diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,10021 @@
+{
+  "best_metric": 0.6475752433152033,
+  "best_model_checkpoint": "./runtime-masked/MiniLMv2-L6-H384-distilled-from-RoBERTa-Large-finetuned-wikitext103-mlm-multi-emails-hq-x2bs/checkpoint-4004",
+  "epoch": 16.0,
+  "global_step": 4928,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4291497975708505e-06,
+      "loss": 7.2679,
+      "step": 3
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.858299595141701e-06,
+      "loss": 7.1451,
+      "step": 6
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.287449392712551e-06,
+      "loss": 7.1002,
+      "step": 9
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.716599190283402e-06,
+      "loss": 6.9508,
+      "step": 12
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2145748987854251e-05,
+      "loss": 6.7874,
+      "step": 15
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4574898785425101e-05,
+      "loss": 6.6279,
+      "step": 18
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7004048582995952e-05,
+      "loss": 6.5254,
+      "step": 21
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9433198380566804e-05,
+      "loss": 6.3327,
+      "step": 24
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1862348178137653e-05,
+      "loss": 6.2761,
+      "step": 27
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4291497975708502e-05,
+      "loss": 6.0968,
+      "step": 30
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6720647773279357e-05,
+      "loss": 5.9384,
+      "step": 33
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9149797570850203e-05,
+      "loss": 5.8496,
+      "step": 36
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.157894736842105e-05,
+      "loss": 5.7136,
+      "step": 39
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.4008097165991904e-05,
+      "loss": 5.6149,
+      "step": 42
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.6437246963562756e-05,
+      "loss": 5.5674,
+      "step": 45
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.886639676113361e-05,
+      "loss": 5.489,
+      "step": 48
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1295546558704454e-05,
+      "loss": 5.3851,
+      "step": 51
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.3724696356275306e-05,
+      "loss": 5.3135,
+      "step": 54
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.615384615384616e-05,
+      "loss": 5.1979,
+      "step": 57
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8582995951417004e-05,
+      "loss": 5.0876,
+      "step": 60
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.101214574898786e-05,
+      "loss": 5.094,
+      "step": 63
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.3441295546558715e-05,
+      "loss": 5.0148,
+      "step": 66
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.587044534412956e-05,
+      "loss": 4.9376,
+      "step": 69
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.8299595141700406e-05,
+      "loss": 4.9033,
+      "step": 72
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.072874493927125e-05,
+      "loss": 4.8783,
+      "step": 75
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.31578947368421e-05,
+      "loss": 4.8382,
+      "step": 78
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.558704453441296e-05,
+      "loss": 4.7009,
+      "step": 81
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.801619433198381e-05,
+      "loss": 4.6597,
+      "step": 84
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.044534412955465e-05,
+      "loss": 4.5674,
+      "step": 87
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.287449392712551e-05,
+      "loss": 4.5938,
+      "step": 90
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.530364372469636e-05,
+      "loss": 4.6061,
+      "step": 93
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.773279352226722e-05,
+      "loss": 4.582,
+      "step": 96
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.016194331983806e-05,
+      "loss": 4.4975,
+      "step": 99
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.259109311740891e-05,
+      "loss": 4.3876,
+      "step": 102
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.502024291497977e-05,
+      "loss": 4.388,
+      "step": 105
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.744939271255061e-05,
+      "loss": 4.2698,
+      "step": 108
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.987854251012147e-05,
+      "loss": 4.3306,
+      "step": 111
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.230769230769232e-05,
+      "loss": 4.3391,
+      "step": 114
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.473684210526316e-05,
+      "loss": 4.1981,
+      "step": 117
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.716599190283401e-05,
+      "loss": 4.3052,
+      "step": 120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.959514170040485e-05,
+      "loss": 4.1384,
+      "step": 123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00010202429149797573,
+      "loss": 4.159,
+      "step": 126
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00010445344129554657,
+      "loss": 4.1178,
+      "step": 129
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00010688259109311743,
+      "loss": 4.0929,
+      "step": 132
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00010931174089068827,
+      "loss": 4.0761,
+      "step": 135
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011174089068825912,
+      "loss": 3.9875,
+      "step": 138
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011417004048582995,
+      "loss": 4.0039,
+      "step": 141
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011659919028340081,
+      "loss": 3.9917,
+      "step": 144
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011902834008097166,
+      "loss": 4.0101,
+      "step": 147
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001214574898785425,
+      "loss": 3.9108,
+      "step": 150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00012388663967611335,
+      "loss": 3.9445,
+      "step": 153
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001263157894736842,
+      "loss": 3.96,
+      "step": 156
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00012874493927125507,
+      "loss": 3.9475,
+      "step": 159
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00013117408906882592,
+      "loss": 3.8582,
+      "step": 162
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00013360323886639676,
+      "loss": 3.8952,
+      "step": 165
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00013603238866396762,
+      "loss": 3.7632,
+      "step": 168
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00013846153846153847,
+      "loss": 3.7845,
+      "step": 171
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001408906882591093,
+      "loss": 3.7638,
+      "step": 174
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00014331983805668017,
+      "loss": 3.8404,
+      "step": 177
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00014574898785425102,
+      "loss": 3.7742,
+      "step": 180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014817813765182186,
+      "loss": 3.7533,
+      "step": 183
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015060728744939272,
+      "loss": 3.7303,
+      "step": 186
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015303643724696357,
+      "loss": 3.7195,
+      "step": 189
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015546558704453443,
+      "loss": 3.7544,
+      "step": 192
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015789473684210527,
+      "loss": 3.6913,
+      "step": 195
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00016032388663967612,
+      "loss": 3.7917,
+      "step": 198
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016275303643724698,
+      "loss": 3.6758,
+      "step": 201
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016518218623481781,
+      "loss": 3.6774,
+      "step": 204
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00016761133603238867,
+      "loss": 3.6199,
+      "step": 207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017004048582995953,
+      "loss": 3.6028,
+      "step": 210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001724696356275304,
+      "loss": 3.6084,
+      "step": 213
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017489878542510122,
+      "loss": 3.6165,
+      "step": 216
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017732793522267208,
+      "loss": 3.5123,
+      "step": 219
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017975708502024294,
+      "loss": 3.5594,
+      "step": 222
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018218623481781377,
+      "loss": 3.6238,
+      "step": 225
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018461538461538463,
+      "loss": 3.4991,
+      "step": 228
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001870445344129555,
+      "loss": 3.5384,
+      "step": 231
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018947368421052632,
+      "loss": 3.5282,
+      "step": 234
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019190283400809716,
+      "loss": 3.574,
+      "step": 237
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019433198380566801,
+      "loss": 3.5391,
+      "step": 240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019676113360323887,
+      "loss": 3.4529,
+      "step": 243
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001991902834008097,
+      "loss": 3.4957,
+      "step": 246
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019999990991501854,
+      "loss": 3.4346,
+      "step": 249
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019999943696930958,
+      "loss": 3.4838,
+      "step": 252
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019999855864354245,
+      "loss": 3.4721,
+      "step": 255
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001999972749412778,
+      "loss": 3.5,
+      "step": 258
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00019999558586771948,
+      "loss": 3.3624,
+      "step": 261
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00019999349142971467,
+      "loss": 3.4138,
+      "step": 264
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00019999099163575389,
+      "loss": 3.4005,
+      "step": 267
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00019998808649597085,
+      "loss": 3.365,
+      "step": 270
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001999847760221425,
+      "loss": 3.3424,
+      "step": 273
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00019998106022768887,
+      "loss": 3.3629,
+      "step": 276
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00019997693912767318,
+      "loss": 3.3722,
+      "step": 279
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00019997241273880158,
+      "loss": 3.3951,
+      "step": 282
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00019996748107942335,
+      "loss": 3.3817,
+      "step": 285
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00019996214416953046,
+      "loss": 3.3289,
+      "step": 288
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00019995640203075788,
+      "loss": 3.3074,
+      "step": 291
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00019995025468638318,
+      "loss": 3.3145,
+      "step": 294
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00019994370216132662,
+      "loss": 3.2853,
+      "step": 297
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000199936744482151,
+      "loss": 3.2416,
+      "step": 300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001999293816770615,
+      "loss": 3.2565,
+      "step": 303
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00019992161377590563,
+      "loss": 3.2947,
+      "step": 306
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.5121698756686252,
+      "eval_loss": 3.0832247734069824,
+      "eval_runtime": 16.2528,
+      "eval_samples_per_second": 135.177,
+      "eval_steps_per_second": 67.619,
+      "step": 308
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001999134408101731,
+      "loss": 3.2464,
+      "step": 309
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00019990486281299568,
+      "loss": 3.2509,
+      "step": 312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00019989587981914704,
+      "loss": 3.284,
+      "step": 315
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00019988649186504262,
+      "loss": 3.1894,
+      "step": 318
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001998766989887396,
+      "loss": 3.3045,
+      "step": 321
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001998665012299365,
+      "loss": 3.2935,
+      "step": 324
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001998558986299733,
+      "loss": 3.1808,
+      "step": 327
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000199844891231831,
+      "loss": 3.1295,
+      "step": 330
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00019983347908013172,
+      "loss": 3.262,
+      "step": 333
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00019982166222113826,
+      "loss": 3.1685,
+      "step": 336
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00019980944070275406,
+      "loss": 3.1682,
+      "step": 339
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00019979681457452304,
+      "loss": 3.2196,
+      "step": 342
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001997837838876293,
+      "loss": 3.1117,
+      "step": 345
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001997703486948969,
+      "loss": 3.2077,
+      "step": 348
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00019975650905078976,
+      "loss": 3.1355,
+      "step": 351
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00019974226501141137,
+      "loss": 3.2277,
+      "step": 354
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00019972761663450452,
+      "loss": 3.14,
+      "step": 357
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001997125639794512,
+      "loss": 3.1937,
+      "step": 360
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00019969710710727214,
+      "loss": 3.1441,
+      "step": 363
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00019968124608062682,
+      "loss": 3.1193,
+      "step": 366
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.000199664980963813,
+      "loss": 3.1849,
+      "step": 369
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00019964831182276663,
+      "loss": 3.1189,
+      "step": 372
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019963123872506147,
+      "loss": 3.0733,
+      "step": 375
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001996137617399088,
+      "loss": 3.101,
+      "step": 378
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00019959588093815728,
+      "loss": 3.074,
+      "step": 381
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019957759639229247,
+      "loss": 3.1142,
+      "step": 384
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019955890817643674,
+      "loss": 3.1246,
+      "step": 387
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001995398163663488,
+      "loss": 3.104,
+      "step": 390
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00019952032103942347,
+      "loss": 3.1105,
+      "step": 393
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001995004222746913,
+      "loss": 3.0985,
+      "step": 396
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00019948012015281853,
+      "loss": 3.1341,
+      "step": 399
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00019945941475610623,
+      "loss": 3.0335,
+      "step": 402
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001994383061684905,
+      "loss": 3.0978,
+      "step": 405
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00019941679447554175,
+      "loss": 3.0737,
+      "step": 408
