816612244,
-      "learning_rate": 0.00013392336751214248,
-      "loss": 3.2151,
+      "grad_norm": 0.8173696994781494,
+      "learning_rate": 0.00013391041554236373,
+      "loss": 3.22,
       "step": 72100
     },
     {
       "epoch": 7.7789757412398925,
-      "grad_norm": 0.7723524570465088,
-      "learning_rate": 0.00013359956826767403,
-      "loss": 3.229,
+      "grad_norm": 0.7856990098953247,
+      "learning_rate": 0.00013358661629789528,
+      "loss": 3.2322,
       "step": 72150
     },
     {
       "epoch": 7.784366576819407,
-      "grad_norm": 0.7924665808677673,
-      "learning_rate": 0.0001332757690232056,
-      "loss": 3.2323,
+      "grad_norm": 0.8363749980926514,
+      "learning_rate": 0.00013326281705342685,
+      "loss": 3.2338,
       "step": 72200
     },
     {
       "epoch": 7.789757412398922,
-      "grad_norm": 0.8395488262176514,
-      "learning_rate": 0.00013295196977873718,
-      "loss": 3.2297,
+      "grad_norm": 0.8107422590255737,
+      "learning_rate": 0.00013293901780895843,
+      "loss": 3.2322,
       "step": 72250
     },
     {
       "epoch": 7.795148247978437,
-      "grad_norm": 0.8542680144309998,
-      "learning_rate": 0.00013262817053426876,
-      "loss": 3.2238,
+      "grad_norm": 0.8467124104499817,
+      "learning_rate": 0.00013261521856449,
+      "loss": 3.2286,
       "step": 72300
     },
     {
       "epoch": 7.800539083557951,
-      "grad_norm": 0.804538905620575,
-      "learning_rate": 0.0001323043712898003,
-      "loss": 3.2303,
+      "grad_norm": 0.8180131316184998,
+      "learning_rate": 0.00013229141932002156,
+      "loss": 3.2327,
       "step": 72350
     },
     {
       "epoch": 7.8059299191374665,
-      "grad_norm": 0.8471668362617493,
-      "learning_rate": 0.0001319805720453319,
-      "loss": 3.2319,
+      "grad_norm": 0.7756885290145874,
+      "learning_rate": 0.00013196762007555314,
+      "loss": 3.2329,
       "step": 72400
     },
     {
       "epoch": 7.811320754716981,
-      "grad_norm": 0.917702317237854,
-      "learning_rate": 0.00013165677280086346,
-      "loss": 3.2264,
+      "grad_norm": 0.8815840482711792,
+      "learning_rate": 0.00013164382083108472,
+      "loss": 3.2278,
       "step": 72450
     },
     {
       "epoch": 7.816711590296496,
-      "grad_norm": 0.8071184754371643,
-      "learning_rate": 0.00013133297355639502,
-      "loss": 3.2349,
+      "grad_norm": 0.8170946836471558,
+      "learning_rate": 0.0001313200215866163,
+      "loss": 3.2392,
       "step": 72500
     },
     {
       "epoch": 7.822102425876011,
-      "grad_norm": 0.8377707004547119,
-      "learning_rate": 0.0001310091743119266,
-      "loss": 3.2281,
+      "grad_norm": 0.8364844918251038,
+      "learning_rate": 0.00013099622234214784,
+      "loss": 3.2302,
       "step": 72550
     },
     {
       "epoch": 7.827493261455525,
-      "grad_norm": 0.8600809574127197,
-      "learning_rate": 0.00013068537506745817,
-      "loss": 3.2396,
+      "grad_norm": 0.834038496017456,
+      "learning_rate": 0.00013067242309767942,
+      "loss": 3.2414,
       "step": 72600
     },
     {
       "epoch": 7.83288409703504,
-      "grad_norm": 0.7609186768531799,
-      "learning_rate": 0.00013036157582298972,
-      "loss": 3.2438,
+      "grad_norm": 0.7629231810569763,
+      "learning_rate": 0.000130348623853211,
+      "loss": 3.2461,
       "step": 72650
     },
     {
       "epoch": 7.8382749326145555,
-      "grad_norm": 0.7975425720214844,
-      "learning_rate": 0.0001300377765785213,
-      "loss": 3.2397,
+      "grad_norm": 0.7637219429016113,
+      "learning_rate": 0.00013002482460874258,
+      "loss": 3.2435,
       "step": 72700
     },
     {
       "epoch": 7.84366576819407,
-      "grad_norm": 0.8560848236083984,
-      "learning_rate": 0.00012971397733405288,
-      "loss": 3.2357,
+      "grad_norm": 0.8876663446426392,
+      "learning_rate": 0.00012970102536427413,
+      "loss": 3.2362,
       "step": 72750
     },
     {
       "epoch": 7.849056603773585,
-      "grad_norm": 0.8832368850708008,
-      "learning_rate": 0.00012939017808958443,
-      "loss": 3.2582,
+      "grad_norm": 0.8259429931640625,
+      "learning_rate": 0.0001293772261198057,
+      "loss": 3.2611,
       "step": 72800
     },
     {
       "epoch": 7.8544474393531,
-      "grad_norm": 0.8199717402458191,
-      "learning_rate": 0.000129066378845116,
-      "loss": 3.2274,
+      "grad_norm": 0.8051719069480896,
+      "learning_rate": 0.00012905342687533728,
+      "loss": 3.2314,
       "step": 72850
     },
     {
       "epoch": 7.859838274932614,
-      "grad_norm": 0.8567049503326416,
-      "learning_rate": 0.00012874257960064758,
-      "loss": 3.2322,
+      "grad_norm": 0.7907696962356567,
+      "learning_rate": 0.00012872962763086886,
+      "loss": 3.2336,
       "step": 72900
     },
     {
       "epoch": 7.8652291105121295,
-      "grad_norm": 0.7869696021080017,
-      "learning_rate": 0.00012841878035617916,
-      "loss": 3.2524,
+      "grad_norm": 0.7773061990737915,
+      "learning_rate": 0.00012840582838640044,
+      "loss": 3.2552,
       "step": 72950
     },
     {
       "epoch": 7.870619946091644,
-      "grad_norm": 0.790312647819519,
-      "learning_rate": 0.0001280949811117107,
-      "loss": 3.2268,
+      "grad_norm": 0.8075227737426758,
+      "learning_rate": 0.000128082029141932,
+      "loss": 3.2294,
       "step": 73000
     },
     {
       "epoch": 7.870619946091644,
-      "eval_accuracy": 0.3900164945948991,
-      "eval_loss": 3.331883430480957,
-      "eval_runtime": 184.8094,
-      "eval_samples_per_second": 97.457,
-      "eval_steps_per_second": 6.093,
+      "eval_accuracy": 0.38974986040820236,
+      "eval_loss": 3.3344857692718506,
+      "eval_runtime": 185.203,
+      "eval_samples_per_second": 97.25,
+      "eval_steps_per_second": 6.08,
       "step": 73000
     },
     {
       "epoch": 7.876010781671159,
-      "grad_norm": 0.8404228687286377,
-      "learning_rate": 0.0001277711818672423,
-      "loss": 3.2261,
+      "grad_norm": 0.8024936318397522,
+      "learning_rate": 0.00012775822989746357,
+      "loss": 3.228,
       "step": 73050
     },
     {
       "epoch": 7.881401617250674,
-      "grad_norm": 0.8575640320777893,
-      "learning_rate": 0.00012744738262277387,
-      "loss": 3.2354,
+      "grad_norm": 0.8461624979972839,
+      "learning_rate": 0.00012743443065299514,
+      "loss": 3.2375,
       "step": 73100
     },
     {
       "epoch": 7.886792452830189,
-      "grad_norm": 0.8023410439491272,
-      "learning_rate": 0.00012712358337830544,
-      "loss": 3.2411,
+      "grad_norm": 0.8133740425109863,
+      "learning_rate": 0.00012711063140852672,
+      "loss": 3.243,
       "step": 73150
     },
     {
       "epoch": 7.892183288409703,
-      "grad_norm": 0.7935019731521606,
-      "learning_rate": 0.00012679978413383702,
-      "loss": 3.2188,
+      "grad_norm": 0.7898610234260559,
+      "learning_rate": 0.00012678683216405827,
+      "loss": 3.2223,
       "step": 73200
     },
     {
       "epoch": 7.8975741239892185,
-      "grad_norm": 0.8237470984458923,
-      "learning_rate": 0.00012647598488936857,
-      "loss": 3.2369,
+      "grad_norm": 0.8022027015686035,
+      "learning_rate": 0.00012646303291958985,
+      "loss": 3.2392,
       "step": 73250
     },
     {
       "epoch": 7.902964959568733,
-      "grad_norm": 0.8367178440093994,
-      "learning_rate": 0.00012615218564490015,
-      "loss": 3.2359,
+      "grad_norm": 0.8332421779632568,
+      "learning_rate": 0.00012613923367512143,
+      "loss": 3.239,
       "step": 73300
     },
     {
       "epoch": 7.908355795148248,
-      "grad_norm": 0.8428457975387573,
-      "learning_rate": 0.00012582838640043173,
-      "loss": 3.2394,
+      "grad_norm": 0.8167179226875305,
+      "learning_rate": 0.00012581543443065298,
+      "loss": 3.2418,
       "step": 73350
     },
     {
       "epoch": 7.913746630727763,
-      "grad_norm": 0.8914994597434998,
-      "learning_rate": 0.0001255045871559633,
-      "loss": 3.2437,
+      "grad_norm": 0.8627277612686157,
+      "learning_rate": 0.00012549163518618456,
+      "loss": 3.2455,
       "step": 73400
     },
     {
       "epoch": 7.919137466307277,
-      "grad_norm": 0.8220916986465454,
-      "learning_rate": 0.00012518078791149486,
-      "loss": 3.2222,
+      "grad_norm": 0.8191093802452087,
+      "learning_rate": 0.00012516783594171613,
+      "loss": 3.2264,
       "step": 73450
     },
     {
       "epoch": 7.9245283018867925,
-      "grad_norm": 0.8281764388084412,
-      "learning_rate": 0.00012485698866702643,
-      "loss": 3.2444,
+      "grad_norm": 0.8071380853652954,
+      "learning_rate": 0.00012484403669724768,
+      "loss": 3.2465,
       "step": 73500
     },
     {
       "epoch": 7.929919137466308,
-      "grad_norm": 0.8000036478042603,
-      "learning_rate": 0.000124533189422558,
-      "loss": 3.2307,
+      "grad_norm": 0.8399831056594849,
+      "learning_rate": 0.00012452023745277926,
+      "loss": 3.2325,
       "step": 73550
     },
     {
       "epoch": 7.935309973045822,
-      "grad_norm": 0.857599675655365,
-      "learning_rate": 0.0001242093901780896,
-      "loss": 3.2393,
+      "grad_norm": 0.8575718402862549,
+      "learning_rate": 0.00012419643820831084,
+      "loss": 3.2416,
       "step": 73600
     },
     {
       "epoch": 7.940700808625337,
-      "grad_norm": 0.8535449504852295,
-      "learning_rate": 0.00012388559093362114,
-      "loss": 3.2199,
+      "grad_norm": 0.9044980406761169,
+      "learning_rate": 0.0001238726389638424,
+      "loss": 3.222,
       "step": 73650
     },
     {
       "epoch": 7.946091644204852,
-      "grad_norm": 0.8354669213294983,
-      "learning_rate": 0.0001235682676740421,
-      "loss": 3.2238,
+      "grad_norm": 0.815284788608551,
+      "learning_rate": 0.00012354883971937397,
+      "loss": 3.2261,
       "step": 73700
     },
     {
       "epoch": 7.951482479784366,
-      "grad_norm": 0.8469333052635193,
-      "learning_rate": 0.00012324446842957367,
-      "loss": 3.2163,
+      "grad_norm": 0.7698843479156494,
+      "learning_rate": 0.00012322504047490555,
+      "loss": 3.2182,
       "step": 73750
     },
     {
       "epoch": 7.9568733153638815,
-      "grad_norm": 0.8277457356452942,
-      "learning_rate": 0.00012292066918510522,
-      "loss": 3.196,
+      "grad_norm": 0.804814875125885,
+      "learning_rate": 0.00012290124123043712,
+      "loss": 3.1994,
       "step": 73800
     },
     {
       "epoch": 7.962264150943396,
-      "grad_norm": 0.8391556739807129,
-      "learning_rate": 0.0001225968699406368,
-      "loss": 3.2214,
+      "grad_norm": 0.7834069132804871,
+      "learning_rate": 0.00012257744198596867,
+      "loss": 3.223,
       "step": 73850
     },
     {
       "epoch": 7.967654986522911,
-      "grad_norm": 0.8240298628807068,
-      "learning_rate": 0.00012227307069616837,
-      "loss": 3.2273,
+      "grad_norm": 0.8091722130775452,
+      "learning_rate": 0.00012225364274150025,
+      "loss": 3.231,
       "step": 73900
     },
     {
       "epoch": 7.973045822102426,
-      "grad_norm": 0.8295006155967712,
-      "learning_rate": 0.00012194927145169994,
-      "loss": 3.251,
+      "grad_norm": 0.8159931898117065,
+      "learning_rate": 0.00012192984349703183,
+      "loss": 3.2533,
       "step": 73950
     },
     {
       "epoch": 7.97843665768194,
-      "grad_norm": 0.8376235961914062,
-      "learning_rate": 0.0001216254722072315,
-      "loss": 3.2412,
+      "grad_norm": 0.8482950925827026,
+      "learning_rate": 0.0001216060442525634,
+      "loss": 3.2455,
       "step": 74000
     },
     {
       "epoch": 7.97843665768194,
-      "eval_accuracy": 0.39033039278779513,
-      "eval_loss": 3.3302652835845947,
-      "eval_runtime": 185.3426,
-      "eval_samples_per_second": 97.177,
-      "eval_steps_per_second": 6.075,
+      "eval_accuracy": 0.38994032892054115,
+      "eval_loss": 3.3322672843933105,
+      "eval_runtime": 185.1982,
+      "eval_samples_per_second": 97.253,
+      "eval_steps_per_second": 6.08,
       "step": 74000
     },
     {
       "epoch": 7.9838274932614555,
-      "grad_norm": 0.8279843330383301,
-      "learning_rate": 0.00012130167296276308,
-      "loss": 3.2328,
+      "grad_norm": 0.7893300652503967,
+      "learning_rate": 0.00012128224500809496,
+      "loss": 3.2372,
       "step": 74050
     },
     {
       "epoch": 7.989218328840971,
-      "grad_norm": 0.7964900135993958,
-      "learning_rate": 0.00012097787371829464,
-      "loss": 3.2259,
+      "grad_norm": 0.841288685798645,
+      "learning_rate": 0.00012095844576362653,
+      "loss": 3.2274,
       "step": 74100
     },
     {
       "epoch": 7.994609164420485,
-      "grad_norm": 0.8571212887763977,
-      "learning_rate": 0.00012065407447382622,
-      "loss": 3.2181,
+      "grad_norm": 0.9836897253990173,
+      "learning_rate": 0.00012063464651915811,
+      "loss": 3.22,
       "step": 74150
     },
     {
       "epoch": 8.0,
-      "grad_norm": 1.8256008625030518,
-      "learning_rate": 0.00012033027522935779,
-      "loss": 3.2142,
+      "grad_norm": 1.7988165616989136,
+      "learning_rate": 0.00012031084727468969,
+      "loss": 3.2174,