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00019939487976446468,
+      "loss": 3.0942,
+      "step": 411
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00019937256212409756,
+      "loss": 3.0471,
+      "step": 414
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00019934984164491227,
+      "loss": 3.0165,
+      "step": 417
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00019932671841901354,
+      "loss": 2.9886,
+      "step": 420
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00019930319254013887,
+      "loss": 3.0629,
+      "step": 423
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000199279264103658,
+      "loss": 3.0518,
+      "step": 426
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00019925493320657262,
+      "loss": 2.9858,
+      "step": 429
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00019923019994751585,
+      "loss": 3.0696,
+      "step": 432
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.000199205064426752,
+      "loss": 3.1251,
+      "step": 435
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000199179526746176,
+      "loss": 3.0311,
+      "step": 438
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019915358700931313,
+      "loss": 2.9571,
+      "step": 441
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019912724532131847,
+      "loss": 2.9914,
+      "step": 444
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00019910050178897657,
+      "loss": 2.9803,
+      "step": 447
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00019907335652070103,
+      "loss": 3.0183,
+      "step": 450
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001990458096265339,
+      "loss": 3.0207,
+      "step": 453
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00019901786121814547,
+      "loss": 2.9883,
+      "step": 456
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00019898951140883369,
+      "loss": 2.924,
+      "step": 459
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001989607603135236,
+      "loss": 3.0618,
+      "step": 462
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00019893160804876708,
+      "loss": 3.0179,
+      "step": 465
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00019890205473274236,
+      "loss": 2.9295,
+      "step": 468
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00019887210048525323,
+      "loss": 2.9724,
+      "step": 471
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00019884174542772899,
+      "loss": 2.9413,
+      "step": 474
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019881098968322367,
+      "loss": 3.0484,
+      "step": 477
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019877983337641565,
+      "loss": 2.9098,
+      "step": 480
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019874827663360706,
+      "loss": 2.9568,
+      "step": 483
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019871631958272336,
+      "loss": 2.9348,
+      "step": 486
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00019868396235331282,
+      "loss": 2.9615,
+      "step": 489
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00019865120507654593,
+      "loss": 2.9036,
+      "step": 492
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00019861804788521493,
+      "loss": 2.8977,
+      "step": 495
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00019858449091373313,
+      "loss": 3.0531,
+      "step": 498
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00019855053429813463,
+      "loss": 2.9548,
+      "step": 501
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00019851617817607354,
+      "loss": 2.9541,
+      "step": 504
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00019848142268682356,
+      "loss": 2.8871,
+      "step": 507
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00019844626797127724,
+      "loss": 2.8821,
+      "step": 510
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00019841071417194561,
+      "loss": 2.9179,
+      "step": 513
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00019837476143295748,
+      "loss": 2.9251,
+      "step": 516
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00019833840990005893,
+      "loss": 2.8764,
+      "step": 519
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00019830165972061265,
+      "loss": 2.8817,
+      "step": 522
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00019826451104359738,
+      "loss": 2.8707,
+      "step": 525
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00019822696401960727,
+      "loss": 2.8489,
+      "step": 528
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00019818901880085137,
+      "loss": 2.9285,
+      "step": 531
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00019815067554115282,
+      "loss": 2.8707,
+      "step": 534
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001981119343959485,
+      "loss": 2.8615,
+      "step": 537
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00019807279552228816,
+      "loss": 2.8273,
+      "step": 540
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00019803325907883385,
+      "loss": 2.9113,
+      "step": 543
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00019799332522585936,
+      "loss": 2.8561,
+      "step": 546
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00019795299412524945,
+      "loss": 2.8562,
+      "step": 549
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00019791226594049932,
+      "loss": 2.8861,
+      "step": 552
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00019787114083671375,
+      "loss": 2.8196,
+      "step": 555
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00019782961898060677,
+      "loss": 2.8737,
+      "step": 558
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00019778770054050058,
+      "loss": 2.9101,
+      "step": 561
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00019774538568632515,
+      "loss": 2.9066,
+      "step": 564
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00019770267458961741,
+      "loss": 2.8889,
+      "step": 567
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00019765956742352062,
+      "loss": 2.8761,
+      "step": 570
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00019761606436278362,
+      "loss": 2.8484,
+      "step": 573
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00019757216558376013,
+      "loss": 2.8575,
+      "step": 576
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00019752787126440803,
+      "loss": 2.785,
+      "step": 579
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001974831815842887,
+      "loss": 2.8186,
+      "step": 582
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00019743809672456618,
+      "loss": 2.8184,
+      "step": 585
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001973926168680066,
+      "loss": 2.8115,
+      "step": 588
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00019734674219897718,
+      "loss": 2.8627,
+      "step": 591
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00019730047290344578,
+      "loss": 2.8416,
+      "step": 594
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001972538091689799,
+      "loss": 2.8565,
+      "step": 597
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00019720675118474614,
+      "loss": 2.8373,
+      "step": 600
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00019715929914150923,
+      "loss": 2.8271,
+      "step": 603
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00019711145323163137,
+      "loss": 2.8548,
+      "step": 606
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00019706321364907142,
+      "loss": 2.8292,
+      "step": 609
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019701458058938418,
+      "loss": 2.9203,
+      "step": 612
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019696555424971943,
+      "loss": 2.8727,
+      "step": 615
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.5661561892883697,
+      "eval_loss": 2.672184705734253,
+      "eval_runtime": 16.2659,
+      "eval_samples_per_second": 135.068,
+      "eval_steps_per_second": 67.565,
+      "step": 616
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001969161348288213,
+      "loss": 2.7287,
+      "step": 618
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019686632252702743,
+      "loss": 2.7983,
+      "step": 621
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019681611754626807,
+      "loss": 2.7829,
+      "step": 624
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019676552009006534,
+      "loss": 2.8671,
+      "step": 627
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001967145303635324,
+      "loss": 2.7472,
+      "step": 630
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019666314857337262,
+      "loss": 2.7506,
+      "step": 633
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019661137492787867,
+      "loss": 2.7307,
+      "step": 636
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019655920963693174,
+      "loss": 2.7653,
+      "step": 639
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019650665291200082,
+      "loss": 2.8072,
+      "step": 642
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019645370496614145,
+      "loss": 2.781,
+      "step": 645
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019640036601399535,
+      "loss": 2.6695,
+      "step": 648
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00019634663627178918,
+      "loss": 2.7504,
+      "step": 651
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00019629251595733383,
+      "loss": 2.7793,
+      "step": 654
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00019623800529002347,
+      "loss": 2.7255,
+      "step": 657
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00019618310449083477,
+      "loss": 2.7955,
+      "step": 660
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00019612781378232583,
+      "loss": 2.6888,
+      "step": 663
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00019607213338863547,
+      "loss": 2.8287,
+      "step": 666
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001960160635354821,
+      "loss": 2.7925,
+      "step": 669
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00019595960445016307,
+      "loss": 2.8107,
+      "step": 672
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00019590275636155352,
+      "loss": 2.7144,
+      "step": 675
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00019584551950010555,
+      "loss": 2.8271,
+      "step": 678
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00019578789409784727,
+      "loss": 2.685,
+      "step": 681
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00019572988038838194,
+      "loss": 2.7504,
+      "step": 684
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00019567147860688686,
+      "loss": 2.7186,
+      "step": 687
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00019561268899011256,
+      "loss": 2.7287,
+      "step": 690
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00019555351177638172,
+      "loss": 2.7973,
+      "step": 693
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00019549394720558833,
+      "loss": 2.7732,
+      "step": 696
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00019543399551919668,
+      "loss": 2.7949,
+      "step": 699
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001953736569602403,
+      "loss": 2.893,
+      "step": 702
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00019531293177332102,
+      "loss": 2.7169,
+      "step": 705
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00019525182020460803,
+      "loss": 2.7442,
+      "step": 708
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001951903225018369,
+      "loss": 2.7373,
+      "step": 711
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001951284389143084,
+      "loss": 2.6366,
+      "step": 714
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00019506616969288768,
+      "loss": 2.7411,
+      "step": 717
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00019500351509000314,
+      "loss": 2.7378,
+      "step": 720
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00019494047535964553,
+      "loss": 2.6151,
+      "step": 723
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00019487705075736672,
+      "loss": 2.8224,
+      "step": 726
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00019481324154027894,
+      "loss": 2.8167,
+      "step": 729
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00019474904796705337,
+      "loss": 2.6986,
+      "step": 732
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001946844702979195,
+      "loss": 2.7249,
+      "step": 735
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00019461950879466383,
+      "loss": 2.6904,
+      "step": 738
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001945541637206287,
+      "loss": 2.7191,
+      "step": 741
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00019448843534071163,
+      "loss": 2.804,
+      "step": 744
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00019442232392136375,
+      "loss": 2.7587,
+      "step": 747
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00019435582973058915,
+      "loss": 2.6742,
+      "step": 750
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00019428895303794352,
+      "loss": 2.7017,
+      "step": 753
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00019422169411453317,
+      "loss": 2.7544,
+      "step": 756
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001941540532330139,
+      "loss": 2.7186,
+      "step": 759
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00019408603066758988,
+      "loss": 2.7649,
+      "step": 762
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00019401762669401257,
+      "loss": 2.8109,
+      "step": 765
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00019394884158957965,
+      "loss": 2.7248,
+      "step": 768
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00019387967563313377,
+      "loss": 2.6719,
+      "step": 771
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00019381012910506146,
+      "loss": 2.7268,
+      "step": 774
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00019374020228729206,
+      "loss": 2.7121,
+      "step": 777
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001936698954632966,
+      "loss": 2.6516,
+      "step": 780
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00019359920891808647,
+      "loss": 2.7795,
+      "step": 783
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00019352814293821248,
+      "loss": 2.7295,
+      "step": 786
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00019345669781176356,
+      "loss": 2.6901,
+      "step": 789
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00019338487382836565,
+      "loss": 2.7171,
+      "step": 792
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00019331267127918044,
+      "loss": 2.5934,
+      "step": 795
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00019324009045690438,
+      "loss": 2.677,
+      "step": 798
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00019316713165576726,
+      "loss": 2.7009,
+      "step": 801
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001930937951715312,
+      "loss": 2.7384,
+      "step": 804
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00019302008130148932,
+      "loss": 2.6525,
+      "step": 807
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00019294599034446467,
+      "loss": 2.626,
+      "step": 810
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00019287152260080888,
+      "loss": 2.6826,
+      "step": 813
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00019279667837240105,
+      "loss": 2.7388,
+      "step": 816
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00019272145796264648,
+      "loss": 2.7281,
+      "step": 819
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001926458616764754,
+      "loss": 2.5821,
+      "step": 822
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00019256988982034178,
+      "loss": 2.6877,
+      "step": 825
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00019249354270222218,
+      "loss": 2.6533,
+      "step": 828
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00019241682063161428,
+      "loss": 2.7017,
+      "step": 831
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00019233972391953584,
+      "loss": 2.5812,
+      "step": 834
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00019226225287852325,
+      "loss": 2.6331,
+      "step": 837
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001921844078226305,
+      "loss": 2.6415,
+      "step": 840
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001921061890674277,
+      "loss": 2.6382,
+      "step": 843
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001920275969299998,
+      "loss": 2.6692,
+      "step": 846
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00019194863172894552,
+      "loss": 2.6522,
+      "step": 849
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00019186929378437582,
+      "loss": 2.6311,
+      "step": 852
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00019178958341791268,
+      "loss": 2.6869,
+      "step": 855
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00019170950095268792,
+      "loss": 2.6057,
+      "step": 858
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00019162904671334163,
+      "loss": 2.6813,
+      "step": 861
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00019154822102602115,
+      "loss": 2.6263,
+      "step": 864
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0001914670242183795,
+      "loss": 2.6506,
+      "step": 867
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00019138545661957426,
+      "loss": 2.6605,
+      "step": 870
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00019130351856026597,
+      "loss": 2.6763,
+      "step": 873
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00019122121037261719,
+      "loss": 2.6219,
+      "step": 876
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00019113853239029064,
+      "loss": 2.6208,
+      "step": 879
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00019105548494844835,
+      "loss": 2.7369,
+      "step": 882
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00019097206838374997,
+      "loss": 2.5989,
+      "step": 885
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0001908882830343515,
+      "loss": 2.5859,
+      "step": 888
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00019080412923990395,
+      "loss": 2.6183,
+      "step": 891
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00019071960734155194,
+      "loss": 2.6804,
+      "step": 894
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00019063471768193235,
+      "loss": 2.6772,
+      "step": 897
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00019054946060517283,
+      "loss": 2.6334,
+      "step": 900
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00019046383645689055,
+      "loss": 2.7288,
+      "step": 903
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00019037784558419065,
+      "loss": 2.6486,
+      "step": 906
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00019029148833566497,
+      "loss": 2.6382,
+      "step": 909
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00019020476506139057,
+      "loss": 2.5683,
+      "step": 912
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00019011767611292819,
+      "loss": 2.6047,
+      "step": 915
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00019003022184332116,
+      "loss": 2.6616,
+      "step": 918
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001899424026070936,
+      "loss": 2.6049,
+      "step": 921
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00018985421876024916,
+      "loss": 2.6339,
+      "step": 924
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.5878054172915932,
+      "eval_loss": 2.479712724685669,
+      "eval_runtime": 16.3394,
+      "eval_samples_per_second": 134.46,
+      "eval_steps_per_second": 67.261,
+      "step": 924
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0001897656706602696,
+      "loss": 2.5743,
+      "step": 927
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0001896767586661133,
+      "loss": 2.5385,
+      "step": 930
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0001895874831382138,
+      "loss": 2.6556,
+      "step": 933
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00018949784443847824,
+      "loss": 2.5895,
+      "step": 936
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00018940784293028617,
+      "loss": 2.5747,
+      "step": 939
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00018931747897848778,
+      "loss": 2.6032,
+      "step": 942
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00018922675294940256,
+      "loss": 2.5687,
+      "step": 945
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00018913566521081777,
+      "loss": 2.5473,
+      "step": 948
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00018904421613198712,
+      "loss": 2.6586,
+      "step": 951
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00018895240608362895,
+      "loss": 2.6245,
+      "step": 954
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.000188860235437925,
+      "loss": 2.6062,
+      "step": 957
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00018876770456851877,
+      "loss": 2.4521,
+      "step": 960
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0001886748138505141,
+      "loss": 2.698,
+      "step": 963
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00018858156366047358,
+      "loss": 2.5416,
+      "step": 966
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00018848795437641697,
+      "loss": 2.6364,
+      "step": 969
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00018839398637781972,
+      "loss": 2.5949,
+      "step": 972
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00018829966004561163,
+      "loss": 2.5397,
+      "step": 975
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00018820497576217492,
+      "loss": 2.5792,
+      "step": 978
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00018810993391134295,
+      "loss": 2.5549,
+      "step": 981
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00018801453487839862,
+      "loss": 2.6141,
+      "step": 984
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00018791877905007277,
+      "loss": 2.6055,
+      "step": 987
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00018782266681454255,
+      "loss": 2.5834,
+      "step": 990
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00018772619856143009,
+      "loss": 2.6272,
+      "step": 993
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0001876293746818006,
+      "loss": 2.5862,
+      "step": 996
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.000187532195568161,
+      "loss": 2.6453,
+      "step": 999
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00018743466161445823,
+      "loss": 2.5199,
+      "step": 1002
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00018733677321607775,
+      "loss": 2.5887,
+      "step": 1005
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0001872385307698418,
+      "loss": 2.5769,
+      "step": 1008
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00018713993467400796,
+      "loss": 2.6303,
+      "step": 1011
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00018704098532826735,
+      "loss": 2.5144,
+      "step": 1014
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0001869416831337432,
+      "loss": 2.5568,
+      "step": 1017
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00018684202849298897,
+      "loss": 2.6413,
+      "step": 1020
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00018674202180998708,
+      "loss": 2.5877,
+      "step": 1023
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0001866416634901469,
+      "loss": 2.5414,
+      "step": 1026
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00018654095394030334,
+      "loss": 2.5394,
+      "step": 1029
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00018643989356871514,
+      "loss": 2.5929,
+      "step": 1032
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00018633848278506323,
+      "loss": 2.6068,
+      "step": 1035
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00018623672200044898,
+      "loss": 2.6195,
+      "step": 1038
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00018613461162739263,
+      "loss": 2.5121,
+      "step": 1041
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00018603215207983165,
+      "loss": 2.5959,
+      "step": 1044
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0001859293437731189,
+      "loss": 2.5925,
+      "step": 1047
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00018582618712402113,
+      "loss": 2.608,
+      "step": 1050
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00018572268255071718,
+      "loss": 2.6188,
+      "step": 1053
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0001856188304727963,
+      "loss": 2.5683,
+      "step": 1056
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00018551463131125649,
+      "loss": 2.5835,
+      "step": 1059
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00018541008548850273,
+      "loss": 2.5374,
+      "step": 1062
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0001853051934283453,
+      "loss": 2.5489,
+      "step": 1065
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00018519995555599817,
+      "loss": 2.4947,
+      "step": 1068
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.000185094372298077,
+      "loss": 2.4604,
+      "step": 1071
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00018498844408259773,
+      "loss": 2.5453,
+      "step": 1074
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00018488217133897462,
+      "loss": 2.5738,
+      "step": 1077
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00018477555449801863,
+      "loss": 2.5437,
+      "step": 1080
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00018466859399193555,
+      "loss": 2.46,
+      "step": 1083
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00018456129025432442,
+      "loss": 2.5457,
+      "step": 1086
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00018445364372017564,
+      "loss": 2.5188,
+      "step": 1089
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00018434565482586924,
+      "loss": 2.5652,
+      "step": 1092
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00018423732400917316,
+      "loss": 2.544,
+      "step": 1095
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00018412865170924135,
+      "loss": 2.6398,
+      "step": 1098
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00018401963836661218,
+      "loss": 2.6341,
+      "step": 1101
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00018391028442320644,
+      "loss": 2.5351,
+      "step": 1104
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0001838005903223257,
+      "loss": 2.5473,
+      "step": 1107
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00018369055650865052,
+      "loss": 2.5146,
+      "step": 1110
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00018358018342823855,
+      "loss": 2.5715,
+      "step": 1113
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0001834694715285227,
+      "loss": 2.5376,
+      "step": 1116
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00018335842125830954,
+      "loss": 2.6296,
+      "step": 1119
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00018324703306777718,
+      "loss": 2.4321,
+      "step": 1122
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00018313530740847375,
+      "loss": 2.5319,
+      "step": 1125
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0001830232447333153,
+      "loss": 2.5231,
+      "step": 1128
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00018291084549658412,
+      "loss": 2.5694,
+      "step": 1131
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00018279811015392685,
+      "loss": 2.499,
+      "step": 1134
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00018268503916235273,
+      "loss": 2.5255,
+      "step": 1137
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00018257163298023151,
+      "loss": 2.5671,
+      "step": 1140
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0001824578920672919,
+      "loss": 2.4801,
+      "step": 1143
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00018234381688461942,
+      "loss": 2.5006,
+      "step": 1146
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00018222940789465475,
+      "loss": 2.5033,
+      "step": 1149
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00018211466556119173,
+      "loss": 2.565,
+      "step": 1152
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0001819995903493755,
+      "loss": 2.5869,
+      "step": 1155
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00018188418272570061,
+      "loss": 2.422,
+      "step": 1158
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00018176844315800924,
+      "loss": 2.4295,
+      "step": 1161
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0001816523721154892,
+      "loss": 2.4252,
+      "step": 1164
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00018153597006867188,
+      "loss": 2.5694,
+      "step": 1167
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00018141923748943073,
+      "loss": 2.4952,
+      "step": 1170
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00018130217485097893,
+      "loss": 2.4748,
+      "step": 1173
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00018118478262786782,
+      "loss": 2.5343,
+      "step": 1176
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0001810670612959847,
+      "loss": 2.4971,
+      "step": 1179
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00018094901133255105,
+      "loss": 2.4903,
+      "step": 1182
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00018083063321612056,
+      "loss": 2.5106,
+      "step": 1185
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0001807119274265773,
+      "loss": 2.4929,
+      "step": 1188
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00018059289444513347,
+      "loss": 2.5104,
+      "step": 1191
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00018047353475432782,
+      "loss": 2.4528,
+      "step": 1194
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00018035384883802346,
+      "loss": 2.4571,
+      "step": 1197
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00018023383718140593,
+      "loss": 2.5601,
+      "step": 1200
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00018011350027098127,
+      "loss": 2.4553,
+      "step": 1203
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00017999283859457412,
+      "loss": 2.5075,
+      "step": 1206
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0001798718526413256,
+      "loss": 2.5526,
+      "step": 1209
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00017975054290169138,
+      "loss": 2.4881,
+      "step": 1212
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0001796289098674397,
+      "loss": 2.4973,
+      "step": 1215
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00017950695403164943,
+      "loss": 2.4831,
+      "step": 1218
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.000179384675888708,
+      "loss": 2.5187,
+      "step": 1221
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0001792620759343094,
+      "loss": 2.4949,
+      "step": 1224
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00017913915466545217,
+      "loss": 2.533,
+      "step": 1227
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00017901591258043747,
+      "loss": 2.5053,
+      "step": 1230
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.6025211491749728,
+      "eval_loss": 2.383329153060913,
+      "eval_runtime": 16.3176,
+      "eval_samples_per_second": 134.64,