       "step": 74200
     },
     {
       "epoch": 8.005390835579515,
-      "grad_norm": 0.787792980670929,
-      "learning_rate": 0.00012000647598488935,
-      "loss": 3.157,
+      "grad_norm": 0.8070291876792908,
+      "learning_rate": 0.00011998704803022124,
+      "loss": 3.1629,
       "step": 74250
     },
     {
       "epoch": 8.01078167115903,
-      "grad_norm": 0.8355326652526855,
-      "learning_rate": 0.00011968267674042093,
-      "loss": 3.1554,
+      "grad_norm": 0.8832926750183105,
+      "learning_rate": 0.00011966324878575282,
+      "loss": 3.1593,
       "step": 74300
     },
     {
       "epoch": 8.016172506738544,
-      "grad_norm": 0.8311036229133606,
-      "learning_rate": 0.0001193588774959525,
-      "loss": 3.1441,
+      "grad_norm": 0.8364817500114441,
+      "learning_rate": 0.0001193394495412844,
+      "loss": 3.1457,
       "step": 74350
     },
     {
       "epoch": 8.021563342318059,
-      "grad_norm": 0.8239858746528625,
-      "learning_rate": 0.00011903507825148406,
-      "loss": 3.1528,
+      "grad_norm": 0.8132591843605042,
+      "learning_rate": 0.00011901565029681597,
+      "loss": 3.1558,
       "step": 74400
     },
     {
       "epoch": 8.026954177897574,
-      "grad_norm": 0.8341767191886902,
-      "learning_rate": 0.00011871127900701563,
-      "loss": 3.157,
+      "grad_norm": 0.8514834046363831,
+      "learning_rate": 0.00011869185105234754,
+      "loss": 3.1608,
       "step": 74450
     },
     {
       "epoch": 8.032345013477089,
-      "grad_norm": 0.8128954768180847,
-      "learning_rate": 0.00011838747976254721,
-      "loss": 3.1587,
+      "grad_norm": 0.8021072149276733,
+      "learning_rate": 0.0001183680518078791,
+      "loss": 3.1607,
       "step": 74500
     },
     {
       "epoch": 8.037735849056604,
-      "grad_norm": 0.8075172305107117,
-      "learning_rate": 0.00011806368051807879,
-      "loss": 3.1719,
+      "grad_norm": 0.8027331233024597,
+      "learning_rate": 0.00011804425256341068,
+      "loss": 3.1741,
       "step": 74550
     },
     {
       "epoch": 8.04312668463612,
-      "grad_norm": 0.7725398540496826,
-      "learning_rate": 0.00011773988127361035,
-      "loss": 3.1608,
+      "grad_norm": 0.7780774831771851,
+      "learning_rate": 0.00011772045331894224,
+      "loss": 3.1636,
       "step": 74600
     },
     {
       "epoch": 8.048517520215633,
-      "grad_norm": 0.8969375491142273,
-      "learning_rate": 0.00011741608202914192,
-      "loss": 3.1593,
+      "grad_norm": 0.8480037450790405,
+      "learning_rate": 0.00011739665407447382,
+      "loss": 3.164,
       "step": 74650
     },
     {
       "epoch": 8.053908355795148,
-      "grad_norm": 0.7979028224945068,
-      "learning_rate": 0.0001170922827846735,
-      "loss": 3.1489,
+      "grad_norm": 0.7819019556045532,
+      "learning_rate": 0.00011707933081489476,
+      "loss": 3.1516,
       "step": 74700
     },
     {
       "epoch": 8.059299191374663,
-      "grad_norm": 0.8518366813659668,
-      "learning_rate": 0.00011676848354020506,
-      "loss": 3.1652,
+      "grad_norm": 0.7888557314872742,
+      "learning_rate": 0.00011675553157042632,
+      "loss": 3.1707,
       "step": 74750
     },
     {
       "epoch": 8.064690026954178,
-      "grad_norm": 0.8133111000061035,
-      "learning_rate": 0.00011644468429573664,
-      "loss": 3.1594,
+      "grad_norm": 0.8173406720161438,
+      "learning_rate": 0.00011643173232595789,
+      "loss": 3.1638,
       "step": 74800
     },
     {
       "epoch": 8.070080862533693,
-      "grad_norm": 0.8464822769165039,
-      "learning_rate": 0.0001161208850512682,
-      "loss": 3.1603,
+      "grad_norm": 0.9043296575546265,
+      "learning_rate": 0.00011610793308148947,
+      "loss": 3.1623,
       "step": 74850
     },
     {
       "epoch": 8.075471698113208,
-      "grad_norm": 0.8304824829101562,
-      "learning_rate": 0.00011579708580679978,
-      "loss": 3.1717,
+      "grad_norm": 0.8537508249282837,
+      "learning_rate": 0.00011578413383702104,
+      "loss": 3.1742,
       "step": 74900
     },
     {
       "epoch": 8.080862533692722,
-      "grad_norm": 0.8468976616859436,
-      "learning_rate": 0.00011547328656233134,
-      "loss": 3.1532,
+      "grad_norm": 0.8383791446685791,
+      "learning_rate": 0.00011546033459255261,
+      "loss": 3.1563,
       "step": 74950
     },
     {
       "epoch": 8.086253369272237,
-      "grad_norm": 0.8231673240661621,
-      "learning_rate": 0.00011514948731786292,
-      "loss": 3.1538,
+      "grad_norm": 0.7972344756126404,
+      "learning_rate": 0.00011514301133297355,
+      "loss": 3.1569,
       "step": 75000
     },
     {
       "epoch": 8.086253369272237,
-      "eval_accuracy": 0.3899742286215393,
-      "eval_loss": 3.336721420288086,
-      "eval_runtime": 184.8716,
-      "eval_samples_per_second": 97.424,
-      "eval_steps_per_second": 6.091,
+      "eval_accuracy": 0.38994478368894153,
+      "eval_loss": 3.3383166790008545,
+      "eval_runtime": 184.7956,
+      "eval_samples_per_second": 97.464,
+      "eval_steps_per_second": 6.093,
       "step": 75000
     },
     {
       "epoch": 8.091644204851752,
-      "grad_norm": 0.8552824258804321,
-      "learning_rate": 0.00011482568807339448,
-      "loss": 3.1624,
+      "grad_norm": 0.8097219467163086,
+      "learning_rate": 0.00011481921208850512,
+      "loss": 3.1658,
       "step": 75050
     },
     {
       "epoch": 8.097035040431267,
-      "grad_norm": 0.8481261134147644,
-      "learning_rate": 0.00011450188882892605,
-      "loss": 3.153,
+      "grad_norm": 0.7911630272865295,
+      "learning_rate": 0.00011449541284403669,
+      "loss": 3.1553,
       "step": 75100
     },
     {
       "epoch": 8.102425876010782,
-      "grad_norm": 0.8056097626686096,
-      "learning_rate": 0.00011417808958445763,
-      "loss": 3.1682,
+      "grad_norm": 0.8461593985557556,
+      "learning_rate": 0.00011417161359956825,
+      "loss": 3.1711,
       "step": 75150
     },
     {
       "epoch": 8.107816711590296,
-      "grad_norm": 0.8345311284065247,
-      "learning_rate": 0.0001138542903399892,
-      "loss": 3.1789,
+      "grad_norm": 0.7964184284210205,
+      "learning_rate": 0.00011384781435509983,
+      "loss": 3.1809,
       "step": 75200
     },
     {
       "epoch": 8.11320754716981,
-      "grad_norm": 0.7978355884552002,
-      "learning_rate": 0.00011353049109552075,
-      "loss": 3.1634,
+      "grad_norm": 0.789570152759552,
+      "learning_rate": 0.0001135240151106314,
+      "loss": 3.1654,
       "step": 75250
     },
     {
       "epoch": 8.118598382749326,
-      "grad_norm": 0.810292661190033,
-      "learning_rate": 0.00011320669185105233,
-      "loss": 3.1758,
+      "grad_norm": 0.792686939239502,
+      "learning_rate": 0.00011320021586616297,
+      "loss": 3.1782,
       "step": 75300
     },
     {
       "epoch": 8.123989218328841,
-      "grad_norm": 0.8161046504974365,
-      "learning_rate": 0.00011288289260658391,
-      "loss": 3.1726,
+      "grad_norm": 0.835252583026886,
+      "learning_rate": 0.00011287641662169454,
+      "loss": 3.1779,
       "step": 75350
     },
     {
       "epoch": 8.129380053908356,
-      "grad_norm": 0.7999998331069946,
-      "learning_rate": 0.00011255909336211549,
-      "loss": 3.1797,
+      "grad_norm": 0.7725673317909241,
+      "learning_rate": 0.0001125526173772261,
+      "loss": 3.182,
       "step": 75400
     },
     {
       "epoch": 8.134770889487871,
-      "grad_norm": 0.8012352585792542,
-      "learning_rate": 0.00011224177010253641,
-      "loss": 3.1674,
+      "grad_norm": 0.8120366930961609,
+      "learning_rate": 0.00011222881813275768,
+      "loss": 3.1711,
       "step": 75450
     },
     {
       "epoch": 8.140161725067385,
-      "grad_norm": 0.8193981051445007,
-      "learning_rate": 0.00011191797085806799,
-      "loss": 3.1668,
+      "grad_norm": 0.8159014582633972,
+      "learning_rate": 0.00011190501888828925,
+      "loss": 3.1689,
       "step": 75500
     },
     {
       "epoch": 8.1455525606469,
-      "grad_norm": 0.8131717443466187,
-      "learning_rate": 0.00011159417161359957,
-      "loss": 3.1712,
+      "grad_norm": 0.8142615556716919,
+      "learning_rate": 0.00011158121964382083,
+      "loss": 3.1743,
       "step": 75550
     },
     {
       "epoch": 8.150943396226415,
-      "grad_norm": 0.8060713410377502,
-      "learning_rate": 0.00011127037236913112,
-      "loss": 3.1755,
+      "grad_norm": 0.8550556302070618,
+      "learning_rate": 0.00011125742039935238,
+      "loss": 3.1787,
       "step": 75600
     },
     {
       "epoch": 8.15633423180593,
-      "grad_norm": 0.8505592942237854,
-      "learning_rate": 0.0001109465731246627,
-      "loss": 3.16,
+      "grad_norm": 0.7932506799697876,
+      "learning_rate": 0.00011093362115488396,
+      "loss": 3.1626,
       "step": 75650
     },
     {
       "epoch": 8.161725067385445,
-      "grad_norm": 0.8405550122261047,
-      "learning_rate": 0.00011062277388019427,
-      "loss": 3.1734,
+      "grad_norm": 0.8494173884391785,
+      "learning_rate": 0.00011060982191041554,
+      "loss": 3.1756,
       "step": 75700
     },
     {
       "epoch": 8.167115902964959,
-      "grad_norm": 0.8327785134315491,
-      "learning_rate": 0.00011029897463572585,
-      "loss": 3.1702,
+      "grad_norm": 0.8864437341690063,
+      "learning_rate": 0.00011028602266594712,
+      "loss": 3.1729,
       "step": 75750
     },
     {
       "epoch": 8.172506738544474,
-      "grad_norm": 0.838642954826355,
-      "learning_rate": 0.0001099751753912574,
-      "loss": 3.1898,
+      "grad_norm": 0.806351900100708,
+      "learning_rate": 0.00010996222342147867,
+      "loss": 3.1927,
       "step": 75800
     },
     {
       "epoch": 8.177897574123989,
-      "grad_norm": 0.8472077250480652,
-      "learning_rate": 0.00010965137614678898,
-      "loss": 3.1944,
+      "grad_norm": 0.8257133960723877,
+      "learning_rate": 0.00010963842417701024,
+      "loss": 3.1989,
       "step": 75850
     },
     {
       "epoch": 8.183288409703504,
-      "grad_norm": 0.852802038192749,
-      "learning_rate": 0.00010932757690232056,
-      "loss": 3.1576,
+      "grad_norm": 0.8057741522789001,
+      "learning_rate": 0.00010931462493254182,
+      "loss": 3.159,
       "step": 75900
     },
     {
       "epoch": 8.18867924528302,
-      "grad_norm": 0.8112123608589172,
-      "learning_rate": 0.00010900377765785212,
-      "loss": 3.176,
+      "grad_norm": 0.7742595672607422,
+      "learning_rate": 0.00010899082568807339,
+      "loss": 3.1796,
       "step": 75950
     },
     {
       "epoch": 8.194070080862534,
-      "grad_norm": 0.8221772313117981,
-      "learning_rate": 0.0001086799784133837,
-      "loss": 3.1823,
+      "grad_norm": 0.8021875619888306,
+      "learning_rate": 0.00010866702644360495,
+      "loss": 3.185,
       "step": 76000
     },
     {
       "epoch": 8.194070080862534,
-      "eval_accuracy": 0.3900471347092628,
-      "eval_loss": 3.335939407348633,
-      "eval_runtime": 185.1744,
-      "eval_samples_per_second": 97.265,
-      "eval_steps_per_second": 6.081,
+      "eval_accuracy": 0.38998737562096486,
+      "eval_loss": 3.3373751640319824,
+      "eval_runtime": 185.1022,
+      "eval_samples_per_second": 97.303,
+      "eval_steps_per_second": 6.083,
       "step": 76000
     },
     {
       "epoch": 8.199460916442048,
-      "grad_norm": 0.8041417002677917,
-      "learning_rate": 0.00010835617916891526,
-      "loss": 3.1784,
+      "grad_norm": 0.7910264134407043,
+      "learning_rate": 0.00010834322719913653,
+      "loss": 3.1801,
       "step": 76050
     },
     {
       "epoch": 8.204851752021563,
-      "grad_norm": 0.8178136944770813,
-      "learning_rate": 0.00010803237992444683,
-      "loss": 3.1516,
+      "grad_norm": 0.7795304656028748,
+      "learning_rate": 0.00010801942795466809,
+      "loss": 3.155,
       "step": 76100
     },
     {
       "epoch": 8.210242587601078,
-      "grad_norm": 0.8160865902900696,
-      "learning_rate": 0.0001077085806799784,
-      "loss": 3.1834,
+      "grad_norm": 0.8299195170402527,
+      "learning_rate": 0.00010769562871019967,
+      "loss": 3.1855,
       "step": 76150
     },
     {
       "epoch": 8.215633423180593,
-      "grad_norm": 0.8639083504676819,
-      "learning_rate": 0.00010738478143550998,
-      "loss": 3.1641,
+      "grad_norm": 0.8401374220848083,
+      "learning_rate": 0.00010737182946573123,
+      "loss": 3.1657,
       "step": 76200
     },
     {
       "epoch": 8.221024258760108,
-      "grad_norm": 0.8077586889266968,
-      "learning_rate": 0.00010706098219104153,
-      "loss": 3.1617,
+      "grad_norm": 0.7858666777610779,
+      "learning_rate": 0.0001070480302212628,
+      "loss": 3.1654,
       "step": 76250
     },
     {
       "epoch": 8.226415094339623,
-      "grad_norm": 0.8162104487419128,
-      "learning_rate": 0.00010673718294657311,
-      "loss": 3.1768,
+      "grad_norm": 0.8238749504089355,
+      "learning_rate": 0.00010672423097679438,
+      "loss": 3.1798,
       "step": 76300
     },
     {
       "epoch": 8.231805929919137,
-      "grad_norm": 0.8558855056762695,
-      "learning_rate": 0.00010641338370210469,