+      "eval_steps_per_second": 67.351,
+      "step": 1232
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0001788923501788669,
+      "loss": 2.5555,
+      "step": 1233
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00017876846796164068,
+      "loss": 2.4955,
+      "step": 1236
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001786442664309554,
+      "loss": 2.5338,
+      "step": 1239
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0001785197460903021,
+      "loss": 2.4958,
+      "step": 1242
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0001783949074444643,
+      "loss": 2.4291,
+      "step": 1245
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00017826975099951583,
+      "loss": 2.4112,
+      "step": 1248
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0001781442772628188,
+      "loss": 2.4166,
+      "step": 1251
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017801848674302154,
+      "loss": 2.5571,
+      "step": 1254
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017789237995005668,
+      "loss": 2.4778,
+      "step": 1257
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0001777659573951388,
+      "loss": 2.5491,
+      "step": 1260
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00017763921959076273,
+      "loss": 2.4311,
+      "step": 1263
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00017751216705070105,
+      "loss": 2.4439,
+      "step": 1266
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00017738480029000234,
+      "loss": 2.4623,
+      "step": 1269
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.000177257119824989,
+      "loss": 2.5291,
+      "step": 1272
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00017712912617325502,
+      "loss": 2.4934,
+      "step": 1275
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0001770008198536641,
+      "loss": 2.5712,
+      "step": 1278
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0001768722013863474,
+      "loss": 2.5426,
+      "step": 1281
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00017674327129270148,
+      "loss": 2.4619,
+      "step": 1284
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00017661403009538616,
+      "loss": 2.484,
+      "step": 1287
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00017648447831832242,
+      "loss": 2.4566,
+      "step": 1290
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0001763546164866903,
+      "loss": 2.5267,
+      "step": 1293
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00017622444512692672,
+      "loss": 2.4614,
+      "step": 1296
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00017609396476672343,
+      "loss": 2.4796,
+      "step": 1299
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0001759631759350247,
+      "loss": 2.3971,
+      "step": 1302
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0001758320791620254,
+      "loss": 2.4879,
+      "step": 1305
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0001757006749791687,
+      "loss": 2.5324,
+      "step": 1308
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00017556896391914394,
+      "loss": 2.3853,
+      "step": 1311
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0001754369465158845,
+      "loss": 2.455,
+      "step": 1314
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0001753046233045656,
+      "loss": 2.4964,
+      "step": 1317
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0001751719948216022,
+      "loss": 2.4615,
+      "step": 1320
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00017503906160464672,
+      "loss": 2.4745,
+      "step": 1323
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00017490582419258697,
+      "loss": 2.4068,
+      "step": 1326
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00017477228312554388,
+      "loss": 2.4827,
+      "step": 1329
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00017463843894486937,
+      "loss": 2.4002,
+      "step": 1332
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00017450429219314408,
+      "loss": 2.3769,
+      "step": 1335
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00017436984341417532,
+      "loss": 2.4319,
+      "step": 1338
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00017423509315299458,
+      "loss": 2.4803,
+      "step": 1341
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00017410004195585573,
+      "loss": 2.4286,
+      "step": 1344
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00017396469037023242,
+      "loss": 2.4884,
+      "step": 1347
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00017382903894481611,
+      "loss": 2.4376,
+      "step": 1350
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00017369308822951367,
+      "loss": 2.4966,
+      "step": 1353
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00017355683877544532,
+      "loss": 2.5422,
+      "step": 1356
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00017342029113494233,
+      "loss": 2.4287,
+      "step": 1359
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00017328344586154467,
+      "loss": 2.445,
+      "step": 1362
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0001731463035099989,
+      "loss": 2.4555,
+      "step": 1365
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00017300886463625595,
+      "loss": 2.4885,
+      "step": 1368
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00017287112979746868,
+      "loss": 2.4137,
+      "step": 1371
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00017273309955198984,
+      "loss": 2.494,
+      "step": 1374
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0001725947744593697,
+      "loss": 2.4098,
+      "step": 1377
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0001724561550803537,
+      "loss": 2.4887,
+      "step": 1380
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00017231724197688033,
+      "loss": 2.4888,
+      "step": 1383
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0001721780357120788,
+      "loss": 2.5013,
+      "step": 1386
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00017203853685026675,
+      "loss": 2.422,
+      "step": 1389
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00017189874595694788,
+      "loss": 2.4682,
+      "step": 1392
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00017175866359880982,
+      "loss": 2.3751,
+      "step": 1395
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00017161829034372168,
+      "loss": 2.4773,
+      "step": 1398
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00017147762676073187,
+      "loss": 2.4837,
+      "step": 1401
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0001713366734200657,
+      "loss": 2.4487,
+      "step": 1404
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00017119543089312317,
+      "loss": 2.4272,
+      "step": 1407
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00017105389975247647,
+      "loss": 2.4699,
+      "step": 1410
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00017091208057186792,
+      "loss": 2.4837,
+      "step": 1413
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00017076997392620737,
+      "loss": 2.4351,
+      "step": 1416
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0001706275803915701,
+      "loss": 2.3902,
+      "step": 1419
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00017048490054519434,
+      "loss": 2.3355,
+      "step": 1422
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00017034193496547902,
+      "loss": 2.3968,
+      "step": 1425
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00017019868423198134,
+      "loss": 2.4197,
+      "step": 1428
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00017005514892541444,
+      "loss": 2.4192,
+      "step": 1431
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00016991132962764516,
+      "loss": 2.359,
+      "step": 1434
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00016976722692169148,
+      "loss": 2.4068,
+      "step": 1437
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00016962284139172037,
+      "loss": 2.4527,
+      "step": 1440
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00016947817362304525,
+      "loss": 2.4723,
+      "step": 1443
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00016933322420212372,
+      "loss": 2.4029,
+      "step": 1446
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00016918799371655512,
+      "loss": 2.4574,
+      "step": 1449
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00016904248275507818,
+      "loss": 2.4426,
+      "step": 1452
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00016889669190756868,
+      "loss": 2.4689,
+      "step": 1455
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00016875062176503693,
+      "loss": 2.407,
+      "step": 1458
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0001686042729196255,
+      "loss": 2.3877,
+      "step": 1461
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0001684576459646068,
+      "loss": 2.394,
+      "step": 1464
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00016831074149438056,
+      "loss": 2.4613,
+      "step": 1467
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00016816356010447163,
+      "loss": 2.3824,
+      "step": 1470
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0001680161023915273,
+      "loss": 2.5161,
+      "step": 1473
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00016786836895331514,
+      "loss": 2.3828,
+      "step": 1476
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00016772036038872039,
+      "loss": 2.4413,
+      "step": 1479
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0001675720772977437,
+      "loss": 2.3981,
+      "step": 1482
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00016742352028149843,
+      "loss": 2.4631,
+      "step": 1485
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0001672746899422086,
+      "loss": 2.4012,
+      "step": 1488
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.000167125586883206,
+      "loss": 2.4043,
+      "step": 1491
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00016697621170892824,
+      "loss": 2.3889,
+      "step": 1494
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00016682656502491576,
+      "loss": 2.4473,
+      "step": 1497
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0001666766474378099,
+      "loss": 2.5165,
+      "step": 1500
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00016652645955535006,
+      "loss": 2.5155,
+      "step": 1503
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00016637600198637133,
+      "loss": 2.3923,
+      "step": 1506
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00016622527534080218,
+      "loss": 2.4353,
+      "step": 1509
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00016607428022966182,
+      "loss": 2.3346,
+      "step": 1512
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00016592301726505771,
+      "loss": 2.3873,
+      "step": 1515
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00016577148706018328,
+      "loss": 2.4799,
+      "step": 1518
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00016561969022931515,
+      "loss": 2.4228,
+      "step": 1521
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0001654676273878109,
+      "loss": 2.4024,
+      "step": 1524
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00016531529915210643,
+      "loss": 2.386,
+      "step": 1527
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0001651627061397135,
+      "loss": 2.4365,
+      "step": 1530
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00016500984896921725,
+      "loss": 2.3997,
+      "step": 1533
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00016485672826027363,
+      "loss": 2.3659,
+      "step": 1536
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00016470334463360698,
+      "loss": 2.4531,
+      "step": 1539
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.6106430794745761,
+      "eval_loss": 2.3084843158721924,
+      "eval_runtime": 16.3383,
+      "eval_samples_per_second": 134.469,
+      "eval_steps_per_second": 67.265,
+      "step": 1540
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00016454969871100743,
+      "loss": 2.376,
+      "step": 1542
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0001643957911153284,
+      "loss": 2.2957,
+      "step": 1545
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00016424162247048412,
+      "loss": 2.3557,
+      "step": 1548
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00016408719340144705,
+      "loss": 2.3722,
+      "step": 1551
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00016393250453424534,
+      "loss": 2.3903,
+      "step": 1554
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0001637775564959604,
+      "loss": 2.3326,
+      "step": 1557
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00016362234991472416,
+      "loss": 2.3791,
+      "step": 1560
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00016346688541971668,
+      "loss": 2.4608,
+      "step": 1563
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00016331116364116363,
+      "loss": 2.3049,
+      "step": 1566
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00016315518521033354,
+      "loss": 2.437,
+      "step": 1569
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00016299895075953547,
+      "loss": 2.4088,
+      "step": 1572
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0001628424609221163,
+      "loss": 2.4097,
+      "step": 1575
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00016268571633245812,
+      "loss": 2.3635,
+      "step": 1578
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00016252871762597592,
+      "loss": 2.4373,
+      "step": 1581
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00016237146543911463,
+      "loss": 2.2713,
+      "step": 1584
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00016221396040934694,
+      "loss": 2.4049,
+      "step": 1587
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00016205620317517034,
+      "loss": 2.3796,
+      "step": 1590
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00016189819437610484,
+      "loss": 2.3642,
+      "step": 1593
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00016173993465269022,
+      "loss": 2.3668,
+      "step": 1596
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00016158142464648342,
+      "loss": 2.4196,
+      "step": 1599
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00016142266500005604,
+      "loss": 2.488,
+      "step": 1602
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00016126365635699166,
+      "loss": 2.3974,
+      "step": 1605
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00016110439936188318,
+      "loss": 2.4516,
+      "step": 1608
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00016094489466033043,
+      "loss": 2.3589,
+      "step": 1611
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0001607851428989372,
+      "loss": 2.4077,
+      "step": 1614
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00016062514472530898,
+      "loss": 2.3902,
+      "step": 1617
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0001604649007880501,
+      "loss": 2.3319,
+      "step": 1620
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00016030441173676117,
+      "loss": 2.3729,
+      "step": 1623
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00016014367822203646,
+      "loss": 2.3052,
+      "step": 1626
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0001599827008954613,
+      "loss": 2.3613,
+      "step": 1629
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0001598214804096093,
+      "loss": 2.4415,
+      "step": 1632
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00015966001741803983,
+      "loss": 2.3959,
+      "step": 1635
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0001594983125752954,
+      "loss": 2.3294,
+      "step": 1638
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0001593363665368988,