-      "loss": 3.1665,
+      "grad_norm": 0.8600948452949524,
+      "learning_rate": 0.00010640043173232595,
+      "loss": 3.1702,
       "step": 76350
     },
     {
       "epoch": 8.237196765498652,
-      "grad_norm": 0.817091703414917,
-      "learning_rate": 0.00010608958445763627,
-      "loss": 3.1608,
+      "grad_norm": 0.7855995893478394,
+      "learning_rate": 0.00010607663248785753,
+      "loss": 3.164,
       "step": 76400
     },
     {
       "epoch": 8.242587601078167,
-      "grad_norm": 0.8458694815635681,
-      "learning_rate": 0.00010576578521316782,
-      "loss": 3.1833,
+      "grad_norm": 0.8233217000961304,
+      "learning_rate": 0.00010575283324338908,
+      "loss": 3.1892,
       "step": 76450
     },
     {
       "epoch": 8.247978436657682,
-      "grad_norm": 0.8939496278762817,
-      "learning_rate": 0.0001054419859686994,
-      "loss": 3.1853,
+      "grad_norm": 0.8577181100845337,
+      "learning_rate": 0.00010542903399892066,
+      "loss": 3.1883,
       "step": 76500
     },
     {
       "epoch": 8.253369272237197,
-      "grad_norm": 0.8105524778366089,
-      "learning_rate": 0.00010511818672423097,
-      "loss": 3.1898,
+      "grad_norm": 0.8304762840270996,
+      "learning_rate": 0.00010510523475445224,
+      "loss": 3.1926,
       "step": 76550
     },
     {
       "epoch": 8.25876010781671,
-      "grad_norm": 0.8321719169616699,
-      "learning_rate": 0.00010479438747976254,
-      "loss": 3.1769,
+      "grad_norm": 0.8044326305389404,
+      "learning_rate": 0.0001047814355099838,
+      "loss": 3.1809,
       "step": 76600
     },
     {
       "epoch": 8.264150943396226,
-      "grad_norm": 0.8863265514373779,
-      "learning_rate": 0.0001044705882352941,
-      "loss": 3.1695,
+      "grad_norm": 0.8440804481506348,
+      "learning_rate": 0.00010445763626551537,
+      "loss": 3.1729,
       "step": 76650
     },
     {
       "epoch": 8.269541778975741,
-      "grad_norm": 0.8106926679611206,
-      "learning_rate": 0.00010414678899082568,
-      "loss": 3.1919,
+      "grad_norm": 0.8171114921569824,
+      "learning_rate": 0.00010413383702104694,
+      "loss": 3.1942,
       "step": 76700
     },
     {
       "epoch": 8.274932614555256,
-      "grad_norm": 0.8519572019577026,
-      "learning_rate": 0.00010382298974635726,
-      "loss": 3.1769,
+      "grad_norm": 0.8302549123764038,
+      "learning_rate": 0.00010381003777657852,
+      "loss": 3.1802,
       "step": 76750
     },
     {
       "epoch": 8.280323450134771,
-      "grad_norm": 0.8171219229698181,
-      "learning_rate": 0.00010349919050188882,
-      "loss": 3.1732,
+      "grad_norm": 0.7697561979293823,
+      "learning_rate": 0.00010348623853211008,
+      "loss": 3.1741,
       "step": 76800
     },
     {
       "epoch": 8.285714285714286,
-      "grad_norm": 0.8138165473937988,
-      "learning_rate": 0.0001031753912574204,
-      "loss": 3.1747,
+      "grad_norm": 0.7852888703346252,
+      "learning_rate": 0.00010316243928764165,
+      "loss": 3.1784,
       "step": 76850
     },
     {
       "epoch": 8.2911051212938,
-      "grad_norm": 0.8623942732810974,
-      "learning_rate": 0.00010285159201295196,
-      "loss": 3.1877,
+      "grad_norm": 0.8156181573867798,
+      "learning_rate": 0.00010283864004317323,
+      "loss": 3.1924,
       "step": 76900
     },
     {
       "epoch": 8.296495956873315,
-      "grad_norm": 0.8288198709487915,
-      "learning_rate": 0.00010252779276848353,
-      "loss": 3.1723,
+      "grad_norm": 0.8077170848846436,
+      "learning_rate": 0.00010251484079870479,
+      "loss": 3.1766,
       "step": 76950
     },
     {
       "epoch": 8.30188679245283,
-      "grad_norm": 0.8703798651695251,
-      "learning_rate": 0.0001022039935240151,
-      "loss": 3.1717,
+      "grad_norm": 0.8286692500114441,
+      "learning_rate": 0.00010219104155423637,
+      "loss": 3.1749,
       "step": 77000
     },
     {
       "epoch": 8.30188679245283,
-      "eval_accuracy": 0.3907528352156179,
-      "eval_loss": 3.3312199115753174,
-      "eval_runtime": 185.2399,
-      "eval_samples_per_second": 97.231,
-      "eval_steps_per_second": 6.079,
+      "eval_accuracy": 0.390606914387283,
+      "eval_loss": 3.3321518898010254,
+      "eval_runtime": 185.3128,
+      "eval_samples_per_second": 97.192,
+      "eval_steps_per_second": 6.076,
       "step": 77000
     },
     {
       "epoch": 8.307277628032345,
-      "grad_norm": 0.8199688196182251,
-      "learning_rate": 0.00010188019427954668,
-      "loss": 3.1804,
+      "grad_norm": 0.7854719758033752,
+      "learning_rate": 0.0001018737182946573,
+      "loss": 3.1833,
       "step": 77050
     },
     {
       "epoch": 8.31266846361186,
-      "grad_norm": 0.8455313444137573,
-      "learning_rate": 0.00010155639503507823,
-      "loss": 3.1848,
+      "grad_norm": 0.8314083218574524,
+      "learning_rate": 0.00010154991905018887,
+      "loss": 3.1864,
       "step": 77100
     },
     {
       "epoch": 8.318059299191376,
-      "grad_norm": 0.8879739046096802,
-      "learning_rate": 0.00010123259579060981,
-      "loss": 3.1738,
+      "grad_norm": 0.8903999328613281,
+      "learning_rate": 0.00010122611980572045,
+      "loss": 3.1781,
       "step": 77150
     },
     {
       "epoch": 8.323450134770889,
-      "grad_norm": 0.849547266960144,
-      "learning_rate": 0.00010090879654614139,
-      "loss": 3.1672,
+      "grad_norm": 0.8407829999923706,
+      "learning_rate": 0.00010090232056125201,
+      "loss": 3.1695,
       "step": 77200
     },
     {
       "epoch": 8.328840970350404,
-      "grad_norm": 0.8199661374092102,
-      "learning_rate": 0.00010058499730167296,
-      "loss": 3.1712,
+      "grad_norm": 0.7816826105117798,
+      "learning_rate": 0.00010057852131678359,
+      "loss": 3.1748,
       "step": 77250
     },
     {
       "epoch": 8.33423180592992,
-      "grad_norm": 0.9070252180099487,
-      "learning_rate": 0.00010026119805720452,
-      "loss": 3.1767,
+      "grad_norm": 0.830484926700592,
+      "learning_rate": 0.00010025472207231515,
+      "loss": 3.1788,
       "step": 77300
     },
     {
       "epoch": 8.339622641509434,
-      "grad_norm": 0.8134100437164307,
-      "learning_rate": 9.993739881273609e-05,
-      "loss": 3.1671,
+      "grad_norm": 0.8447725772857666,
+      "learning_rate": 9.993092282784673e-05,
+      "loss": 3.168,
       "step": 77350
     },
     {
       "epoch": 8.34501347708895,
-      "grad_norm": 0.8291438221931458,
-      "learning_rate": 9.961359956826767e-05,
-      "loss": 3.2002,
+      "grad_norm": 0.8344404101371765,
+      "learning_rate": 9.96071235833783e-05,
+      "loss": 3.2012,
       "step": 77400
     },
     {
       "epoch": 8.350404312668463,
-      "grad_norm": 0.8317015767097473,
-      "learning_rate": 9.928980032379923e-05,
-      "loss": 3.1749,
+      "grad_norm": 0.863575279712677,
+      "learning_rate": 9.928332433890986e-05,
+      "loss": 3.1774,
       "step": 77450
     },
     {
       "epoch": 8.355795148247978,
-      "grad_norm": 0.8489733338356018,
-      "learning_rate": 9.896600107933081e-05,
-      "loss": 3.1672,
+      "grad_norm": 0.8597997426986694,
+      "learning_rate": 9.895952509444144e-05,
+      "loss": 3.1703,
       "step": 77500
     },
     {
       "epoch": 8.361185983827493,
-      "grad_norm": 0.8064641952514648,
-      "learning_rate": 9.864220183486238e-05,
-      "loss": 3.1891,
+      "grad_norm": 0.8186684846878052,
+      "learning_rate": 9.863572584997302e-05,
+      "loss": 3.192,
       "step": 77550
     },
     {
       "epoch": 8.366576819407008,
-      "grad_norm": 0.8400290012359619,
-      "learning_rate": 9.831840259039394e-05,
-      "loss": 3.1807,
+      "grad_norm": 0.8455076217651367,
+      "learning_rate": 9.831192660550457e-05,
+      "loss": 3.1844,
       "step": 77600
     },
     {
       "epoch": 8.371967654986523,
-      "grad_norm": 0.8481228351593018,
-      "learning_rate": 9.799460334592552e-05,
-      "loss": 3.1785,
+      "grad_norm": 0.8197668790817261,
+      "learning_rate": 9.798812736103614e-05,
+      "loss": 3.1818,
       "step": 77650
     },
     {
       "epoch": 8.377358490566039,
-      "grad_norm": 0.8531959652900696,
-      "learning_rate": 9.76708041014571e-05,
-      "loss": 3.1811,
+      "grad_norm": 0.8690318465232849,
+      "learning_rate": 9.766432811656772e-05,
+      "loss": 3.1833,
       "step": 77700
     },
     {
       "epoch": 8.382749326145552,
-      "grad_norm": 0.8155908584594727,
-      "learning_rate": 9.734700485698865e-05,
-      "loss": 3.1734,
+      "grad_norm": 0.8059806227684021,
+      "learning_rate": 9.73405288720993e-05,
+      "loss": 3.176,
       "step": 77750
     },
     {
       "epoch": 8.388140161725067,
-      "grad_norm": 0.8197899460792542,
-      "learning_rate": 9.702320561252022e-05,
-      "loss": 3.1783,
+      "grad_norm": 0.7999092936515808,
+      "learning_rate": 9.701672962763086e-05,
+      "loss": 3.1828,
       "step": 77800
     },
     {
       "epoch": 8.393530997304582,
-      "grad_norm": 0.785799503326416,
-      "learning_rate": 9.66994063680518e-05,
-      "loss": 3.1647,
+      "grad_norm": 0.7501765489578247,
+      "learning_rate": 9.669293038316243e-05,
+      "loss": 3.1667,
       "step": 77850
     },
     {
       "epoch": 8.398921832884097,
-      "grad_norm": 0.8561741709709167,
-      "learning_rate": 9.637560712358338e-05,
-      "loss": 3.172,
+      "grad_norm": 0.7897477149963379,
+      "learning_rate": 9.6369131138694e-05,
+      "loss": 3.1748,
       "step": 77900
     },
     {
       "epoch": 8.404312668463612,
-      "grad_norm": 0.8725782036781311,
-      "learning_rate": 9.605180787911493e-05,
-      "loss": 3.1793,
+      "grad_norm": 0.824909508228302,
+      "learning_rate": 9.604533189422557e-05,
+      "loss": 3.1822,
       "step": 77950
     },
     {
       "epoch": 8.409703504043126,
-      "grad_norm": 0.8261856436729431,
-      "learning_rate": 9.572800863464651e-05,
-      "loss": 3.1889,
+      "grad_norm": 0.8260047435760498,
+      "learning_rate": 9.572153264975715e-05,
+      "loss": 3.1912,
       "step": 78000
     },
     {
       "epoch": 8.409703504043126,
-      "eval_accuracy": 0.3913926920719573,
-      "eval_loss": 3.326793670654297,
-      "eval_runtime": 184.8117,
-      "eval_samples_per_second": 97.456,
-      "eval_steps_per_second": 6.093,
+      "eval_accuracy": 0.3911025888614927,
+      "eval_loss": 3.328890085220337,
+      "eval_runtime": 184.8301,
+      "eval_samples_per_second": 97.446,
+      "eval_steps_per_second": 6.092,
       "step": 78000
     },
     {
       "epoch": 8.415094339622641,
-      "grad_norm": 0.8366762399673462,
-      "learning_rate": 9.540420939017809e-05,
-      "loss": 3.1671,
+      "grad_norm": 0.8345364332199097,
+      "learning_rate": 9.539773340528871e-05,
+      "loss": 3.1706,
       "step": 78050
     },
     {
       "epoch": 8.420485175202156,
-      "grad_norm": 0.8470886945724487,
-      "learning_rate": 9.508041014570966e-05,
-      "loss": 3.1645,
+      "grad_norm": 0.8107349872589111,
+      "learning_rate": 9.507393416082027e-05,
+      "loss": 3.1668,
       "step": 78100
     },
     {
       "epoch": 8.425876010781671,
-      "grad_norm": 0.8738471865653992,
-      "learning_rate": 9.475661090124121e-05,
-      "loss": 3.1871,
+      "grad_norm": 0.873961865901947,
+      "learning_rate": 9.475013491635185e-05,
+      "loss": 3.1897,
       "step": 78150
     },
     {
       "epoch": 8.431266846361186,
-      "grad_norm": 0.8640100955963135,
-      "learning_rate": 9.443281165677279e-05,
-      "loss": 3.1946,
+      "grad_norm": 0.8412361741065979,
+      "learning_rate": 9.442633567188343e-05,
+      "loss": 3.1988,
       "step": 78200
     },
     {
       "epoch": 8.436657681940702,
-      "grad_norm": 0.8384734392166138,
-      "learning_rate": 9.410901241230437e-05,
-      "loss": 3.1726,
+      "grad_norm": 0.8533317446708679,
+      "learning_rate": 9.410253642741498e-05,
+      "loss": 3.1762,
       "step": 78250
     },
     {
       "epoch": 8.442048517520215,
-      "grad_norm": 0.8679486513137817,
-      "learning_rate": 9.378521316783593e-05,
-      "loss": 3.1655,
+      "grad_norm": 0.8340447545051575,
+      "learning_rate": 9.377873718294656e-05,
+      "loss": 3.1693,
       "step": 78300
     },
     {
       "epoch": 8.44743935309973,
-      "grad_norm": 0.8307101130485535,
-      "learning_rate": 9.346141392336751e-05,
-      "loss": 3.1704,
+      "grad_norm": 0.8339968323707581,
+      "learning_rate": 9.345493793847814e-05,
+      "loss": 3.173,
       "step": 78350
     },
     {
       "epoch": 8.452830188679245,
-      "grad_norm": 0.8091768622398376,
-      "learning_rate": 9.313761467889907e-05,
-      "loss": 3.1647,
+      "grad_norm": 0.800408124923706,
+      "learning_rate": 9.313113869400971e-05,
+      "loss": 3.168,
       "step": 78400
     },
     {
       "epoch": 8.45822102425876,
-      "grad_norm": 0.8790152668952942,
-      "learning_rate": 9.281381543443064e-05,
-      "loss": 3.1761,
+      "grad_norm": 0.9167557954788208,
+      "learning_rate": 9.280733944954126e-05,
+      "loss": 3.1787,
       "step": 78450
     },
     {
       "epoch": 8.463611859838275,