+      "loss": 2.3211,
+      "step": 1641
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00015917417995935077,
+      "loss": 2.3113,
+      "step": 1644
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00015901175350012698,
+      "loss": 2.3507,
+      "step": 1647
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00015884908781767565,
+      "loss": 2.3533,
+      "step": 1650
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00015868618357141472,
+      "loss": 2.4636,
+      "step": 1653
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00015852304142172923,
+      "loss": 2.4222,
+      "step": 1656
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00015835966202996867,
+      "loss": 2.4257,
+      "step": 1659
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00015819604605844418,
+      "loss": 2.3802,
+      "step": 1662
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00015803219417042608,
+      "loss": 2.2824,
+      "step": 1665
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00015786810703014096,
+      "loss": 2.4089,
+      "step": 1668
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0001577037853027691,
+      "loss": 2.3537,
+      "step": 1671
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00015753922965444184,
+      "loss": 2.3758,
+      "step": 1674
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0001573744407522386,
+      "loss": 2.3749,
+      "step": 1677
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00015720941926418455,
+      "loss": 2.3841,
+      "step": 1680
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0001570441658592477,
+      "loss": 2.3546,
+      "step": 1683
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00015687868120733614,
+      "loss": 2.3845,
+      "step": 1686
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00015671296597929535,
+      "loss": 2.2959,
+      "step": 1689
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00015654702084690568,
+      "loss": 2.4619,
+      "step": 1692
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0001563808464828794,
+      "loss": 2.2943,
+      "step": 1695
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00015621444356085803,
+      "loss": 2.3027,
+      "step": 1698
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00015604781275540956,
+      "loss": 2.4349,
+      "step": 1701
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00015588095474202595,
+      "loss": 2.3143,
+      "step": 1704
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00015571387019712004,
+      "loss": 2.3555,
+      "step": 1707
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0001555465597980231,
+      "loss": 2.4337,
+      "step": 1710
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00015537902422298197,
+      "loss": 2.393,
+      "step": 1713
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00015521126415115623,
+      "loss": 2.3029,
+      "step": 1716
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00015504328026261566,
+      "loss": 2.3065,
+      "step": 1719
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.0001548750732383372,
+      "loss": 2.3214,
+      "step": 1722
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00015470664376020246,
+      "loss": 2.3422,
+      "step": 1725
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00015453799251099478,
+      "loss": 2.3227,
+      "step": 1728
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00015436912017439657,
+      "loss": 2.3816,
+      "step": 1731
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00015420002743498645,
+      "loss": 2.3966,
+      "step": 1734
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00015403071497823652,
+      "loss": 2.2734,
+      "step": 1737
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0001538611834905096,
+      "loss": 2.384,
+      "step": 1740
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00015369143365905635,
+      "loss": 2.3495,
+      "step": 1743
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00015352146617201266,
+      "loss": 2.3252,
+      "step": 1746
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00015335128171839671,
+      "loss": 2.349,
+      "step": 1749
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00015318088098810622,
+      "loss": 2.3845,
+      "step": 1752
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0001530102646719156,
+      "loss": 2.3248,
+      "step": 1755
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0001528394334614733,
+      "loss": 2.3738,
+      "step": 1758
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00015266838804929892,
+      "loss": 2.3512,
+      "step": 1761
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00015249712912878031,
+      "loss": 2.4223,
+      "step": 1764
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00015232565739417092,
+      "loss": 2.3593,
+      "step": 1767
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00015215397354058686,
+      "loss": 2.3934,
+      "step": 1770
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00015198207826400413,
+      "loss": 2.4059,
+      "step": 1773
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00015180997226125592,
+      "loss": 2.3081,
+      "step": 1776
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00015163765623002945,
+      "loss": 2.2689,
+      "step": 1779
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00015146513086886356,
+      "loss": 2.3314,
+      "step": 1782
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00015129239687714557,
+      "loss": 2.2807,
+      "step": 1785
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00015111945495510857,
+      "loss": 2.3481,
+      "step": 1788
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0001509463058038286,
+      "loss": 2.3724,
+      "step": 1791
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00015077295012522174,
+      "loss": 2.3322,
+      "step": 1794
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00015059938862204127,
+      "loss": 2.2553,
+      "step": 1797
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0001504256219978749,
+      "loss": 2.2632,
+      "step": 1800
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0001502516509571418,
+      "loss": 2.27,
+      "step": 1803
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00015007747620508988,
+      "loss": 2.3615,
+      "step": 1806
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00014990309844779284,
+      "loss": 2.2815,
+      "step": 1809
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0001497285183921473,
+      "loss": 2.3518,
+      "step": 1812
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00014955373674586996,
+      "loss": 2.3358,
+      "step": 1815
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00014937875421749472,
+      "loss": 2.3217,
+      "step": 1818
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00014920357151636992,
+      "loss": 2.3558,
+      "step": 1821
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00014902818935265527,
+      "loss": 2.3474,
+      "step": 1824
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00014885260843731905,
+      "loss": 2.3579,
+      "step": 1827
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00014867682948213536,
+      "loss": 2.3964,
+      "step": 1830
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.000148500853199681,
+      "loss": 2.3697,
+      "step": 1833
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00014832468030333265,
+      "loss": 2.3099,
+      "step": 1836
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00014814831150726428,
+      "loss": 2.3651,
+      "step": 1839
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00014797174752644382,
+      "loss": 2.304,
+      "step": 1842
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00014779498907663033,
+      "loss": 2.3598,
+      "step": 1845
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.0001476180368743715,
+      "loss": 2.2852,
+      "step": 1848
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.6175154625608633,
+      "eval_loss": 2.245072364807129,
+      "eval_runtime": 16.3034,
+      "eval_samples_per_second": 134.757,
+      "eval_steps_per_second": 67.409,
+      "step": 1848
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00014744089163700025,
+      "loss": 2.395,
+      "step": 1851
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0001472635540826321,
+      "loss": 2.3843,
+      "step": 1854
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00014708602493016218,
+      "loss": 2.3417,
+      "step": 1857
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001469083048992623,
+      "loss": 2.2776,
+      "step": 1860
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00014673039471037807,
+      "loss": 2.2988,
+      "step": 1863
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.000146552295084726,
+      "loss": 2.2561,
+      "step": 1866
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00014637400674429057,
+      "loss": 2.3886,
+      "step": 1869
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00014619553041182116,
+      "loss": 2.2859,
+      "step": 1872
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00014601686681082934,
+      "loss": 2.3711,
+      "step": 1875
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00014583801666558576,
+      "loss": 2.315,
+      "step": 1878
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00014565898070111735,
+      "loss": 2.2431,
+      "step": 1881
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.0001454797596432043,
+      "loss": 2.3051,
+      "step": 1884
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00014530035421837716,
+      "loss": 2.3228,
+      "step": 1887
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00014512076515391375,
+      "loss": 2.3054,
+      "step": 1890
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0001449409931778365,
+      "loss": 2.3504,
+      "step": 1893
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.0001447610390189092,
+      "loss": 2.328,
+      "step": 1896
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00014458090340663428,
+      "loss": 2.268,
+      "step": 1899
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00014440058707124967,
+      "loss": 2.3306,
+      "step": 1902
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00014422009074372604,
+      "loss": 2.2928,
+      "step": 1905
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00014403941515576344,
+      "loss": 2.3729,
+      "step": 1908
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00014385856103978894,
+      "loss": 2.2654,
+      "step": 1911
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.0001436775291289532,
+      "loss": 2.3404,
+      "step": 1914
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00014349632015712752,
+      "loss": 2.2935,
+      "step": 1917
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00014331493485890114,
+      "loss": 2.2743,
+      "step": 1920
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00014313337396957803,
+      "loss": 2.4234,
+      "step": 1923
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00014295163822517393,
+      "loss": 2.3393,
+      "step": 1926
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.0001427697283624135,
+      "loss": 2.3336,
+      "step": 1929
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00014258764511872716,
+      "loss": 2.3729,
+      "step": 1932
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00014240538923224823,
+      "loss": 2.3284,
+      "step": 1935
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00014222296144180994,
+      "loss": 2.3265,
+      "step": 1938
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00014204036248694225,
+      "loss": 2.3156,
+      "step": 1941
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00014185759310786917,
+      "loss": 2.2792,
+      "step": 1944
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00014167465404550542,
+      "loss": 2.2745,
+      "step": 1947
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00014149154604145366,
+      "loss": 2.2669,
+      "step": 1950
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00014130826983800145,
+      "loss": 2.2678,
+      "step": 1953
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.0001411248261781181,
+      "loss": 2.3385,
+      "step": 1956
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00014094121580545183,
+      "loss": 2.3564,
+      "step": 1959
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.0001407574394643267,
+      "loss": 2.2749,
+      "step": 1962
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00014057349789973946,
+      "loss": 2.2941,
+      "step": 1965
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00014038939185735683,
+      "loss": 2.2932,
+      "step": 1968
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.0001402051220835121,
+      "loss": 2.2875,
+      "step": 1971
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00014002068932520247,
+      "loss": 2.3496,
+      "step": 1974
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00013983609433008574,
+      "loss": 2.273,
+      "step": 1977
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.0001396513378464774,
+      "loss": 2.2399,
+      "step": 1980
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00013946642062334766,
+      "loss": 2.3506,
+      "step": 1983
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00013928134341031825,
+      "loss": 2.2949,
+      "step": 1986
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00013909610695765948,
+      "loss": 2.3473,
+      "step": 1989
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00013891071201628728,
+      "loss": 2.2964,
+      "step": 1992
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00013872515933776,
+      "loss": 2.3721,
+      "step": 1995
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00013853944967427535,
+      "loss": 2.3076,
+      "step": 1998
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00013835358377866763,
+      "loss": 2.3327,
+      "step": 2001
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00013816756240440424,
+      "loss": 2.2885,
+      "step": 2004
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00013798138630558303,
+      "loss": 2.3773,
+      "step": 2007
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00013779505623692909,
+      "loss": 2.3261,
+      "step": 2010
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00013760857295379154,
+      "loss": 2.3456,
+      "step": 2013
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00013742193721214064,
+      "loss": 2.422,
+      "step": 2016
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00013723514976856483,
+      "loss": 2.2936,
+      "step": 2019
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00013704821138026737,
+      "loss": 2.3485,
+      "step": 2022
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00013686112280506346,
+      "loss": 2.2716,
+      "step": 2025
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00013667388480137716,
+      "loss": 2.291,
+      "step": 2028
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0001364864981282383,
+      "loss": 2.3078,
+      "step": 2031
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00013629896354527932,
+      "loss": 2.3167,
+      "step": 2034
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.0001361112818127323,
+      "loss": 2.3923,
+      "step": 2037
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013592345369142585,
+      "loss": 2.3003,
+      "step": 2040
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013573547994278205,
+      "loss": 2.2712,
+      "step": 2043
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0001355473613288132,
+      "loss": 2.3246,
+      "step": 2046
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00013535909861211903,
+      "loss": 2.2397,
+      "step": 2049