-      "grad_norm": 0.8548306822776794,
-      "learning_rate": 9.249001618996222e-05,
-      "loss": 3.1806,
+      "grad_norm": 0.8427416086196899,
+      "learning_rate": 9.248354020507284e-05,
+      "loss": 3.1839,
       "step": 78500
     },
     {
       "epoch": 8.46900269541779,
-      "grad_norm": 0.8710213303565979,
-      "learning_rate": 9.21662169454938e-05,
-      "loss": 3.1752,
+      "grad_norm": 0.8728715181350708,
+      "learning_rate": 9.215974096060442e-05,
+      "loss": 3.1779,
       "step": 78550
     },
     {
       "epoch": 8.474393530997304,
-      "grad_norm": 0.8840611577033997,
-      "learning_rate": 9.184241770102534e-05,
-      "loss": 3.1864,
+      "grad_norm": 0.8693714141845703,
+      "learning_rate": 9.1835941716136e-05,
+      "loss": 3.1893,
       "step": 78600
     },
     {
       "epoch": 8.479784366576819,
-      "grad_norm": 0.8206093907356262,
-      "learning_rate": 9.151861845655692e-05,
-      "loss": 3.1722,
+      "grad_norm": 0.8074076771736145,
+      "learning_rate": 9.151214247166756e-05,
+      "loss": 3.1732,
       "step": 78650
     },
     {
       "epoch": 8.485175202156334,
-      "grad_norm": 0.8537209033966064,
-      "learning_rate": 9.11948192120885e-05,
-      "loss": 3.1831,
+      "grad_norm": 0.8128515481948853,
+      "learning_rate": 9.118834322719913e-05,
+      "loss": 3.1889,
       "step": 78700
     },
     {
       "epoch": 8.49056603773585,
-      "grad_norm": 0.8518295884132385,
-      "learning_rate": 9.087101996762008e-05,
-      "loss": 3.1831,
+      "grad_norm": 0.8658214807510376,
+      "learning_rate": 9.08645439827307e-05,
+      "loss": 3.187,
       "step": 78750
     },
     {
       "epoch": 8.495956873315365,
-      "grad_norm": 0.8618102073669434,
-      "learning_rate": 9.054722072315163e-05,
-      "loss": 3.193,
+      "grad_norm": 0.922846257686615,
+      "learning_rate": 9.054074473826227e-05,
+      "loss": 3.1961,
       "step": 78800
     },
     {
       "epoch": 8.501347708894878,
-      "grad_norm": 0.8804048299789429,
-      "learning_rate": 9.02234214786832e-05,
-      "loss": 3.1665,
+      "grad_norm": 0.7982033491134644,
+      "learning_rate": 9.021694549379385e-05,
+      "loss": 3.1698,
       "step": 78850
     },
     {
       "epoch": 8.506738544474393,
-      "grad_norm": 0.807671844959259,
-      "learning_rate": 8.989962223421478e-05,
-      "loss": 3.1756,
+      "grad_norm": 0.8339309692382812,
+      "learning_rate": 8.989314624932541e-05,
+      "loss": 3.1777,
       "step": 78900
     },
     {
       "epoch": 8.512129380053908,
-      "grad_norm": 0.8216286301612854,
-      "learning_rate": 8.957582298974635e-05,
-      "loss": 3.176,
+      "grad_norm": 0.7865361571311951,
+      "learning_rate": 8.956934700485697e-05,
+      "loss": 3.1804,
       "step": 78950
     },
     {
       "epoch": 8.517520215633423,
-      "grad_norm": 0.9073687791824341,
-      "learning_rate": 8.925202374527791e-05,
-      "loss": 3.1733,
+      "grad_norm": 0.8305343389511108,
+      "learning_rate": 8.924554776038855e-05,
+      "loss": 3.1771,
       "step": 79000
     },
     {
       "epoch": 8.517520215633423,
-      "eval_accuracy": 0.3913608567758277,
-      "eval_loss": 3.324925184249878,
-      "eval_runtime": 185.0203,
-      "eval_samples_per_second": 97.346,
-      "eval_steps_per_second": 6.086,
+      "eval_accuracy": 0.3911895111717443,
+      "eval_loss": 3.326295852661133,
+      "eval_runtime": 185.0767,
+      "eval_samples_per_second": 97.316,
+      "eval_steps_per_second": 6.084,
       "step": 79000
     },
     {
       "epoch": 8.522911051212938,
-      "grad_norm": 0.8593799471855164,
-      "learning_rate": 8.892822450080949e-05,
-      "loss": 3.1864,
+      "grad_norm": 0.8321008682250977,
+      "learning_rate": 8.892174851592013e-05,
+      "loss": 3.1894,
       "step": 79050
     },
     {
       "epoch": 8.528301886792454,
-      "grad_norm": 0.8459751605987549,
-      "learning_rate": 8.860442525634107e-05,
-      "loss": 3.1834,
+      "grad_norm": 0.8203441500663757,
+      "learning_rate": 8.859794927145168e-05,
+      "loss": 3.1856,
       "step": 79100
     },
     {
       "epoch": 8.533692722371967,
-      "grad_norm": 0.9012542963027954,
-      "learning_rate": 8.828062601187263e-05,
-      "loss": 3.1701,
+      "grad_norm": 0.8526754379272461,
+      "learning_rate": 8.827415002698326e-05,
+      "loss": 3.1741,
       "step": 79150
     },
     {
       "epoch": 8.539083557951482,
-      "grad_norm": 0.8935059309005737,
-      "learning_rate": 8.795682676740421e-05,
-      "loss": 3.1855,
+      "grad_norm": 0.8762194514274597,
+      "learning_rate": 8.795035078251483e-05,
+      "loss": 3.1888,
       "step": 79200
     },
     {
       "epoch": 8.544474393530997,
-      "grad_norm": 0.8865764141082764,
-      "learning_rate": 8.763950350782515e-05,
-      "loss": 3.1828,
+      "grad_norm": 0.822303831577301,
+      "learning_rate": 8.762655153804641e-05,
+      "loss": 3.1832,
       "step": 79250
     },
     {
       "epoch": 8.549865229110512,
-      "grad_norm": 0.8719243407249451,
-      "learning_rate": 8.731570426335671e-05,
-      "loss": 3.1741,
+      "grad_norm": 0.8480991125106812,
+      "learning_rate": 8.730275229357798e-05,
+      "loss": 3.1773,
       "step": 79300
     },
     {
       "epoch": 8.555256064690028,
-      "grad_norm": 0.8688287734985352,
-      "learning_rate": 8.699190501888828e-05,
-      "loss": 3.1992,
+      "grad_norm": 0.8900675177574158,
+      "learning_rate": 8.697895304910954e-05,
+      "loss": 3.2036,
       "step": 79350
     },
     {
       "epoch": 8.560646900269543,
-      "grad_norm": 0.8754032254219055,
-      "learning_rate": 8.666810577441985e-05,
-      "loss": 3.1734,
+      "grad_norm": 0.8703182339668274,
+      "learning_rate": 8.665515380464112e-05,
+      "loss": 3.1759,
       "step": 79400
     },
     {
       "epoch": 8.566037735849056,
-      "grad_norm": 0.8592960834503174,
-      "learning_rate": 8.634430652995142e-05,
-      "loss": 3.1913,
+      "grad_norm": 0.845158040523529,
+      "learning_rate": 8.633135456017268e-05,
+      "loss": 3.1934,
       "step": 79450
     },
     {
       "epoch": 8.571428571428571,
-      "grad_norm": 0.8773514628410339,
-      "learning_rate": 8.6020507285483e-05,
-      "loss": 3.1695,
+      "grad_norm": 0.9047236442565918,
+      "learning_rate": 8.600755531570426e-05,
+      "loss": 3.1729,
       "step": 79500
     },
     {
       "epoch": 8.576819407008086,
-      "grad_norm": 1.158023715019226,
-      "learning_rate": 8.569670804101456e-05,
-      "loss": 3.1684,
+      "grad_norm": 0.8393561244010925,
+      "learning_rate": 8.568375607123582e-05,
+      "loss": 3.1729,
       "step": 79550
     },
     {
       "epoch": 8.582210242587601,
-      "grad_norm": 0.8309387564659119,
-      "learning_rate": 8.537290879654612e-05,
-      "loss": 3.1909,
+      "grad_norm": 0.811484158039093,
+      "learning_rate": 8.53599568267674e-05,
+      "loss": 3.192,
       "step": 79600
     },
     {
       "epoch": 8.587601078167117,
-      "grad_norm": 0.8641684055328369,
-      "learning_rate": 8.50491095520777e-05,
-      "loss": 3.1692,
+      "grad_norm": 0.8052199482917786,
+      "learning_rate": 8.503615758229897e-05,
+      "loss": 3.1714,
       "step": 79650
     },
     {
       "epoch": 8.59299191374663,
-      "grad_norm": 0.9096925854682922,
-      "learning_rate": 8.472531030760928e-05,
-      "loss": 3.1758,
+      "grad_norm": 0.9134714007377625,
+      "learning_rate": 8.471235833783054e-05,
+      "loss": 3.1793,
       "step": 79700
     },
     {
       "epoch": 8.598382749326145,
-      "grad_norm": 0.8475040793418884,
-      "learning_rate": 8.440151106314086e-05,
-      "loss": 3.176,
+      "grad_norm": 0.8519928455352783,
+      "learning_rate": 8.438855909336211e-05,
+      "loss": 3.1796,
       "step": 79750
     },
     {
       "epoch": 8.60377358490566,
-      "grad_norm": 0.8620229959487915,
-      "learning_rate": 8.407771181867241e-05,
-      "loss": 3.1735,
+      "grad_norm": 0.842822790145874,
+      "learning_rate": 8.406475984889367e-05,
+      "loss": 3.1762,
       "step": 79800
     },
     {
       "epoch": 8.609164420485175,
-      "grad_norm": 0.8247419595718384,
-      "learning_rate": 8.375391257420398e-05,
-      "loss": 3.172,
+      "grad_norm": 0.7788348197937012,
+      "learning_rate": 8.374096060442525e-05,
+      "loss": 3.1748,
       "step": 79850
     },
     {
       "epoch": 8.61455525606469,
-      "grad_norm": 0.8362515568733215,
-      "learning_rate": 8.343011332973556e-05,
-      "loss": 3.187,
+      "grad_norm": 0.8229261040687561,
+      "learning_rate": 8.341716135995683e-05,
+      "loss": 3.1902,
       "step": 79900
     },
     {
       "epoch": 8.619946091644206,
-      "grad_norm": 0.8225242495536804,
-      "learning_rate": 8.310631408526714e-05,
-      "loss": 3.1912,
+      "grad_norm": 0.8361340165138245,
+      "learning_rate": 8.309336211548838e-05,
+      "loss": 3.1947,
       "step": 79950
     },
     {
       "epoch": 8.625336927223719,
-      "grad_norm": 0.7935653328895569,
-      "learning_rate": 8.278251484079869e-05,
-      "loss": 3.1708,
+      "grad_norm": 0.7869669795036316,
+      "learning_rate": 8.276956287101996e-05,
+      "loss": 3.1761,
       "step": 80000
     },
     {
       "epoch": 8.625336927223719,
-      "eval_accuracy": 0.3918086153265112,
-      "eval_loss": 3.321773052215576,
-      "eval_runtime": 185.1286,
-      "eval_samples_per_second": 97.289,
-      "eval_steps_per_second": 6.082,
+      "eval_accuracy": 0.3917423370649443,
+      "eval_loss": 3.3230607509613037,
+      "eval_runtime": 185.1693,
+      "eval_samples_per_second": 97.268,
+      "eval_steps_per_second": 6.081,
       "step": 80000
     },
     {
       "epoch": 8.630727762803234,
-      "grad_norm": 0.8339737057685852,
-      "learning_rate": 8.245871559633027e-05,
-      "loss": 3.1691,
+      "grad_norm": 0.8103577494621277,
+      "learning_rate": 8.244576362655153e-05,
+      "loss": 3.1737,
       "step": 80050
     },
     {
       "epoch": 8.63611859838275,
-      "grad_norm": 0.8542758822441101,
-      "learning_rate": 8.213491635186185e-05,
-      "loss": 3.1696,
+      "grad_norm": 0.8197165131568909,
+      "learning_rate": 8.212196438208311e-05,
+      "loss": 3.1753,
       "step": 80100
     },
     {
       "epoch": 8.641509433962264,
-      "grad_norm": 0.8964957594871521,
-      "learning_rate": 8.181111710739341e-05,
-      "loss": 3.1944,
+      "grad_norm": 0.9167805910110474,
+      "learning_rate": 8.179816513761467e-05,
+      "loss": 3.1966,
       "step": 80150
     },
     {
       "epoch": 8.64690026954178,
-      "grad_norm": 0.8350257277488708,
-      "learning_rate": 8.148731786292497e-05,
-      "loss": 3.1732,
+      "grad_norm": 0.8052075505256653,
+      "learning_rate": 8.147436589314624e-05,
+      "loss": 3.178,
       "step": 80200
     },
     {
       "epoch": 8.652291105121293,
-      "grad_norm": 0.8604266047477722,
-      "learning_rate": 8.116351861845655e-05,
-      "loss": 3.1959,
+      "grad_norm": 0.8420871496200562,
+      "learning_rate": 8.115056664867782e-05,
+      "loss": 3.1983,
       "step": 80250
     },
     {
       "epoch": 8.657681940700808,
-      "grad_norm": 0.8317278027534485,
-      "learning_rate": 8.083971937398812e-05,
-      "loss": 3.1793,
+      "grad_norm": 0.8218672275543213,
+      "learning_rate": 8.082676740420938e-05,
+      "loss": 3.1813,
       "step": 80300
     },
     {
       "epoch": 8.663072776280323,
-      "grad_norm": 0.8608930706977844,
-      "learning_rate": 8.05159201295197e-05,
-      "loss": 3.1823,
+      "grad_norm": 0.8376367092132568,
+      "learning_rate": 8.050296815974096e-05,
+      "loss": 3.1852,
       "step": 80350
     },
     {
       "epoch": 8.668463611859838,
-      "grad_norm": 0.885753333568573,
-      "learning_rate": 8.019212088505126e-05,
-      "loss": 3.1764,
+      "grad_norm": 0.8482858538627625,
+      "learning_rate": 8.017916891527252e-05,
+      "loss": 3.1787,
       "step": 80400
     },
     {
       "epoch": 8.673854447439354,
-      "grad_norm": 0.8569355010986328,
-      "learning_rate": 7.986832164058282e-05,
-      "loss": 3.1739,
+      "grad_norm": 0.8449885249137878,
+      "learning_rate": 7.985536967080409e-05,
+      "loss": 3.1745,
       "step": 80450
     },
     {
       "epoch": 8.679245283018869,
-      "grad_norm": 0.8573786616325378,
-      "learning_rate": 7.95445223961144e-05,
-      "loss": 3.1716,
+      "grad_norm": 0.8417103290557861,
+      "learning_rate": 7.953157042633566e-05,
+      "loss": 3.1751,
       "step": 80500
     },
     {
       "epoch": 8.684636118598382,
-      "grad_norm": 0.948824405670166,
-      "learning_rate": 7.922072315164598e-05,
-      "loss": 3.1796,
+      "grad_norm": 0.9321690201759338,
+      "learning_rate": 7.920777118186724e-05,
+      "loss": 3.1817,
       "step": 80550
     },
     {
       "epoch": 8.690026954177897,
-      "grad_norm": 0.8927190899848938,
-      "learning_rate": 7.889692390717755e-05,
-      "loss": 3.1731,
+      "grad_norm": 0.8538167476654053,
+      "learning_rate": 7.888397193739879e-05,
+      "loss": 3.1763,
       "step": 80600
     },
     {
       "epoch": 8.695417789757412,
-      "grad_norm": 0.876378059387207,
-      "learning_rate": 7.85731246627091e-05,
-      "loss": 3.1849,
+      "grad_norm": 0.8338154554367065,
+      "learning_rate": 7.856017269293037e-05,
+      "loss": 3.1875,
       "step": 80650
     },
     {
       "epoch": 8.700808625336927,
-      "grad_norm": 0.8848509192466736,
-      "learning_rate": 7.824932541824068e-05,
-      "loss": 3.1881,
+      "grad_norm": 0.8730599880218506,
+      "learning_rate": 7.823637344846195e-05,
+      "loss": 3.1911,
       "step": 80700
     },
     {
       "epoch": 8.706199460916443,
-      "grad_norm": 0.8308282494544983,
-      "learning_rate": 7.792552617377226e-05,
-      "loss": 3.1794,
+      "grad_norm": 0.8520699739456177,
+      "learning_rate": 7.791257420399353e-05,
+      "loss": 3.1807,
       "step": 80750
     },
     {
       "epoch": 8.711590296495956,
-      "grad_norm": 0.8364676237106323,
-      "learning_rate": 7.760172692930382e-05,
-      "loss": 3.151,
+      "grad_norm": 0.8210703730583191,
+      "learning_rate": 7.758877495952508e-05,
+      "loss": 3.1525,
       "step": 80800
     },
     {
       "epoch": 8.716981132075471,
-      "grad_norm": 0.8687278032302856,
-      "learning_rate": 7.727792768483539e-05,
-      "loss": 3.1811,
+      "grad_norm": 0.8489146828651428,
+      "learning_rate": 7.726497571505665e-05,
+      "loss": 3.1826,
       "step": 80850
     },
     {
       "epoch": 8.722371967654986,
-      "grad_norm": 0.8405010104179382,
-      "learning_rate": 7.695412844036697e-05,
-      "loss": 3.1711,
+      "grad_norm": 0.816367506980896,
+      "learning_rate": 7.694117647058823e-05,
+      "loss": 3.1729,
       "step": 80900
     },
     {
       "epoch": 8.727762803234501,
-      "grad_norm": 0.8751876950263977,
-      "learning_rate": 7.663032919589854e-05,
-      "loss": 3.1721,
+      "grad_norm": 0.8610575199127197,
+      "learning_rate": 7.661737722611981e-05,
+      "loss": 3.1732,
       "step": 80950
     },
     {
       "epoch": 8.733153638814017,
-      "grad_norm": 0.8680465817451477,
-      "learning_rate": 7.630652995143011e-05,
-      "loss": 3.1854,
+      "grad_norm": 0.8475139141082764,
+      "learning_rate": 7.629357798165137e-05,
+      "loss": 3.1889,
       "step": 81000
     },
     {
       "epoch": 8.733153638814017,
-      "eval_accuracy": 0.3922179107549083,
-      "eval_loss": 3.3174245357513428,
-      "eval_runtime": 185.0813,
-      "eval_samples_per_second": 97.314,
-      "eval_steps_per_second": 6.084,
+      "eval_accuracy": 0.3920232047799448,
+      "eval_loss": 3.319646120071411,
+      "eval_runtime": 185.4331,
+      "eval_samples_per_second": 97.129,
+      "eval_steps_per_second": 6.072,
       "step": 81000
     },
     {
       "epoch": 8.738544474393532,
-      "grad_norm": 0.8618085384368896,
-      "learning_rate": 7.598273070696167e-05,
-      "loss": 3.194,
+      "grad_norm": 0.8521168231964111,
+      "learning_rate": 7.596977873718294e-05,
+      "loss": 3.1971,
       "step": 81050
     },
     {
       "epoch": 8.743935309973045,
-      "grad_norm": 0.8313191533088684,
-      "learning_rate": 7.565893146249325e-05,
-      "loss": 3.1953,
+      "grad_norm": 0.8344953060150146,
+      "learning_rate": 7.565245547760389e-05,
+      "loss": 3.1993,
       "step": 81100
     },
     {
       "epoch": 8.74932614555256,
-      "grad_norm": 0.875768780708313,
-      "learning_rate": 7.533513221802481e-05,
-      "loss": 3.1711,
+      "grad_norm": 0.8405035138130188,
+      "learning_rate": 7.532865623313544e-05,
+      "loss": 3.1744,
       "step": 81150
     },
     {
       "epoch": 8.754716981132075,
-      "grad_norm": 0.8554214239120483,
+      "grad_norm": 0.8584547638893127,
       "learning_rate": 7.501133297355639e-05,
-      "loss": 3.1901,
+      "loss": 3.1933,
       "step": 81200
     },
     {
       "epoch": 8.76010781671159,
-      "grad_norm": 0.8244317173957825,
+      "grad_norm": 0.8843427896499634,
       "learning_rate": 7.468753372908796e-05,
-      "loss": 3.1755,
+      "loss": 3.1785,
       "step": 81250
     },
     {
       "epoch": 8.765498652291106,
-      "grad_norm": 0.8972374200820923,
+      "grad_norm": 0.9259549975395203,
       "learning_rate": 7.436373448461953e-05,
-      "loss": 3.162,
+      "loss": 3.1642,
       "step": 81300
     },
     {
       "epoch": 8.77088948787062,
-      "grad_norm": 0.8698784112930298,
+      "grad_norm": 0.8856974244117737,
       "learning_rate": 7.40399352401511e-05,
-      "loss": 3.1873,
+      "loss": 3.1904,
       "step": 81350
     },
     {
       "epoch": 8.776280323450134,
-      "grad_norm": 0.8932819962501526,
+      "grad_norm": 0.8694860935211182,
       "learning_rate": 7.371613599568268e-05,
-      "loss": 3.2003,
+      "loss": 3.2025,
       "step": 81400
     },
     {
       "epoch": 8.78167115902965,
-      "grad_norm": 0.8630633354187012,
+      "grad_norm": 0.8153300285339355,
       "learning_rate": 7.339233675121424e-05,
-      "loss": 3.1777,
+      "loss": 3.1802,
       "step": 81450
     },
     {
       "epoch": 8.787061994609164,
-      "grad_norm": 0.8544013500213623,
+      "grad_norm": 0.8006735444068909,
       "learning_rate": 7.306853750674582e-05,
-      "loss": 3.1815,
+      "loss": 3.1858,
       "step": 81500
     },
     {
       "epoch": 8.79245283018868,
-      "grad_norm": 0.8761833906173706,
+      "grad_norm": 0.873319685459137,
       "learning_rate": 7.274473826227738e-05,
-      "loss": 3.182,
+      "loss": 3.1848,
       "step": 81550
     },
     {
       "epoch": 8.797843665768195,
-      "grad_norm": 0.830359935760498,
+      "grad_norm": 0.8179894089698792,
       "learning_rate": 7.242093901780896e-05,
-      "loss": 3.157,
+      "loss": 3.1606,
       "step": 81600
     },
     {
       "epoch": 8.80323450134771,
-      "grad_norm": 0.8646902441978455,
+      "grad_norm": 1.1829814910888672,
       "learning_rate": 7.209713977334052e-05,
-      "loss": 3.1892,
+      "loss": 3.1938,
       "step": 81650
     },
     {
       "epoch": 8.808625336927223,
-      "grad_norm": 0.8650271892547607,
+      "grad_norm": 0.8417500257492065,
       "learning_rate": 7.17733405288721e-05,
-      "loss": 3.1926,
+      "loss": 3.1965,
       "step": 81700
     },
     {
       "epoch": 8.814016172506738,
-      "grad_norm": 0.8778845071792603,
+      "grad_norm": 0.849403977394104,
       "learning_rate": 7.144954128440366e-05,
-      "loss": 3.182,
+      "loss": 3.1847,
       "step": 81750
     },
     {
       "epoch": 8.819407008086253,
-      "grad_norm": 0.852878749370575,
+      "grad_norm": 0.8199416399002075,
       "learning_rate": 7.112574203993523e-05,
-      "loss": 3.1683,
+      "loss": 3.1699,
       "step": 81800
     },
     {
       "epoch": 8.824797843665769,
-      "grad_norm": 0.8865038752555847,
+      "grad_norm": 0.8438077569007874,
       "learning_rate": 7.08019427954668e-05,
-      "loss": 3.1871,
+      "loss": 3.19,
       "step": 81850
     },
     {
       "epoch": 8.830188679245284,
-      "grad_norm": 0.9244481921195984,
+      "grad_norm": 0.8480911254882812,
       "learning_rate": 7.047814355099837e-05,
-      "loss": 3.186,
+      "loss": 3.1902,
       "step": 81900
     },
     {
       "epoch": 8.835579514824797,
-      "grad_norm": 0.9070996642112732,
+      "grad_norm": 0.8652701377868652,
       "learning_rate": 7.015434430652993e-05,
-      "loss": 3.174,
+      "loss": 3.1776,
       "step": 81950
     },
     {
       "epoch": 8.840970350404312,
-      "grad_norm": 0.8721534013748169,
+      "grad_norm": 0.8693273067474365,
       "learning_rate": 6.983054506206151e-05,
-      "loss": 3.1885,
+      "loss": 3.1946,
       "step": 82000
     },
     {
       "epoch": 8.840970350404312,
-      "eval_accuracy": 0.39263633302588186,
-      "eval_loss": 3.313530445098877,
-      "eval_runtime": 185.0624,
-      "eval_samples_per_second": 97.324,
-      "eval_steps_per_second": 6.084,
+      "eval_accuracy": 0.3924052283335005,
+      "eval_loss": 3.3151607513427734,
+      "eval_runtime": 184.9732,
+      "eval_samples_per_second": 97.371,
+      "eval_steps_per_second": 6.087,
       "step": 82000
     },
     {
       "epoch": 8.846361185983827,
-      "grad_norm": 0.872122049331665,
+      "grad_norm": 0.8377047777175903,
       "learning_rate": 6.950674581759309e-05,
-      "loss": 3.1817,
+      "loss": 3.1836,
       "step": 82050
     },
     {
       "epoch": 8.851752021563343,
-      "grad_norm": 0.8270566463470459,
+      "grad_norm": 0.8401516079902649,
       "learning_rate": 6.918294657312465e-05,
-      "loss": 3.1714,
+      "loss": 3.1738,
       "step": 82100
     },
     {
       "epoch": 8.857142857142858,
-      "grad_norm": 0.8941061496734619,
+      "grad_norm": 0.8585867285728455,
       "learning_rate": 6.885914732865623e-05,
-      "loss": 3.1884,
+      "loss": 3.1916,
       "step": 82150
     },
     {
       "epoch": 8.862533692722373,
-      "grad_norm": 0.8830512166023254,
+      "grad_norm": 0.8396210074424744,
       "learning_rate": 6.85353480841878e-05,
-      "loss": 3.1766,
+      "loss": 3.1801,
       "step": 82200
     },
     {
       "epoch": 8.867924528301886,
-      "grad_norm": 0.8618978261947632,
+      "grad_norm": 0.8280928730964661,
       "learning_rate": 6.821154883971937e-05,
-      "loss": 3.1935,
+      "loss": 3.1956,
       "step": 82250
     },
     {
       "epoch": 8.873315363881401,
-      "grad_norm": 0.8829900026321411,
+      "grad_norm": 0.906252920627594,
       "learning_rate": 6.788774959525094e-05,
-      "loss": 3.1805,
+      "loss": 3.1818,
       "step": 82300
     },
     {
       "epoch": 8.878706199460916,
-      "grad_norm": 0.8683678507804871,
+      "grad_norm": 0.8322795629501343,
       "learning_rate": 6.756395035078252e-05,
-      "loss": 3.1852,
+      "loss": 3.1908,
       "step": 82350
     },
     {
       "epoch": 8.884097035040432,
-      "grad_norm": 0.8829304575920105,
+      "grad_norm": 0.856133759021759,
       "learning_rate": 6.724015110631408e-05,
-      "loss": 3.1773,
+      "loss": 3.1786,
       "step": 82400
     },
     {
       "epoch": 8.889487870619947,
-      "grad_norm": 0.8580146431922913,
+      "grad_norm": 0.8618354201316833,
       "learning_rate": 6.691635186184566e-05,
-      "loss": 3.1882,
+      "loss": 3.1911,
       "step": 82450
     },
     {
       "epoch": 8.89487870619946,
-      "grad_norm": 0.8446676135063171,
+      "grad_norm": 0.8436751365661621,
       "learning_rate": 6.659255261737722e-05,
-      "loss": 3.1776,
+      "loss": 3.1788,
       "step": 82500
     },
     {
       "epoch": 8.900269541778975,
-      "grad_norm": 0.8247393369674683,
+      "grad_norm": 0.8414259552955627,
       "learning_rate": 6.626875337290879e-05,
-      "loss": 3.1661,
+      "loss": 3.1687,
       "step": 82550
     },
     {
       "epoch": 8.90566037735849,
-      "grad_norm": 0.8302431702613831,
+      "grad_norm": 0.8588948249816895,
       "learning_rate": 6.594495412844036e-05,
-      "loss": 3.1837,
+      "loss": 3.1871,
       "step": 82600
     },
     {
       "epoch": 8.911051212938006,
-      "grad_norm": 0.8455914258956909,
+      "grad_norm": 0.8397443890571594,
       "learning_rate": 6.562115488397193e-05,
-      "loss": 3.1756,
+      "loss": 3.181,
       "step": 82650
     },
     {
       "epoch": 8.91644204851752,
-      "grad_norm": 0.875441312789917,
+      "grad_norm": 0.8405500650405884,
       "learning_rate": 6.52973556395035e-05,
-      "loss": 3.1938,
+      "loss": 3.1951,
       "step": 82700
     },
     {
       "epoch": 8.921832884097036,
-      "grad_norm": 0.8480143547058105,
+      "grad_norm": 0.8125476241111755,
       "learning_rate": 6.497355639503507e-05,
-      "loss": 3.1679,
+      "loss": 3.1696,
       "step": 82750
     },
     {
       "epoch": 8.92722371967655,
-      "grad_norm": 0.8535221815109253,
+      "grad_norm": 0.8378741145133972,
       "learning_rate": 6.464975715056663e-05,
-      "loss": 3.1713,
+      "loss": 3.1755,
       "step": 82800
     },
     {
       "epoch": 8.932614555256064,
-      "grad_norm": 0.8422010540962219,
+      "grad_norm": 0.8491596579551697,
       "learning_rate": 6.432595790609821e-05,
-      "loss": 3.1738,
+      "loss": 3.1784,
       "step": 82850
     },
     {
       "epoch": 8.93800539083558,
-      "grad_norm": 0.8463007211685181,
+      "grad_norm": 0.832361102104187,
       "learning_rate": 6.400215866162979e-05,
-      "loss": 3.1733,
+      "loss": 3.175,
       "step": 82900
     },
     {
       "epoch": 8.943396226415095,
-      "grad_norm": 0.8810790777206421,
+      "grad_norm": 0.8865790963172913,
       "learning_rate": 6.367835941716135e-05,
-      "loss": 3.1733,
+      "loss": 3.1758,
       "step": 82950
     },
     {
       "epoch": 8.94878706199461,
-      "grad_norm": 0.8690757751464844,
+      "grad_norm": 0.8258707523345947,
       "learning_rate": 6.335456017269293e-05,
-      "loss": 3.1691,
+      "loss": 3.1729,
       "step": 83000
     },
     {
       "epoch": 8.94878706199461,
-      "eval_accuracy": 0.3927895335977003,
-      "eval_loss": 3.312455415725708,
-      "eval_runtime": 184.9722,
-      "eval_samples_per_second": 97.371,
-      "eval_steps_per_second": 6.087,
+      "eval_accuracy": 0.3925592981284214,