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013517069255588327,
+      "loss": 2.2596,
+      "step": 2052
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013498214392387083,
+      "loss": 2.1575,
+      "step": 2055
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.0001347934534804246,
+      "loss": 2.2904,
+      "step": 2058
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00013460462199046226,
+      "loss": 2.1767,
+      "step": 2061
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00013441565021947332,
+      "loss": 2.2593,
+      "step": 2064
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00013422653893351604,
+      "loss": 2.3004,
+      "step": 2067
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.0001340372888992141,
+      "loss": 2.2932,
+      "step": 2070
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.0001338479008837538,
+      "loss": 2.3329,
+      "step": 2073
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00013365837565488064,
+      "loss": 2.2765,
+      "step": 2076
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00013346871398089644,
+      "loss": 2.2924,
+      "step": 2079
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00013327891663065614,
+      "loss": 2.2459,
+      "step": 2082
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.0001330889843735647,
+      "loss": 2.1898,
+      "step": 2085
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00013289891797957395,
+      "loss": 2.2234,
+      "step": 2088
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00013270871821917946,
+      "loss": 2.3638,
+      "step": 2091
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00013251838586341745,
+      "loss": 2.3405,
+      "step": 2094
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00013232792168386176,
+      "loss": 2.2713,
+      "step": 2097
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00013213732645262044,
+      "loss": 2.2664,
+      "step": 2100
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00013194660094233298,
+      "loss": 2.2861,
+      "step": 2103
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00013175574592616692,
+      "loss": 2.2772,
+      "step": 2106
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00013156476217781468,
+      "loss": 2.2263,
+      "step": 2109
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00013137365047149078,
+      "loss": 2.3177,
+      "step": 2112
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00013118241158192827,
+      "loss": 2.2316,
+      "step": 2115
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.0001309910462843758,
+      "loss": 2.3542,
+      "step": 2118
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00013079955535459455,
+      "loss": 2.2913,
+      "step": 2121
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.000130607939568855,
+      "loss": 2.2785,
+      "step": 2124
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00013041619970393352,
+      "loss": 2.2508,
+      "step": 2127
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.0001302243365371098,
+      "loss": 2.2514,
+      "step": 2130
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00013003235084616324,
+      "loss": 2.2517,
+      "step": 2133
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00012984024340936983,
+      "loss": 2.2517,
+      "step": 2136
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00012964801500549931,
+      "loss": 2.2079,
+      "step": 2139
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00012945566641381159,
+      "loss": 2.3606,
+      "step": 2142
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00012926319841405394,
+      "loss": 2.3764,
+      "step": 2145
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00012907061178645763,
+      "loss": 2.2781,
+      "step": 2148
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00012887790731173486,
+      "loss": 2.2436,
+      "step": 2151
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.0001286850857710755,
+      "loss": 2.228,
+      "step": 2154
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.6243810506983248,
+      "eval_loss": 2.193675994873047,
+      "eval_runtime": 16.3306,
+      "eval_samples_per_second": 134.533,
+      "eval_steps_per_second": 67.297,
+      "step": 2156
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00012849214794614407,
+      "loss": 2.2569,
+      "step": 2157
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 0.0001282990946190764,
+      "loss": 2.3658,
+      "step": 2160
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.00012810592657247656,
+      "loss": 2.2651,
+      "step": 2163
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.0001279126445894138,
+      "loss": 2.215,
+      "step": 2166
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.00012771924945341906,
+      "loss": 2.2391,
+      "step": 2169
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.00012752574194848211,
+      "loss": 2.2661,
+      "step": 2172
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.00012733212285904818,
+      "loss": 2.1834,
+      "step": 2175
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.0001271383929700149,
+      "loss": 2.2816,
+      "step": 2178
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.00012694455306672895,
+      "loss": 2.2706,
+      "step": 2181
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.00012675060393498318,
+      "loss": 2.2744,
+      "step": 2184
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 0.00012655654636101304,
+      "loss": 2.2792,
+      "step": 2187
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.00012636238113149367,
+      "loss": 2.3216,
+      "step": 2190
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 0.00012616810903353666,
+      "loss": 2.2724,
+      "step": 2193
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.00012597373085468678,
+      "loss": 2.2908,
+      "step": 2196
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.00012577924738291877,
+      "loss": 2.3213,
+      "step": 2199
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 0.0001255846594066344,
+      "loss": 2.2962,
+      "step": 2202
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 0.00012538996771465887,
+      "loss": 2.2341,
+      "step": 2205
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 0.00012519517309623793,
+      "loss": 2.3292,
+      "step": 2208
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.0001250002763410346,
+      "loss": 2.2678,
+      "step": 2211
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 0.0001248052782391259,
+      "loss": 2.2799,
+      "step": 2214
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.00012461017958099966,
+      "loss": 2.2489,
+      "step": 2217
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.00012441498115755146,
+      "loss": 2.3042,
+      "step": 2220
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 0.00012421968376008115,
+      "loss": 2.1692,
+      "step": 2223
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.00012402428818028994,
+      "loss": 2.3398,
+      "step": 2226
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 0.000123828795210277,
+      "loss": 2.2435,
+      "step": 2229
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.00012363320564253637,
+      "loss": 2.2838,
+      "step": 2232
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 0.0001234375202699535,
+      "loss": 2.1099,
+      "step": 2235
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.00012324173988580235,
+      "loss": 2.239,
+      "step": 2238
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 0.0001230458652837421,
+      "loss": 2.2111,
+      "step": 2241
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.00012284989725781377,
+      "loss": 2.2932,
+      "step": 2244
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.0001226538366024371,
+      "loss": 2.2204,
+      "step": 2247
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 0.00012245768411240737,
+      "loss": 2.2478,
+      "step": 2250
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 0.00012226144058289216,
+      "loss": 2.2759,
+      "step": 2253
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.00012206510680942806,
+      "loss": 2.297,
+      "step": 2256
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 0.00012186868358791756,
+      "loss": 2.2437,
+      "step": 2259
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 0.00012167217171462566,
+      "loss": 2.2756,
+      "step": 2262
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.00012147557198617678,
+      "loss": 2.2352,
+      "step": 2265
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.00012127888519955157,
+      "loss": 2.2638,
+      "step": 2268
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 0.0001210821121520835,
+      "loss": 2.1679,
+      "step": 2271
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.0001208852536414557,
+      "loss": 2.2499,
+      "step": 2274
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.00012068831046569789,
+      "loss": 2.2459,
+      "step": 2277
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 0.00012049128342318288,
+      "loss": 2.1518,
+      "step": 2280
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.00012029417331262349,
+      "loss": 2.246,
+      "step": 2283
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.00012009698093306936,
+      "loss": 2.2471,
+      "step": 2286
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 0.00011989970708390353,
+      "loss": 2.2332,
+      "step": 2289
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.00011970235256483934,
+      "loss": 2.2554,
+      "step": 2292
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.00011950491817591717,
+      "loss": 2.2454,
+      "step": 2295
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 0.00011930740471750121,
+      "loss": 2.2378,
+      "step": 2298
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.00011910981299027608,
+      "loss": 2.2978,
+      "step": 2301
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.00011891214379524375,
+      "loss": 2.2042,
+      "step": 2304
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 0.0001187143979337203,
+      "loss": 2.2105,
+      "step": 2307
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 0.00011851657620733243,
+      "loss": 2.2343,
+      "step": 2310
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.00011831867941801455,
+      "loss": 2.2197,
+      "step": 2313
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 0.00011812070836800533,
+      "loss": 2.3094,
+      "step": 2316
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 0.00011792266385984433,
+      "loss": 2.2106,
+      "step": 2319
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.00011772454669636912,
+      "loss": 2.3657,
+      "step": 2322
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 0.00011752635768071167,
+      "loss": 2.3096,
+      "step": 2325
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 0.0001173280976162952,
+      "loss": 2.1856,
+      "step": 2328
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 0.00011712976730683108,
+      "loss": 2.236,
+      "step": 2331
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 0.00011693136755631528,
+      "loss": 2.2304,
+      "step": 2334
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 0.00011673289916902539,
+      "loss": 2.2346,
+      "step": 2337
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 0.00011653436294951724,
+      "loss": 2.2353,
+      "step": 2340
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 0.00011633575970262152,
+      "loss": 2.2511,
+      "step": 2343
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 0.0001161370902334408,
+      "loss": 2.2557,
+      "step": 2346
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 0.00011593835534734596,
+      "loss": 2.2501,
+      "step": 2349
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 0.00011573955584997318,
+      "loss": 2.2231,
+      "step": 2352
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 0.00011554069254722051,
+      "loss": 2.1851,
+      "step": 2355
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 0.00011534176624524464,
+      "loss": 2.2573,
+      "step": 2358
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 0.00011514277775045768,
+      "loss": 2.2314,
+      "step": 2361
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 0.00011494372786952384,
+      "loss": 2.1805,
+      "step": 2364
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 0.00011474461740935621,
+      "loss": 2.2441,
+      "step": 2367
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 0.0001145454471771134,
+      "loss": 2.2018,
+      "step": 2370
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 0.0001143462179801964,
+      "loss": 2.1672,
+      "step": 2373
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 0.00011414693062624515,
+      "loss": 2.2151,
+      "step": 2376
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 0.00011394758592313543,
+      "loss": 2.2059,
+      "step": 2379
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 0.00011374818467897541,
+      "loss": 2.2516,
+      "step": 2382
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 0.00011354872770210256,
+      "loss": 2.1991,
+      "step": 2385
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 0.00011334921580108027,
+      "loss": 2.2307,
+      "step": 2388
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 0.00011314964978469445,
+      "loss": 2.1478,
+      "step": 2391
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 0.00011295003046195058,
+      "loss": 2.1494,
+      "step": 2394
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 0.00011275035864207017,
+      "loss": 2.289,
+      "step": 2397
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 0.00011255063513448743,
+      "loss": 2.1709,
+      "step": 2400
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 0.00011235086074884622,
+      "loss": 2.2586,
+      "step": 2403
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 0.00011215103629499661,
+      "loss": 2.2279,
+      "step": 2406
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 0.00011195116258299169,
+      "loss": 2.2403,
+      "step": 2409
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 0.00011175124042308416,
+      "loss": 2.2453,
+      "step": 2412
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 0.00011155127062572314,
+      "loss": 2.2293,
+      "step": 2415
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 0.00011135125400155091,
+      "loss": 2.2598,
+      "step": 2418
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 0.00011115119136139951,
+      "loss": 2.2331,
+      "step": 2421
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 0.00011095108351628758,
+      "loss": 2.2416,
+      "step": 2424
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 0.00011075093127741695,
+      "loss": 2.1992,
+      "step": 2427
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 0.00011055073545616952,
+      "loss": 2.1727,
+      "step": 2430
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 0.0001103504968641037,
+      "loss": 2.2371,
+      "step": 2433
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 0.00011015021631295149,
+      "loss": 2.23,
+      "step": 2436
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 0.00010994989461461476,
+      "loss": 2.1677,
+      "step": 2439
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 0.00010974953258116238,
+      "loss": 2.2252,
+      "step": 2442
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 0.00010954913102482664,
+      "loss": 2.2119,
+      "step": 2445
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 0.000109348690758,
+      "loss": 2.2567,
+      "step": 2448
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 0.00010914821259323202,
+      "loss": 2.2209,
+      "step": 2451
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 0.00010894769734322567,
+      "loss": 2.2701,
+      "step": 2454
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 0.00010874714582083438,
+      "loss": 2.1552,
+      "step": 2457
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 0.00010854655883905869,
+      "loss": 2.1527,
+      "step": 2460
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 0.0001083459372110427,
+      "loss": 2.2013,
+      "step": 2463
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.6309924029431118,
+      "eval_loss": 2.1445603370666504,
+      "eval_runtime": 16.3353,
+      "eval_samples_per_second": 134.494,
+      "eval_steps_per_second": 67.278,
+      "step": 2464
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 0.00010814528175007108,
+      "loss": 2.22,
+      "step": 2466
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 0.0001079445932695657,
+      "loss": 2.2225,
+      "step": 2469
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 0.00010774387258308217,
+      "loss": 2.2667,
+      "step": 2472
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 0.00010754312050430668,
+      "loss": 2.2468,
+      "step": 2475
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 0.00010734233784705276,
+      "loss": 2.2416,
+      "step": 2478
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 0.00010714152542525781,
+      "loss": 2.1588,
+      "step": 2481
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 0.00010694068405297997,
+      "loss": 2.2093,
+      "step": 2484
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 0.00010673981454439463,
+      "loss": 2.2511,
+      "step": 2487
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 0.00010653891771379134,
+      "loss": 2.2265,
+      "step": 2490
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 0.00010633799437557039,
+      "loss": 2.2257,
+      "step": 2493
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 0.00010613704534423949,
+      "loss": 2.2588,
+      "step": 2496
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 0.00010593607143441053,
+      "loss": 2.1435,
+      "step": 2499
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 0.00010573507346079625,
+      "loss": 2.1494,
+      "step": 2502
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 0.00010553405223820693,
+      "loss": 2.2067,
+      "step": 2505
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 0.00010533300858154715,
+      "loss": 2.2174,
+      "step": 2508
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 0.00010513194330581233,
+      "loss": 2.168,
+      "step": 2511
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 0.00010493085722608562,
+      "loss": 2.2395,
+      "step": 2514
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 0.00010472975115753452,
+      "loss": 2.1584,
+      "step": 2517
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 0.00010452862591540742,
+      "loss": 2.1738,
+      "step": 2520
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 0.0001043274823150306,
+      "loss": 2.2533,
+      "step": 2523
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 0.00010412632117180471,
+      "loss": 2.308,
+      "step": 2526
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 0.00010392514330120145,
+      "loss": 2.1664,
+      "step": 2529
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 0.00010372394951876043,
+      "loss": 2.2669,
+      "step": 2532
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 0.00010352274064008567,
+      "loss": 2.1636,
+      "step": 2535
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 0.00010332151748084242,
+      "loss": 2.2729,
+      "step": 2538
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 0.00010312028085675391,
+      "loss": 2.2097,
+      "step": 2541
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 0.00010291903158359783,
+      "loss": 2.2306,
+      "step": 2544
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 0.0001027177704772032,
+      "loss": 2.1675,
+      "step": 2547
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 0.00010251649835344696,
+      "loss": 2.266,
+      "step": 2550
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 0.0001023152160282508,
+      "loss": 2.2716,
+      "step": 2553
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 0.00010211392431757773,
+      "loss": 2.2013,
+      "step": 2556
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 0.00010191262403742878,
+      "loss": 2.1526,
+      "step": 2559
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 0.00010171131600383974,
+      "loss": 2.1521,
+      "step": 2562
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 0.00010151000103287784,
+      "loss": 2.2781,
+      "step": 2565
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 0.00010130867994063839,
+      "loss": 2.1813,
+      "step": 2568
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 0.00010110735354324159,
+      "loss": 2.2153,
+      "step": 2571
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 0.00010090602265682906,
+      "loss": 2.2226,
+      "step": 2574
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 0.00010070468809756068,
+      "loss": 2.2177,
+      "step": 2577
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 0.00010050335068161123,
+      "loss": 2.186,
+      "step": 2580
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 0.00010030201122516696,
+      "loss": 2.2026,
+      "step": 2583
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 0.00010010067054442251,
+      "loss": 2.2229,
+      "step": 2586
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 9.989932945557751e-05,
+      "loss": 2.2416,
+      "step": 2589
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 9.969798877483308e-05,
+      "loss": 2.2854,
+      "step": 2592
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 9.949664931838882e-05,
+      "loss": 2.158,
+      "step": 2595
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 9.929531190243932e-05,
+      "loss": 2.2394,
+      "step": 2598
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 9.909397734317095e-05,
+      "loss": 2.1703,
+      "step": 2601
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 9.889264645675843e-05,
+      "loss": 2.2031,
+      "step": 2604
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 9.869132005936163e-05,
+      "loss": 2.2224,
+      "step": 2607
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 9.848999896712217e-05,
+      "loss": 2.2693,
+      "step": 2610
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 9.82886839961603e-05,
+      "loss": 2.2381,
+      "step": 2613
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 9.808737596257121e-05,
+      "loss": 2.1839,
+      "step": 2616
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 9.788607568242229e-05,
+      "loss": 2.129,
+      "step": 2619
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 9.768478397174922e-05,
+      "loss": 2.121,
+      "step": 2622
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 9.748350164655306e-05,
+      "loss": 2.2323,
+      "step": 2625
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 9.728222952279684e-05,
+      "loss": 2.2613,
+      "step": 2628
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 9.708096841640222e-05,
+      "loss": 2.1113,
+      "step": 2631
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 9.687971914324607e-05,
+      "loss": 2.1729,
+      "step": 2634
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 9.667848251915758e-05,
+      "loss": 2.2271,
+      "step": 2637
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 9.647725935991436e-05,
+      "loss": 2.2319,
+      "step": 2640
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 9.627605048123959e-05,
+      "loss": 2.2244,
+      "step": 2643
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 9.607485669879857e-05,
+      "loss": 2.1239,
+      "step": 2646
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 9.587367882819532e-05,
+      "loss": 2.2429,
+      "step": 2649
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 9.567251768496938e-05,
+      "loss": 2.1936,
+      "step": 2652
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 9.547137408459257e-05,
+      "loss": 2.2038,
+      "step": 2655
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 9.52702488424655e-05,
+      "loss": 2.1978,
+      "step": 2658
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 9.506914277391439e-05,
+      "loss": 2.1977,
+      "step": 2661
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 9.486805669418769e-05,
+      "loss": 2.1772,
+      "step": 2664
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 9.466699141845287e-05,
+      "loss": 2.1929,
+      "step": 2667
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 9.446594776179306e-05,
+      "loss": 2.2712,
+      "step": 2670
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 9.426492653920375e-05,
+      "loss": 2.2606,
+      "step": 2673
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 9.406392856558949e-05,
+      "loss": 2.1655,
+      "step": 2676
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 9.386295465576053e-05,
+      "loss": 2.1544,
+      "step": 2679
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 9.366200562442963e-05,
+      "loss": 2.2016,
+      "step": 2682
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 9.346108228620868e-05,
+      "loss": 2.3044,
+      "step": 2685
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 9.326018545560542e-05,
+      "loss": 2.2102,
+      "step": 2688
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 9.305931594702007e-05,
+      "loss": 2.1511,
+      "step": 2691
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 9.28584745747422e-05,
+      "loss": 2.212,
+      "step": 2694
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 9.265766215294725e-05,
+      "loss": 2.1871,
+      "step": 2697
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 9.245687949569332e-05,
+      "loss": 2.226,
+      "step": 2700
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 9.225612741691788e-05,
+      "loss": 2.1323,
+      "step": 2703
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 9.205540673043434e-05,
+      "loss": 2.1258,
+      "step": 2706
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 9.185471824992891e-05,
+      "loss": 2.1963,
+      "step": 2709
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 9.165406278895732e-05,
+      "loss": 2.2423,
+      "step": 2712
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 9.145344116094134e-05,
+      "loss": 2.1678,
+      "step": 2715
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 9.125285417916563e-05,
+      "loss": 2.196,
+      "step": 2718
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 9.105230265677437e-05,
+      "loss": 2.1637,
+      "step": 2721
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 9.085178740676803e-05,
+      "loss": 2.1019,
+      "step": 2724
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 9.065130924199998e-05,
+      "loss": 2.1651,
+      "step": 2727
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 9.045086897517337e-05,
+      "loss": 2.2656,
+      "step": 2730
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 9.025046741883764e-05,
+      "loss": 2.2224,
+      "step": 2733
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 9.005010538538527e-05,
+      "loss": 2.1764,
+      "step": 2736
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 8.984978368704855e-05,
+      "loss": 2.1928,
+      "step": 2739
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 8.964950313589633e-05,
+      "loss": 2.111,
+      "step": 2742
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 8.944926454383049e-05,
+      "loss": 2.2286,
+      "step": 2745
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 8.924906872258306e-05,
+      "loss": 2.18,
+      "step": 2748
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 8.904891648371244e-05,
+      "loss": 2.1869,
+      "step": 2751
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 8.884880863860051e-05,
+      "loss": 2.2054,
+      "step": 2754
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 8.864874599844911e-05,
+      "loss": 2.1351,
+      "step": 2757
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 8.84487293742769e-05,
+      "loss": 2.2412,
+      "step": 2760
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 8.824875957691588e-05,
+      "loss": 2.2394,
+      "step": 2763
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 8.804883741700833e-05,
+      "loss": 2.1228,
+      "step": 2766
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 8.78489637050034e-05,
+      "loss": 2.1862,
+      "step": 2769
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 8.764913925115381e-05,
+      "loss": 2.1463,
+      "step": 2772
+    },
+    {
+      "epoch": 9.0,
+      "eval_accuracy": 0.6357306777396295,
+      "eval_loss": 2.106170654296875,
+      "eval_runtime": 16.2999,
+      "eval_samples_per_second": 134.786,
+      "eval_steps_per_second": 67.424,
+      "step": 2772
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 8.744936486551262e-05,
+      "loss": 2.2427,
+      "step": 2775
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 8.724964135792988e-05,
+      "loss": 2.105,
+      "step": 2778
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 8.70499695380494e-05,
+      "loss": 2.2257,
+      "step": 2781
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 8.685035021530554e-05,
+      "loss": 2.2433,
+      "step": 2784
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 8.665078419891977e-05,
+      "loss": 2.1763,
+      "step": 2787
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 8.645127229789746e-05,
+      "loss": 2.2108,
+      "step": 2790
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 8.625181532102463e-05,
+      "loss": 2.115,
+      "step": 2793
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 8.605241407686462e-05,
+      "loss": 2.2288,
+      "step": 2796
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 8.585306937375486e-05,
+      "loss": 2.1351,
+      "step": 2799
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 8.565378201980361e-05,
+      "loss": 2.1875,
+      "step": 2802
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 8.545455282288661e-05,
+      "loss": 2.1558,
+      "step": 2805
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 8.525538259064381e-05,
+      "loss": 2.2156,
+      "step": 2808
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 8.505627213047617e-05,
+      "loss": 2.1982,
+      "step": 2811
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 8.485722224954237e-05,
+      "loss": 2.1447,
+      "step": 2814
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 8.465823375475537e-05,
+      "loss": 2.2311,
+      "step": 2817
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 8.445930745277953e-05,
+      "loss": 2.1685,
+      "step": 2820
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 8.426044415002684e-05,
+      "loss": 2.1237,
+      "step": 2823
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 8.406164465265406e-05,
+      "loss": 2.1082,
+      "step": 2826
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 8.386290976655924e-05,
+      "loss": 2.1352,
+      "step": 2829
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 8.366424029737853e-05,
+      "loss": 2.1588,
+      "step": 2832
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 8.346563705048277e-05,
+      "loss": 2.2979,
+      "step": 2835
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 8.326710083097462e-05,
+      "loss": 2.1507,
+      "step": 2838
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 8.306863244368474e-05,
+      "loss": 2.127,
+      "step": 2841
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 8.287023269316894e-05,
+      "loss": 2.1869,
+      "step": 2844
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 8.267190238370482e-05,
+      "loss": 2.1259,
+      "step": 2847
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 8.247364231928837e-05,