+      "eval_loss": 3.314058303833008,
+      "eval_runtime": 185.21,
+      "eval_samples_per_second": 97.246,
+      "eval_steps_per_second": 6.08,
       "step": 83000
     },
     {
       "epoch": 8.954177897574123,
-      "grad_norm": 0.8817049264907837,
+      "grad_norm": 0.8663061857223511,
       "learning_rate": 6.30307609282245e-05,
-      "loss": 3.1657,
+      "loss": 3.1686,
       "step": 83050
     },
     {
       "epoch": 8.959568733153638,
-      "grad_norm": 0.8801669478416443,
+      "grad_norm": 0.8376678824424744,
       "learning_rate": 6.270696168375607e-05,
-      "loss": 3.1741,
+      "loss": 3.1761,
       "step": 83100
     },
     {
       "epoch": 8.964959568733153,
-      "grad_norm": 0.888874351978302,
+      "grad_norm": 0.8633261322975159,
       "learning_rate": 6.238316243928764e-05,
-      "loss": 3.2002,
+      "loss": 3.2033,
       "step": 83150
     },
     {
       "epoch": 8.970350404312669,
-      "grad_norm": 0.8430107235908508,
+      "grad_norm": 0.9000486731529236,
       "learning_rate": 6.205936319481921e-05,
-      "loss": 3.1797,
+      "loss": 3.1812,
       "step": 83200
     },
     {
       "epoch": 8.975741239892184,
-      "grad_norm": 0.8379260301589966,
+      "grad_norm": 0.8094392418861389,
       "learning_rate": 6.173556395035078e-05,
-      "loss": 3.1659,
+      "loss": 3.1674,
       "step": 83250
     },
     {
       "epoch": 8.981132075471699,
-      "grad_norm": 0.8534399271011353,
-      "learning_rate": 6.141824069077172e-05,
-      "loss": 3.1762,
+      "grad_norm": 0.8410497903823853,
+      "learning_rate": 6.141176470588236e-05,
+      "loss": 3.1784,
       "step": 83300
     },
     {
       "epoch": 8.986522911051212,
-      "grad_norm": 0.8653827905654907,
-      "learning_rate": 6.109444144630328e-05,
-      "loss": 3.1804,
+      "grad_norm": 0.8640261292457581,
+      "learning_rate": 6.108796546141392e-05,
+      "loss": 3.1831,
       "step": 83350
     },
     {
       "epoch": 8.991913746630727,
-      "grad_norm": 0.8666049242019653,
-      "learning_rate": 6.077064220183486e-05,
-      "loss": 3.1691,
+      "grad_norm": 0.8512471318244934,
+      "learning_rate": 6.076416621694549e-05,
+      "loss": 3.1697,
       "step": 83400
     },
     {
       "epoch": 8.997304582210242,
-      "grad_norm": 0.9309250712394714,
-      "learning_rate": 6.044684295736643e-05,
-      "loss": 3.1926,
+      "grad_norm": 0.9322320818901062,
+      "learning_rate": 6.0440366972477055e-05,
+      "loss": 3.1964,
       "step": 83450
     },
     {
       "epoch": 9.002695417789758,
-      "grad_norm": 0.8734882473945618,
-      "learning_rate": 6.0123043712898e-05,
-      "loss": 3.1226,
+      "grad_norm": 0.8278214335441589,
+      "learning_rate": 6.011656772800863e-05,
+      "loss": 3.1252,
       "step": 83500
     },
     {
       "epoch": 9.008086253369273,
-      "grad_norm": 0.8495007157325745,
-      "learning_rate": 5.979924446842957e-05,
-      "loss": 3.1172,
+      "grad_norm": 0.791012167930603,
+      "learning_rate": 5.9792768483540197e-05,
+      "loss": 3.1193,
       "step": 83550
     },
     {
       "epoch": 9.013477088948788,
-      "grad_norm": 0.8585401177406311,
-      "learning_rate": 5.9475445223961135e-05,
-      "loss": 3.1339,
+      "grad_norm": 0.8340629935264587,
+      "learning_rate": 5.946896923907177e-05,
+      "loss": 3.1389,
       "step": 83600
     },
     {
       "epoch": 9.018867924528301,
-      "grad_norm": 0.8981791734695435,
-      "learning_rate": 5.915164597949271e-05,
-      "loss": 3.1283,
+      "grad_norm": 0.8602026104927063,
+      "learning_rate": 5.9145169994603345e-05,
+      "loss": 3.1343,
       "step": 83650
     },
     {
       "epoch": 9.024258760107816,
-      "grad_norm": 0.8400082588195801,
-      "learning_rate": 5.882784673502428e-05,
-      "loss": 3.113,
+      "grad_norm": 0.8203555345535278,
+      "learning_rate": 5.882137075013491e-05,
+      "loss": 3.116,
       "step": 83700
     },
     {
       "epoch": 9.029649595687331,
-      "grad_norm": 0.8486177921295166,
-      "learning_rate": 5.8504047490555855e-05,
-      "loss": 3.1105,
+      "grad_norm": 0.8314825296401978,
+      "learning_rate": 5.849757150566649e-05,
+      "loss": 3.1127,
       "step": 83750
     },
     {
       "epoch": 9.035040431266847,
-      "grad_norm": 0.8828721642494202,
-      "learning_rate": 5.818024824608742e-05,
-      "loss": 3.114,
+      "grad_norm": 0.8461922407150269,
+      "learning_rate": 5.817377226119805e-05,
+      "loss": 3.1168,
       "step": 83800
     },
     {
       "epoch": 9.040431266846362,
-      "grad_norm": 0.8872936964035034,
-      "learning_rate": 5.7856449001618996e-05,
-      "loss": 3.1002,
+      "grad_norm": 0.8548511266708374,
+      "learning_rate": 5.784997301672963e-05,
+      "loss": 3.1039,
       "step": 83850
     },
     {
       "epoch": 9.045822102425875,
-      "grad_norm": 0.8819625377655029,
-      "learning_rate": 5.753264975715056e-05,
-      "loss": 3.1178,
+      "grad_norm": 0.8729279041290283,
+      "learning_rate": 5.752617377226119e-05,
+      "loss": 3.1212,
       "step": 83900
     },
     {
       "epoch": 9.05121293800539,
-      "grad_norm": 0.8307718634605408,
-      "learning_rate": 5.720885051268213e-05,
-      "loss": 3.1203,
+      "grad_norm": 0.84293532371521,
+      "learning_rate": 5.7202374527792764e-05,
+      "loss": 3.123,
       "step": 83950
     },
     {
       "epoch": 9.056603773584905,
-      "grad_norm": 0.8690371513366699,
-      "learning_rate": 5.68850512682137e-05,
-      "loss": 3.104,
+      "grad_norm": 0.8201664686203003,
+      "learning_rate": 5.6878575283324335e-05,
+      "loss": 3.1054,
       "step": 84000
     },
     {
       "epoch": 9.056603773584905,
-      "eval_accuracy": 0.3928119160925901,
-      "eval_loss": 3.3145651817321777,
-      "eval_runtime": 185.1326,
-      "eval_samples_per_second": 97.287,
-      "eval_steps_per_second": 6.082,
+      "eval_accuracy": 0.3924269589110634,
+      "eval_loss": 3.3185200691223145,
+      "eval_runtime": 185.2088,
+      "eval_samples_per_second": 97.247,
+      "eval_steps_per_second": 6.08,
       "step": 84000
     },
     {
       "epoch": 9.06199460916442,
-      "grad_norm": 0.8724967241287231,
-      "learning_rate": 5.656125202374527e-05,
-      "loss": 3.1179,
+      "grad_norm": 0.8599904775619507,
+      "learning_rate": 5.6554776038855905e-05,
+      "loss": 3.1192,
       "step": 84050
     },
     {
       "epoch": 9.067385444743936,
-      "grad_norm": 0.8930646777153015,
-      "learning_rate": 5.623745277927684e-05,
-      "loss": 3.1429,
+      "grad_norm": 0.9081140756607056,
+      "learning_rate": 5.623097679438747e-05,
+      "loss": 3.1465,
       "step": 84100
     },
     {
       "epoch": 9.07277628032345,
-      "grad_norm": 0.8653122782707214,
-      "learning_rate": 5.5913653534808415e-05,
-      "loss": 3.1217,
+      "grad_norm": 0.8966930508613586,
+      "learning_rate": 5.590717754991905e-05,
+      "loss": 3.1251,
       "step": 84150
     },
     {
       "epoch": 9.078167115902964,
-      "grad_norm": 0.8568156957626343,
-      "learning_rate": 5.558985429033998e-05,
-      "loss": 3.1379,
+      "grad_norm": 0.8360949754714966,
+      "learning_rate": 5.558337830545061e-05,
+      "loss": 3.1401,
       "step": 84200
     },
     {
       "epoch": 9.08355795148248,
-      "grad_norm": 0.8342962265014648,
-      "learning_rate": 5.526605504587156e-05,
-      "loss": 3.1325,
+      "grad_norm": 0.8535497784614563,
+      "learning_rate": 5.525957906098219e-05,
+      "loss": 3.1357,
       "step": 84250
     },
     {
       "epoch": 9.088948787061994,
-      "grad_norm": 0.8761492371559143,
-      "learning_rate": 5.494225580140313e-05,
-      "loss": 3.124,
+      "grad_norm": 0.8542355298995972,
+      "learning_rate": 5.493577981651375e-05,
+      "loss": 3.1276,
       "step": 84300
     },
     {
       "epoch": 9.09433962264151,
-      "grad_norm": 0.8720303773880005,
-      "learning_rate": 5.461845655693469e-05,
-      "loss": 3.1166,
+      "grad_norm": 0.8455662727355957,
+      "learning_rate": 5.4611980572045324e-05,
+      "loss": 3.1184,
       "step": 84350
     },
     {
       "epoch": 9.099730458221025,
-      "grad_norm": 0.9339317083358765,
-      "learning_rate": 5.429465731246627e-05,
-      "loss": 3.1074,
+      "grad_norm": 0.9406881928443909,
+      "learning_rate": 5.4288181327576895e-05,
+      "loss": 3.109,
       "step": 84400
     },
     {
       "epoch": 9.10512129380054,
-      "grad_norm": 0.877099871635437,
+      "grad_norm": 0.8666425943374634,
       "learning_rate": 5.3970858067997833e-05,
-      "loss": 3.1322,
+      "loss": 3.1374,
       "step": 84450
     },
     {
       "epoch": 9.110512129380053,
-      "grad_norm": 0.8813628554344177,
-      "learning_rate": 5.364705882352941e-05,
-      "loss": 3.1202,
+      "grad_norm": 0.859734296798706,
+      "learning_rate": 5.365353480841878e-05,
+      "loss": 3.1224,
       "step": 84500
     },
     {
       "epoch": 9.115902964959568,
-      "grad_norm": 0.8760769963264465,
-      "learning_rate": 5.3323259579060975e-05,
-      "loss": 3.1273,
+      "grad_norm": 0.8375653624534607,
+      "learning_rate": 5.332973556395034e-05,
+      "loss": 3.1319,
       "step": 84550
     },
     {
       "epoch": 9.121293800539084,
-      "grad_norm": 0.8799816370010376,
-      "learning_rate": 5.299946033459255e-05,
-      "loss": 3.1158,
+      "grad_norm": 0.8139840960502625,
+      "learning_rate": 5.300593631948192e-05,
+      "loss": 3.119,
       "step": 84600
     },
     {
       "epoch": 9.126684636118599,
-      "grad_norm": 0.8845181465148926,
-      "learning_rate": 5.267566109012412e-05,
-      "loss": 3.1049,
+      "grad_norm": 0.8541390895843506,
+      "learning_rate": 5.2682137075013485e-05,
+      "loss": 3.1071,
       "step": 84650
     },
     {
       "epoch": 9.132075471698114,
-      "grad_norm": 0.8808998465538025,
-      "learning_rate": 5.235186184565569e-05,
-      "loss": 3.1181,
+      "grad_norm": 0.8539107441902161,
+      "learning_rate": 5.2358337830545056e-05,
+      "loss": 3.1217,
       "step": 84700
     },
     {
       "epoch": 9.137466307277627,
-      "grad_norm": 0.9242299795150757,
-      "learning_rate": 5.202806260118726e-05,
-      "loss": 3.121,
+      "grad_norm": 0.900639533996582,
+      "learning_rate": 5.2034538586076626e-05,
+      "loss": 3.123,
       "step": 84750
     },
     {
       "epoch": 9.142857142857142,
-      "grad_norm": 0.9068589210510254,
-      "learning_rate": 5.170426335671883e-05,
-      "loss": 3.1207,
+      "grad_norm": 0.8555846214294434,
+      "learning_rate": 5.17107393416082e-05,
+      "loss": 3.1225,
       "step": 84800
     },
     {
       "epoch": 9.148247978436657,
-      "grad_norm": 0.8964105248451233,
-      "learning_rate": 5.1380464112250394e-05,
-      "loss": 3.1314,
+      "grad_norm": 0.8566033840179443,
+      "learning_rate": 5.1386940097139775e-05,
+      "loss": 3.1336,
       "step": 84850
     },
     {
       "epoch": 9.153638814016173,
-      "grad_norm": 0.8806259036064148,
-      "learning_rate": 5.105666486778197e-05,
-      "loss": 3.1387,
+      "grad_norm": 0.8521566987037659,
+      "learning_rate": 5.106314085267134e-05,
+      "loss": 3.1411,
       "step": 84900
     },
     {
       "epoch": 9.159029649595688,
-      "grad_norm": 0.8335862755775452,
-      "learning_rate": 5.0732865623313536e-05,
-      "loss": 3.1257,
+      "grad_norm": 0.8052482604980469,
+      "learning_rate": 5.073934160820291e-05,
+      "loss": 3.1304,
       "step": 84950
     },
     {
       "epoch": 9.164420485175203,
-      "grad_norm": 0.9166033864021301,
-      "learning_rate": 5.040906637884511e-05,
-      "loss": 3.1398,
+      "grad_norm": 0.873296856880188,
+      "learning_rate": 5.041554236373448e-05,
+      "loss": 3.1447,
       "step": 85000
     },
     {
       "epoch": 9.164420485175203,
-      "eval_accuracy": 0.3929139411542479,
-      "eval_loss": 3.3140039443969727,
-      "eval_runtime": 185.025,
-      "eval_samples_per_second": 97.344,
-      "eval_steps_per_second": 6.086,
+      "eval_accuracy": 0.39260960441547954,
+      "eval_loss": 3.3156075477600098,
+      "eval_runtime": 185.3438,
+      "eval_samples_per_second": 97.176,
+      "eval_steps_per_second": 6.075,
       "step": 85000
     },
     {
       "epoch": 9.169811320754716,
-      "grad_norm": 0.8852344751358032,
-      "learning_rate": 5.0085267134376684e-05,
-      "loss": 3.1307,
+      "grad_norm": 0.8564801216125488,
+      "learning_rate": 5.009174311926605e-05,
+      "loss": 3.1333,
       "step": 85050
     },
     {
       "epoch": 9.175202156334231,
-      "grad_norm": 0.8456209301948547,
+      "grad_norm": 0.8553125858306885,
       "learning_rate": 4.976794387479762e-05,
-      "loss": 3.1277,
+      "loss": 3.133,
       "step": 85100
     },
     {
       "epoch": 9.180592991913747,
-      "grad_norm": 0.8467517495155334,
+      "grad_norm": 0.8583039045333862,
       "learning_rate": 4.9444144630329194e-05,
-      "loss": 3.137,
+      "loss": 3.1389,
       "step": 85150
     },
     {
       "epoch": 9.185983827493262,
-      "grad_norm": 0.905830442905426,
+      "grad_norm": 0.9124622941017151,
       "learning_rate": 4.912034538586076e-05,
-      "loss": 3.1382,
+      "loss": 3.1409,
       "step": 85200
     },
     {
       "epoch": 9.191374663072777,
-      "grad_norm": 0.9719045162200928,
+      "grad_norm": 0.8893402218818665,
       "learning_rate": 4.8796546141392335e-05,
-      "loss": 3.1128,
+      "loss": 3.1163,
       "step": 85250
     },
     {
       "epoch": 9.19676549865229,
-      "grad_norm": 0.9098581075668335,
+      "grad_norm": 0.8861328959465027,
       "learning_rate": 4.84727468969239e-05,
-      "loss": 3.1232,
+      "loss": 3.1273,
       "step": 85300
     },
     {
       "epoch": 9.202156334231805,
-      "grad_norm": 0.8981218338012695,
+      "grad_norm": 0.8711820244789124,
       "learning_rate": 4.814894765245548e-05,
-      "loss": 3.1183,
+      "loss": 3.1204,
       "step": 85350
     },
     {
       "epoch": 9.20754716981132,
-      "grad_norm": 0.8904754519462585,
+      "grad_norm": 0.880292534828186,
       "learning_rate": 4.782514840798704e-05,
-      "loss": 3.1292,
+      "loss": 3.1304,
       "step": 85400
     },
     {
       "epoch": 9.212938005390836,
-      "grad_norm": 0.88540118932724,
+      "grad_norm": 0.8324595093727112,
       "learning_rate": 4.750134916351861e-05,
-      "loss": 3.1313,
+      "loss": 3.1335,
       "step": 85450
     },
     {
       "epoch": 9.21832884097035,
-      "grad_norm": 0.8722923994064331,
+      "grad_norm": 0.8473580479621887,
       "learning_rate": 4.717754991905018e-05,
-      "loss": 3.1197,
+      "loss": 3.1238,
       "step": 85500
     },
     {
       "epoch": 9.223719676549866,
-      "grad_norm": 0.913180410861969,
+      "grad_norm": 0.9067453742027283,
       "learning_rate": 4.6853750674581754e-05,
-      "loss": 3.107,
+      "loss": 3.1113,
       "step": 85550
     },
     {
       "epoch": 9.22911051212938,
-      "grad_norm": 0.8536023497581482,
+      "grad_norm": 0.8589571118354797,
       "learning_rate": 4.652995143011333e-05,
-      "loss": 3.115,
+      "loss": 3.1175,
       "step": 85600
     },
     {
       "epoch": 9.234501347708894,
-      "grad_norm": 0.8618268966674805,
+      "grad_norm": 0.8338851928710938,
       "learning_rate": 4.6206152185644896e-05,
-      "loss": 3.1302,
+      "loss": 3.133,
       "step": 85650
     },
     {
       "epoch": 9.23989218328841,
-      "grad_norm": 0.8930416703224182,
+      "grad_norm": 0.8637981414794922,
       "learning_rate": 4.588235294117647e-05,
-      "loss": 3.1218,
+      "loss": 3.1245,
       "step": 85700
     },
     {
       "epoch": 9.245283018867925,
-      "grad_norm": 0.9080697894096375,
+      "grad_norm": 0.8614128828048706,
       "learning_rate": 4.555855369670804e-05,
-      "loss": 3.1122,
+      "loss": 3.1155,
       "step": 85750
     },
     {
       "epoch": 9.25067385444744,
-      "grad_norm": 0.8830873370170593,
+      "grad_norm": 0.8531436920166016,
       "learning_rate": 4.523475445223961e-05,
-      "loss": 3.122,
+      "loss": 3.1243,
       "step": 85800
     },
     {
       "epoch": 9.256064690026955,
-      "grad_norm": 0.8966952562332153,
+      "grad_norm": 0.8632187247276306,
       "learning_rate": 4.491095520777118e-05,
-      "loss": 3.1461,
+      "loss": 3.1491,
       "step": 85850
     },
     {
       "epoch": 9.261455525606468,
-      "grad_norm": 0.8461965918540955,
+      "grad_norm": 0.8344792723655701,
       "learning_rate": 4.458715596330275e-05,
-      "loss": 3.1312,
+      "loss": 3.1338,
       "step": 85900
     },
     {
       "epoch": 9.266846361185983,
-      "grad_norm": 0.9272150993347168,
+      "grad_norm": 0.8807640075683594,
       "learning_rate": 4.4263356718834314e-05,
-      "loss": 3.1333,
+      "loss": 3.1349,
       "step": 85950
     },
     {
       "epoch": 9.272237196765499,
-      "grad_norm": 0.8832882642745972,
+      "grad_norm": 0.8676424026489258,
       "learning_rate": 4.393955747436589e-05,
-      "loss": 3.1275,
+      "loss": 3.1303,
       "step": 86000
     },
     {
       "epoch": 9.272237196765499,
-      "eval_accuracy": 0.3932997675588771,
-      "eval_loss": 3.311936378479004,
-      "eval_runtime": 184.859,
-      "eval_samples_per_second": 97.431,
-      "eval_steps_per_second": 6.091,
+      "eval_accuracy": 0.393040956380103,
+      "eval_loss": 3.313966989517212,
+      "eval_runtime": 184.9242,
+      "eval_samples_per_second": 97.397,
+      "eval_steps_per_second": 6.089,
       "step": 86000
     },
     {
       "epoch": 9.277628032345014,
-      "grad_norm": 0.8568077683448792,
+      "grad_norm": 0.8324336409568787,
       "learning_rate": 4.3615758229897456e-05,
-      "loss": 3.1382,
+      "loss": 3.1421,
       "step": 86050
     },
     {
       "epoch": 9.283018867924529,
-      "grad_norm": 0.8889806270599365,
+      "grad_norm": 0.8373544216156006,
       "learning_rate": 4.3291958985429034e-05,
-      "loss": 3.1295,
+      "loss": 3.1336,
       "step": 86100
     },
     {
       "epoch": 9.288409703504042,
-      "grad_norm": 0.8781276345252991,
+      "grad_norm": 0.8576855063438416,
       "learning_rate": 4.29681597409606e-05,
-      "loss": 3.1426,
+      "loss": 3.1479,
       "step": 86150
     },
     {
       "epoch": 9.293800539083557,
-      "grad_norm": 0.9067704081535339,
+      "grad_norm": 0.863328754901886,
       "learning_rate": 4.264436049649217e-05,
-      "loss": 3.1376,
+      "loss": 3.1387,
       "step": 86200
     },
     {
       "epoch": 9.299191374663073,
-      "grad_norm": 0.8865089416503906,
+      "grad_norm": 0.8970577716827393,
       "learning_rate": 4.232056125202374e-05,
-      "loss": 3.1408,
+      "loss": 3.1417,
       "step": 86250
     },
     {
       "epoch": 9.304582210242588,
-      "grad_norm": 0.8952741622924805,
+      "grad_norm": 0.9056400060653687,
       "learning_rate": 4.199676200755531e-05,
-      "loss": 3.1384,
+      "loss": 3.1423,
       "step": 86300
     },
     {
       "epoch": 9.309973045822103,
-      "grad_norm": 0.870446503162384,
+      "grad_norm": 0.8215680718421936,
       "learning_rate": 4.167296276308688e-05,
-      "loss": 3.1127,
+      "loss": 3.1151,
       "step": 86350
     },
     {
       "epoch": 9.315363881401618,
-      "grad_norm": 0.8651818633079529,
+      "grad_norm": 0.8935737013816833,
       "learning_rate": 4.134916351861845e-05,
-      "loss": 3.1194,
+      "loss": 3.125,
       "step": 86400
     },
     {
       "epoch": 9.320754716981131,
-      "grad_norm": 0.8986634612083435,
+      "grad_norm": 0.8761557936668396,
       "learning_rate": 4.102536427415003e-05,
-      "loss": 3.1328,
+      "loss": 3.1351,
       "step": 86450
     },
     {
       "epoch": 9.326145552560646,
-      "grad_norm": 0.9104035496711731,
+      "grad_norm": 0.8842337131500244,
       "learning_rate": 4.0701565029681594e-05,
-      "loss": 3.1283,
+      "loss": 3.1319,
       "step": 86500
     },
     {
       "epoch": 9.331536388140162,
-      "grad_norm": 0.901136577129364,
+      "grad_norm": 0.8900733590126038,
       "learning_rate": 4.0377765785213165e-05,
-      "loss": 3.1363,
+      "loss": 3.1398,
       "step": 86550
     },
     {
       "epoch": 9.336927223719677,
-      "grad_norm": 0.9057474732398987,
+      "grad_norm": 0.8664814829826355,
       "learning_rate": 4.0053966540744736e-05,
-      "loss": 3.1485,
+      "loss": 3.1515,
       "step": 86600
     },
     {
       "epoch": 9.342318059299192,
-      "grad_norm": 0.8955982327461243,
+      "grad_norm": 0.888770580291748,
       "learning_rate": 3.973016729627631e-05,
-      "loss": 3.1168,
+      "loss": 3.1203,
       "step": 86650
     },
     {
       "epoch": 9.347708894878707,
-      "grad_norm": 0.8255138397216797,
+      "grad_norm": 0.82951819896698,
       "learning_rate": 3.940636805180787e-05,
-      "loss": 3.1294,
+      "loss": 3.1318,
       "step": 86700
     },
     {
       "epoch": 9.35309973045822,
-      "grad_norm": 0.8693887591362,
+      "grad_norm": 0.866402804851532,
       "learning_rate": 3.908256880733945e-05,
-      "loss": 3.1106,
+      "loss": 3.1143,
       "step": 86750
     },
     {
       "epoch": 9.358490566037736,
-      "grad_norm": 0.8875924944877625,
+      "grad_norm": 0.9193182587623596,
       "learning_rate": 3.875876956287101e-05,
-      "loss": 3.113,
+      "loss": 3.1167,
       "step": 86800
     },
     {
       "epoch": 9.36388140161725,
-      "grad_norm": 0.8646242618560791,
+      "grad_norm": 0.8478342890739441,
       "learning_rate": 3.843497031840259e-05,
-      "loss": 3.131,
+      "loss": 3.1339,
       "step": 86850
     },
     {
       "epoch": 9.369272237196766,
-      "grad_norm": 0.9060271978378296,
+      "grad_norm": 0.8515084981918335,
       "learning_rate": 3.8111171073934154e-05,
-      "loss": 3.1357,
+      "loss": 3.1382,
       "step": 86900
     },
     {
       "epoch": 9.374663072776281,
-      "grad_norm": 0.8785396814346313,
+      "grad_norm": 0.9116027355194092,
       "learning_rate": 3.778737182946573e-05,
-      "loss": 3.1382,
+      "loss": 3.1403,
       "step": 86950
     },
     {
       "epoch": 9.380053908355794,
-      "grad_norm": 0.8670048713684082,
+      "grad_norm": 0.8636597990989685,
       "learning_rate": 3.7463572584997296e-05,
-      "loss": 3.1152,
+      "loss": 3.1186,
       "step": 87000
     },
     {
       "epoch": 9.380053908355794,
-      "eval_accuracy": 0.39350446959951957,
-      "eval_loss": 3.3097434043884277,
-      "eval_runtime": 185.0264,
-      "eval_samples_per_second": 97.343,
-      "eval_steps_per_second": 6.086,
+      "eval_accuracy": 0.3931999155549756,
+      "eval_loss": 3.3115437030792236,
+      "eval_runtime": 185.4557,
+      "eval_samples_per_second": 97.118,
+      "eval_steps_per_second": 6.072,
       "step": 87000
     },
     {
       "epoch": 9.38544474393531,
-      "grad_norm": 0.8645483255386353,
+      "grad_norm": 0.8413622379302979,
       "learning_rate": 3.713977334052887e-05,
-      "loss": 3.1364,
+      "loss": 3.1368,
       "step": 87050
     },
     {
       "epoch": 9.390835579514825,
-      "grad_norm": 0.9023386836051941,
+      "grad_norm": 0.8973217010498047,
       "learning_rate": 3.681597409606044e-05,
-      "loss": 3.1231,
+      "loss": 3.1272,
       "step": 87100
     },
     {
       "epoch": 9.39622641509434,
-      "grad_norm": 0.8484783172607422,
+      "grad_norm": 0.8484401106834412,
       "learning_rate": 3.649217485159201e-05,
-      "loss": 3.1285,
+      "loss": 3.1329,
       "step": 87150
     },
     {
       "epoch": 9.401617250673855,
-      "grad_norm": 0.8609200119972229,
+      "grad_norm": 0.8557623624801636,
       "learning_rate": 3.616837560712358e-05,
-      "loss": 3.1254,
+      "loss": 3.1275,
       "step": 87200
     },
     {
       "epoch": 9.40700808625337,
-      "grad_norm": 0.8669583201408386,
+      "grad_norm": 0.8505693674087524,
       "learning_rate": 3.584457636265515e-05,
-      "loss": 3.1401,
+      "loss": 3.1427,
       "step": 87250
     },
     {
       "epoch": 9.412398921832883,
-      "grad_norm": 0.9098264575004578,
+      "grad_norm": 0.8859766125679016,
       "learning_rate": 3.552077711818672e-05,
-      "loss": 3.1465,
+      "loss": 3.1487,
       "step": 87300
     },
     {
       "epoch": 9.417789757412399,
-      "grad_norm": 0.8986435532569885,
+      "grad_norm": 0.8826354742050171,
       "learning_rate": 3.519697787371829e-05,
-      "loss": 3.1234,
+      "loss": 3.1261,
       "step": 87350
     },
     {
       "epoch": 9.423180592991914,
-      "grad_norm": 0.8918881416320801,
+      "grad_norm": 0.8677520751953125,
       "learning_rate": 3.487317862924986e-05,
-      "loss": 3.1235,
+      "loss": 3.1261,
       "step": 87400
     },
     {
       "epoch": 9.428571428571429,
-      "grad_norm": 0.9000228047370911,
+      "grad_norm": 0.8759662508964539,
       "learning_rate": 3.4549379384781434e-05,
-      "loss": 3.13,
+      "loss": 3.1328,
       "step": 87450
     },
     {
       "epoch": 9.433962264150944,
-      "grad_norm": 0.9257108569145203,
+      "grad_norm": 0.8896969556808472,
       "learning_rate": 3.4225580140313e-05,
-      "loss": 3.1329,
+      "loss": 3.1362,
       "step": 87500
     },
     {
       "epoch": 9.439353099730457,
-      "grad_norm": 0.8658718466758728,
+      "grad_norm": 0.8729029297828674,
       "learning_rate": 3.390178089584457e-05,
-      "loss": 3.1421,
+      "loss": 3.1455,
       "step": 87550
     },
     {
       "epoch": 9.444743935309972,
-      "grad_norm": 0.895495593547821,
+      "grad_norm": 0.8900150656700134,
       "learning_rate": 3.357798165137615e-05,
-      "loss": 3.128,
+      "loss": 3.1304,
       "step": 87600
     },
     {
       "epoch": 9.450134770889488,
-      "grad_norm": 0.882296085357666,
+      "grad_norm": 0.8727821707725525,
       "learning_rate": 3.325418240690772e-05,
-      "loss": 3.1428,
+      "loss": 3.1481,
       "step": 87650
     },
     {
       "epoch": 9.455525606469003,
-      "grad_norm": 0.8577868342399597,
+      "grad_norm": 0.8477314710617065,
       "learning_rate": 3.293038316243929e-05,
-      "loss": 3.1046,
+      "loss": 3.1063,
       "step": 87700
     },
     {
       "epoch": 9.460916442048518,
-      "grad_norm": 0.9138762950897217,
+      "grad_norm": 0.8587415218353271,
       "learning_rate": 3.260658391797086e-05,
-      "loss": 3.1224,
+      "loss": 3.1258,
       "step": 87750
     },
     {
       "epoch": 9.466307277628033,
-      "grad_norm": 0.8505160212516785,
+      "grad_norm": 0.8558546900749207,
       "learning_rate": 3.2282784673502424e-05,
-      "loss": 3.1177,
+      "loss": 3.1225,
       "step": 87800
     },
     {
       "epoch": 9.471698113207546,
-      "grad_norm": 0.8743757009506226,
+      "grad_norm": 0.8536050319671631,
       "learning_rate": 3.1958985429033995e-05,
-      "loss": 3.13,
+      "loss": 3.1322,
       "step": 87850
     },
     {
       "epoch": 9.477088948787062,
-      "grad_norm": 0.8973845839500427,
+      "grad_norm": 0.8786517977714539,
       "learning_rate": 3.1635186184565565e-05,
-      "loss": 3.1383,
+      "loss": 3.1458,
       "step": 87900
     },
     {
       "epoch": 9.482479784366577,
-      "grad_norm": 0.8859951496124268,
+      "grad_norm": 0.8906846642494202,
       "learning_rate": 3.1311386940097136e-05,
-      "loss": 3.1428,
+      "loss": 3.1442,
       "step": 87950
     },
     {
       "epoch": 9.487870619946092,
-      "grad_norm": 0.9165722131729126,
+      "grad_norm": 0.8581869006156921,
       "learning_rate": 3.098758769562871e-05,
-      "loss": 3.1329,
+      "loss": 3.1368,
       "step": 88000
     },
     {
       "epoch": 9.487870619946092,
-      "eval_accuracy": 0.39391898036653183,
-      "eval_loss": 3.3068549633026123,
-      "eval_runtime": 184.8627,
-      "eval_samples_per_second": 97.429,
-      "eval_steps_per_second": 6.091,
+      "eval_accuracy": 0.3937146042845528,
+      "eval_loss": 3.3079490661621094,
+      "eval_runtime": 184.9387,
+      "eval_samples_per_second": 97.389,
+      "eval_steps_per_second": 6.089,
       "step": 88000
     },
     {
       "epoch": 9.493261455525607,
-      "grad_norm": 0.8812258243560791,
+      "grad_norm": 0.8685297966003418,
       "learning_rate": 3.066378845116028e-05,
-      "loss": 3.1294,
+      "loss": 3.134,
       "step": 88050
     },
     {
       "epoch": 9.498652291105122,
-      "grad_norm": 0.8953335881233215,
+      "grad_norm": 0.8372690081596375,
       "learning_rate": 3.033998920669185e-05,
-      "loss": 3.1284,
+      "loss": 3.1328,
       "step": 88100
     },
     {
       "epoch": 9.504043126684635,
-      "grad_norm": 0.8637675642967224,
+      "grad_norm": 0.8183308839797974,
       "learning_rate": 3.0016189962223416e-05,
-      "loss": 3.1266,
+      "loss": 3.1306,
       "step": 88150
     },
     {
       "epoch": 9.50943396226415,
-      "grad_norm": 0.8599790930747986,
+      "grad_norm": 1.0200047492980957,
       "learning_rate": 2.9692390717754987e-05,
-      "loss": 3.1237,
+      "loss": 3.1259,
       "step": 88200
     },
     {
       "epoch": 9.514824797843666,
-      "grad_norm": 0.8580920100212097,
+      "grad_norm": 0.8445395827293396,
       "learning_rate": 2.9368591473286558e-05,
-      "loss": 3.1317,
+      "loss": 3.1334,
       "step": 88250
     },
     {
       "epoch": 9.520215633423181,
-      "grad_norm": 0.8671429753303528,
+      "grad_norm": 0.8794388175010681,
       "learning_rate": 2.904479222881813e-05,
-      "loss": 3.1385,
+      "loss": 3.1418,
       "step": 88300
     },
     {
       "epoch": 9.525606469002696,
-      "grad_norm": 0.8459764122962952,
+      "grad_norm": 0.7987924814224243,
       "learning_rate": 2.87209929843497e-05,
-      "loss": 3.1398,
+      "loss": 3.1419,
       "step": 88350
     },
     {
       "epoch": 9.530997304582211,
-      "grad_norm": 0.8779290914535522,
+      "grad_norm": 0.8669410943984985,
       "learning_rate": 2.8397193739881274e-05,
-      "loss": 3.138,
+      "loss": 3.1422,
       "step": 88400
     },
     {
       "epoch": 9.536388140161725,
-      "grad_norm": 0.8906304240226746,
+      "grad_norm": 0.8429611921310425,
       "learning_rate": 2.8073394495412842e-05,
-      "loss": 3.1318,
+      "loss": 3.1346,
       "step": 88450
     },
     {
       "epoch": 9.54177897574124,
-      "grad_norm": 0.91546231508255,
+      "grad_norm": 0.8398469686508179,
       "learning_rate": 2.7749595250944413e-05,
-      "loss": 3.1251,
+      "loss": 3.1279,
       "step": 88500
     },
     {
       "epoch": 9.547169811320755,
-      "grad_norm": 0.8772543668746948,
+      "grad_norm": 0.8982305526733398,
       "learning_rate": 2.7425796006475984e-05,
       "loss": 3.1203,
       "step": 88550
     },
     {
       "epoch": 9.55256064690027,
-      "grad_norm": 0.9106935262680054,
+      "grad_norm": 0.8920227289199829,
       "learning_rate": 2.7101996762007554e-05,
-      "loss": 3.1328,
+      "loss": 3.1362,
       "step": 88600
     },
     {
       "epoch": 9.557951482479785,
-      "grad_norm": 0.864047110080719,
+      "grad_norm": 0.8532522320747375,
       "learning_rate": 2.6778197517539125e-05,
-      "loss": 3.1012,
+      "loss": 3.1053,
       "step": 88650
     },
     {
       "epoch": 9.563342318059298,
-      "grad_norm": 0.8812419176101685,
+      "grad_norm": 0.8370513319969177,
       "learning_rate": 2.6454398273070693e-05,
-      "loss": 3.1276,
+      "loss": 3.1305,
       "step": 88700
     },
     {
       "epoch": 9.568733153638814,
-      "grad_norm": 0.8692460060119629,
+      "grad_norm": 0.8290605545043945,
       "learning_rate": 2.6130599028602264e-05,
-      "loss": 3.1308,
+      "loss": 3.1349,
       "step": 88750
     },
     {
       "epoch": 9.574123989218329,
-      "grad_norm": 0.9016609191894531,
+      "grad_norm": 0.8801090121269226,
       "learning_rate": 2.5806799784133835e-05,
-      "loss": 3.1092,
+      "loss": 3.113,
       "step": 88800
     },
     {
       "epoch": 9.579514824797844,
-      "grad_norm": 0.875944972038269,
+      "grad_norm": 0.8506044149398804,
       "learning_rate": 2.5483000539665406e-05,
-      "loss": 3.1279,
+      "loss": 3.13,
       "step": 88850
     },
     {
       "epoch": 9.584905660377359,
-      "grad_norm": 0.8492311835289001,
+      "grad_norm": 0.8414507508277893,
       "learning_rate": 2.5159201295196976e-05,
-      "loss": 3.1498,
+      "loss": 3.1535,
       "step": 88900
     },
     {
       "epoch": 9.590296495956874,
-      "grad_norm": 0.879350483417511,
+      "grad_norm": 0.8439317345619202,
       "learning_rate": 2.4835402050728544e-05,
-      "loss": 3.1354,
+      "loss": 3.1387,
       "step": 88950
     },
     {
       "epoch": 9.595687331536388,
-      "grad_norm": 0.8495625853538513,
+      "grad_norm": 0.8427550196647644,
       "learning_rate": 2.4511602806260115e-05,
-      "loss": 3.1273,
+      "loss": 3.13,
       "step": 89000
     },
     {
       "epoch": 9.595687331536388,
-      "eval_accuracy": 0.3940988008958648,
-      "eval_loss": 3.304088592529297,
-      "eval_runtime": 184.961,
-      "eval_samples_per_second": 97.377,
-      "eval_steps_per_second": 6.088,
+      "eval_accuracy": 0.3939106140941701,
+      "eval_loss": 3.3059844970703125,
+      "eval_runtime": 185.1018,
+      "eval_samples_per_second": 97.303,
+      "eval_steps_per_second": 6.083,
       "step": 89000
     },
     {
       "epoch": 9.601078167115903,
-      "grad_norm": 0.8095118403434753,
+      "grad_norm": 0.7962179780006409,
       "learning_rate": 2.4187803561791686e-05,
-      "loss": 3.1248,
+      "loss": 3.1265,
       "step": 89050
     },
     {
       "epoch": 9.606469002695418,
-      "grad_norm": 0.8577584028244019,
-      "learning_rate": 2.3870480302212628e-05,
-      "loss": 3.1295,
+      "grad_norm": 0.834709644317627,
+      "learning_rate": 2.3864004317323257e-05,
+      "loss": 3.1327,
       "step": 89100
     },
     {
       "epoch": 9.611859838274933,
-      "grad_norm": 0.8957887887954712,
-      "learning_rate": 2.35466810577442e-05,
-      "loss": 3.1508,
+      "grad_norm": 0.883066713809967,
+      "learning_rate": 2.354020507285483e-05,
+      "loss": 3.156,
       "step": 89150
     },
     {
       "epoch": 9.617250673854448,
-      "grad_norm": 0.9367175102233887,
-      "learning_rate": 2.3222881813275766e-05,
-      "loss": 3.1187,
+      "grad_norm": 0.932812511920929,
+      "learning_rate": 2.3216405828386402e-05,
+      "loss": 3.1206,
       "step": 89200
     },
     {
       "epoch": 9.622641509433961,
-      "grad_norm": 0.9250417351722717,
-      "learning_rate": 2.2899082568807337e-05,
-      "loss": 3.1394,
+      "grad_norm": 0.9128279089927673,
+      "learning_rate": 2.289260658391797e-05,
+      "loss": 3.1426,
       "step": 89250
     },
     {
       "epoch": 9.628032345013477,
-      "grad_norm": 0.8945813775062561,
-      "learning_rate": 2.2575283324338908e-05,
-      "loss": 3.1292,
+      "grad_norm": 0.8921266198158264,
+      "learning_rate": 2.256880733944954e-05,
+      "loss": 3.1294,
       "step": 89300
     },
     {
       "epoch": 9.633423180592992,
-      "grad_norm": 0.8463922142982483,
-      "learning_rate": 2.225148407987048e-05,
-      "loss": 3.1253,
+      "grad_norm": 0.8413317799568176,
+      "learning_rate": 2.224500809498111e-05,
+      "loss": 3.1273,
       "step": 89350
     },
     {
       "epoch": 9.638814016172507,
-      "grad_norm": 0.9221121072769165,
-      "learning_rate": 2.192768483540205e-05,
-      "loss": 3.1276,
+      "grad_norm": 0.8817403316497803,
+      "learning_rate": 2.1921208850512682e-05,
+      "loss": 3.1315,
       "step": 89400
     },
     {
       "epoch": 9.644204851752022,
-      "grad_norm": 0.882972240447998,
-      "learning_rate": 2.1603885590933617e-05,
-      "loss": 3.1281,
+      "grad_norm": 0.8911518454551697,
+      "learning_rate": 2.1597409606044253e-05,
+      "loss": 3.1327,
       "step": 89450
     },
     {
       "epoch": 9.649595687331537,
-      "grad_norm": 0.8939113616943359,
-      "learning_rate": 2.1280086346465188e-05,
-      "loss": 3.1432,
+      "grad_norm": 0.8857812285423279,
+      "learning_rate": 2.127361036157582e-05,
+      "loss": 3.1459,
       "step": 89500
     },
     {
       "epoch": 9.65498652291105,
-      "grad_norm": 0.896306037902832,
-      "learning_rate": 2.095628710199676e-05,
-      "loss": 3.1255,
+      "grad_norm": 0.8955504298210144,
+      "learning_rate": 2.094981111710739e-05,
+      "loss": 3.1307,
       "step": 89550
     },
     {
       "epoch": 9.660377358490566,
-      "grad_norm": 0.8766800165176392,
-      "learning_rate": 2.063248785752833e-05,
-      "loss": 3.1184,
+      "grad_norm": 0.86283940076828,
+      "learning_rate": 2.0626011872638962e-05,
+      "loss": 3.1206,
       "step": 89600
     },
     {
       "epoch": 9.66576819407008,
-      "grad_norm": 0.8930668830871582,
-      "learning_rate": 2.03086886130599e-05,
-      "loss": 3.1188,
+      "grad_norm": 0.8852249979972839,
+      "learning_rate": 2.0302212628170533e-05,
+      "loss": 3.1227,
       "step": 89650
     },
     {
       "epoch": 9.671159029649596,
-      "grad_norm": 0.9026961326599121,
-      "learning_rate": 1.9984889368591468e-05,
-      "loss": 3.1218,
+      "grad_norm": 0.878625214099884,
+      "learning_rate": 1.99784133837021e-05,
+      "loss": 3.1262,
       "step": 89700
     },
     {
       "epoch": 9.676549865229111,
-      "grad_norm": 0.9059770107269287,
-      "learning_rate": 1.9661090124123042e-05,
-      "loss": 3.1375,
+      "grad_norm": 0.850557267665863,
+      "learning_rate": 1.965461413923367e-05,
+      "loss": 3.1391,
       "step": 89750
     },
     {
       "epoch": 9.681940700808624,
-      "grad_norm": 0.8790046572685242,
-      "learning_rate": 1.9337290879654613e-05,
-      "loss": 3.1196,
+      "grad_norm": 0.8788577318191528,
+      "learning_rate": 1.9330814894765242e-05,
+      "loss": 3.1231,
       "step": 89800
     },
     {
       "epoch": 9.68733153638814,
-      "grad_norm": 0.8520891070365906,
-      "learning_rate": 1.9013491635186184e-05,
-      "loss": 3.1237,
+      "grad_norm": 0.8096227049827576,
+      "learning_rate": 1.9007015650296813e-05,
+      "loss": 3.1254,
       "step": 89850
     },
     {
       "epoch": 9.692722371967655,
-      "grad_norm": 0.8755563497543335,
-      "learning_rate": 1.868969239071775e-05,
-      "loss": 3.1437,
+      "grad_norm": 0.8480028510093689,
+      "learning_rate": 1.8683216405828384e-05,
+      "loss": 3.1454,
       "step": 89900
     },
     {
       "epoch": 9.69811320754717,
-      "grad_norm": 0.8892934322357178,
-      "learning_rate": 1.8365893146249326e-05,
-      "loss": 3.1458,
+      "grad_norm": 0.8541979193687439,
+      "learning_rate": 1.8359417161359955e-05,
+      "loss": 3.1467,
       "step": 89950
     },
     {
       "epoch": 9.703504043126685,
-      "grad_norm": 0.8864810466766357,
-      "learning_rate": 1.8042093901780893e-05,
-      "loss": 3.129,
+      "grad_norm": 0.8675863742828369,
+      "learning_rate": 1.8035617916891526e-05,
+      "loss": 3.1323,
       "step": 90000
     },
     {
       "epoch": 9.703504043126685,
-      "eval_accuracy": 0.39435913321506827,
-      "eval_loss": 3.302138566970825,
-      "eval_runtime": 184.8974,
-      "eval_samples_per_second": 97.411,
-      "eval_steps_per_second": 6.09,
+      "eval_accuracy": 0.39413606883638513,
+      "eval_loss": 3.304030179977417,
+      "eval_runtime": 185.3134,
+      "eval_samples_per_second": 97.192,
+      "eval_steps_per_second": 6.076,
       "step": 90000
     }
   ],