+      "loss": 2.2649,
+      "step": 2850
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 8.227545330363087e-05,
+      "loss": 2.1888,
+      "step": 2853
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 8.207733614015566e-05,
+      "loss": 2.1949,
+      "step": 2856
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 8.18792916319947e-05,
+      "loss": 2.1379,
+      "step": 2859
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 8.168132058198546e-05,
+      "loss": 2.1585,
+      "step": 2862
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 8.148342379266759e-05,
+      "loss": 2.1941,
+      "step": 2865
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 8.128560206627974e-05,
+      "loss": 2.1447,
+      "step": 2868
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 8.108785620475624e-05,
+      "loss": 2.1413,
+      "step": 2871
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 8.089018700972393e-05,
+      "loss": 2.25,
+      "step": 2874
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 8.069259528249882e-05,
+      "loss": 2.2048,
+      "step": 2877
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 8.049508182408284e-05,
+      "loss": 2.1432,
+      "step": 2880
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 8.029764743516068e-05,
+      "loss": 2.2312,
+      "step": 2883
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 8.01002929160965e-05,
+      "loss": 2.1929,
+      "step": 2886
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 7.990301906693069e-05,
+      "loss": 2.1687,
+      "step": 2889
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 7.970582668737652e-05,
+      "loss": 2.1176,
+      "step": 2892
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 7.950871657681716e-05,
+      "loss": 2.1793,
+      "step": 2895
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 7.931168953430213e-05,
+      "loss": 2.1538,
+      "step": 2898
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 7.91147463585443e-05,
+      "loss": 2.1519,
+      "step": 2901
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 7.891788784791655e-05,
+      "loss": 2.1821,
+      "step": 2904
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 7.872111480044847e-05,
+      "loss": 2.145,
+      "step": 2907
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 7.852442801382322e-05,
+      "loss": 2.1574,
+      "step": 2910
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 7.832782828537437e-05,
+      "loss": 2.2808,
+      "step": 2913
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 7.813131641208245e-05,
+      "loss": 2.1604,
+      "step": 2916
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 7.793489319057195e-05,
+      "loss": 2.1786,
+      "step": 2919
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 7.773855941710786e-05,
+      "loss": 2.2453,
+      "step": 2922
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 7.754231588759265e-05,
+      "loss": 2.2529,
+      "step": 2925
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 7.734616339756291e-05,
+      "loss": 2.1199,
+      "step": 2928
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 7.715010274218625e-05,
+      "loss": 2.2108,
+      "step": 2931
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 7.695413471625792e-05,
+      "loss": 2.2136,
+      "step": 2934
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 7.675826011419766e-05,
+      "loss": 2.1859,
+      "step": 2937
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 7.656247973004656e-05,
+      "loss": 2.1492,
+      "step": 2940
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 7.63667943574637e-05,
+      "loss": 2.2054,
+      "step": 2943
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 7.617120478972297e-05,
+      "loss": 2.1295,
+      "step": 2946
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 7.597571181971006e-05,
+      "loss": 2.0959,
+      "step": 2949
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 7.578031623991886e-05,
+      "loss": 2.2796,
+      "step": 2952
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 7.558501884244857e-05,
+      "loss": 2.0984,
+      "step": 2955
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 7.538982041900033e-05,
+      "loss": 2.2107,
+      "step": 2958
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 7.519472176087414e-05,
+      "loss": 2.1768,
+      "step": 2961
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 7.49997236589654e-05,
+      "loss": 2.1817,
+      "step": 2964
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 7.480482690376207e-05,
+      "loss": 2.1886,
+      "step": 2967
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 7.461003228534115e-05,
+      "loss": 2.2048,
+      "step": 2970
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 7.441534059336563e-05,
+      "loss": 2.1702,
+      "step": 2973
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 7.422075261708125e-05,
+      "loss": 2.1699,
+      "step": 2976
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 7.402626914531328e-05,
+      "loss": 2.1161,
+      "step": 2979
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 7.383189096646335e-05,
+      "loss": 2.1665,
+      "step": 2982
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 7.363761886850633e-05,
+      "loss": 2.196,
+      "step": 2985
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 7.344345363898697e-05,
+      "loss": 2.1785,
+      "step": 2988
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 7.324939606501685e-05,
+      "loss": 2.1046,
+      "step": 2991
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 7.305544693327106e-05,
+      "loss": 2.1544,
+      "step": 2994
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 7.286160702998515e-05,
+      "loss": 2.1705,
+      "step": 2997
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 7.266787714095182e-05,
+      "loss": 2.2029,
+      "step": 3000
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 7.247425805151788e-05,
+      "loss": 2.1483,
+      "step": 3003
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.228075054658096e-05,
+      "loss": 2.1243,
+      "step": 3006
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.208735541058622e-05,
+      "loss": 2.0956,
+      "step": 3009
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.189407342752345e-05,
+      "loss": 2.1788,
+      "step": 3012
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 7.170090538092367e-05,
+      "loss": 2.2902,
+      "step": 3015
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 7.150785205385596e-05,
+      "loss": 2.0941,
+      "step": 3018
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 7.131491422892454e-05,
+      "loss": 2.1134,
+      "step": 3021
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 7.112209268826517e-05,
+      "loss": 2.1195,
+      "step": 3024
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 7.092938821354238e-05,
+      "loss": 2.2264,
+      "step": 3027
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 7.073680158594609e-05,
+      "loss": 2.205,
+      "step": 3030
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 7.054433358618847e-05,
+      "loss": 2.1962,
+      "step": 3033
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 7.035198499450071e-05,
+      "loss": 2.1881,
+      "step": 3036
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 7.015975659063017e-05,
+      "loss": 2.1807,
+      "step": 3039
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 6.996764915383681e-05,
+      "loss": 2.1038,
+      "step": 3042
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 6.97756634628902e-05,
+      "loss": 2.1783,
+      "step": 3045
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 6.95838002960665e-05,
+      "loss": 2.1281,
+      "step": 3048
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 6.939206043114506e-05,
+      "loss": 2.1633,
+      "step": 3051
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 6.920044464540543e-05,
+      "loss": 2.1237,
+      "step": 3054
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 6.900895371562419e-05,
+      "loss": 2.1732,
+      "step": 3057
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 6.881758841807176e-05,
+      "loss": 2.139,
+      "step": 3060
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 6.862634952850926e-05,
+      "loss": 2.1296,
+      "step": 3063
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 6.843523782218534e-05,
+      "loss": 2.1914,
+      "step": 3066
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 6.824425407383315e-05,
+      "loss": 2.1584,
+      "step": 3069
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 6.805339905766706e-05,
+      "loss": 2.128,
+      "step": 3072
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 6.786267354737955e-05,
+      "loss": 2.082,
+      "step": 3075
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 6.767207831613828e-05,
+      "loss": 2.0882,
+      "step": 3078
+    },
+    {
+      "epoch": 10.0,
+      "eval_accuracy": 0.636978362829467,
+      "eval_loss": 2.084742307662964,
+      "eval_runtime": 16.3444,
+      "eval_samples_per_second": 134.419,
+      "eval_steps_per_second": 67.24,
+      "step": 3080
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 6.748161413658256e-05,
+      "loss": 2.0895,
+      "step": 3081
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 6.729128178082058e-05,
+      "loss": 2.143,
+      "step": 3084
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 6.71010820204261e-05,
+      "loss": 2.1523,
+      "step": 3087
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 6.691101562643534e-05,
+      "loss": 2.1812,
+      "step": 3090
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 6.672108336934386e-05,
+      "loss": 2.1807,
+      "step": 3093
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 6.653128601910357e-05,
+      "loss": 2.1916,
+      "step": 3096
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 6.63416243451194e-05,
+      "loss": 2.2036,
+      "step": 3099
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 6.615209911624623e-05,
+      "loss": 2.1112,
+      "step": 3102
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 6.596271110078591e-05,
+      "loss": 2.0984,
+      "step": 3105
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 6.577346106648399e-05,
+      "loss": 2.2862,
+      "step": 3108
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 6.558434978052667e-05,
+      "loss": 2.1379,
+      "step": 3111
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 6.539537800953777e-05,
+      "loss": 2.1475,
+      "step": 3114
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 6.520654651957543e-05,
+      "loss": 2.144,
+      "step": 3117
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 6.50178560761292e-05,
+      "loss": 2.1383,
+      "step": 3120
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 6.482930744411677e-05,
+      "loss": 2.0835,
+      "step": 3123
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 6.464090138788102e-05,
+      "loss": 2.0889,
+      "step": 3126
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 6.445263867118679e-05,
+      "loss": 2.2135,
+      "step": 3129
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 6.426452005721797e-05,
+      "loss": 2.1301,
+      "step": 3132
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 6.407654630857416e-05,
+      "loss": 2.0498,
+      "step": 3135
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 6.388871818726774e-05,
+      "loss": 2.0766,
+      "step": 3138
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 6.370103645472072e-05,
+      "loss": 2.195,
+      "step": 3141
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 6.351350187176176e-05,
+      "loss": 2.2205,
+      "step": 3144
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 6.332611519862284e-05,
+      "loss": 2.1172,
+      "step": 3147
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 6.313887719493657e-05,
+      "loss": 2.1688,
+      "step": 3150
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 6.295178861973267e-05,
+      "loss": 2.0986,
+      "step": 3153
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 6.27648502314352e-05,
+      "loss": 2.0658,
+      "step": 3156
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 6.257806278785937e-05,
+      "loss": 2.1681,
+      "step": 3159
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 6.239142704620853e-05,
+      "loss": 2.2028,
+      "step": 3162
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 6.220494376307094e-05,
+      "loss": 2.1707,
+      "step": 3165
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 6.201861369441697e-05,
+      "loss": 2.2144,
+      "step": 3168
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 6.183243759559579e-05,
+      "loss": 2.2155,
+      "step": 3171
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 6.164641622133241e-05,
+      "loss": 2.1628,
+      "step": 3174
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 6.146055032572466e-05,
+      "loss": 2.1457,
+      "step": 3177
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 6.127484066224005e-05,
+      "loss": 2.079,
+      "step": 3180
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 6.108928798371272e-05,
+      "loss": 2.1211,
+      "step": 3183
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 6.090389304234052e-05,
+      "loss": 2.1723,
+      "step": 3186
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 6.0718656589681764e-05,
+      "loss": 2.1723,
+      "step": 3189
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 6.053357937665237e-05,
+      "loss": 2.1179,
+      "step": 3192
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 6.034866215352262e-05,
+      "loss": 2.2066,
+      "step": 3195
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 6.016390566991429e-05,
+      "loss": 2.1562,
+      "step": 3198
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 5.997931067479753e-05,
+      "loss": 2.1374,
+      "step": 3201
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 5.979487791648789e-05,
+      "loss": 2.1595,
+      "step": 3204
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 5.961060814264321e-05,
+      "loss": 2.194,
+      "step": 3207
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 5.942650210026055e-05,
+      "loss": 2.1749,
+      "step": 3210
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 5.9242560535673344e-05,
+      "loss": 2.207,
+      "step": 3213
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 5.905878419454821e-05,
+      "loss": 2.1641,
+      "step": 3216
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 5.8875173821881904e-05,
+      "loss": 2.2015,
+      "step": 3219
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 5.869173016199858e-05,
+      "loss": 2.0588,
+      "step": 3222
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 5.850845395854636e-05,
+      "loss": 2.1809,
+      "step": 3225
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 5.8325345954494633e-05,
+      "loss": 2.0862,
+      "step": 3228
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 5.814240689213086e-05,
+      "loss": 2.1122,
+      "step": 3231
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 5.795963751305777e-05,
+      "loss": 2.1289,
+      "step": 3234
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 5.77770385581901e-05,
+      "loss": 2.1519,
+      "step": 3237
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 5.759461076775177e-05,
+      "loss": 2.1731,
+      "step": 3240
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 5.7412354881272865e-05,
+      "loss": 2.1847,
+      "step": 3243
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 5.7230271637586555e-05,
+      "loss": 2.2063,
+      "step": 3246
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 5.7048361774826086e-05,
+      "loss": 2.1409,
+      "step": 3249
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 5.686662603042201e-05,
+      "loss": 2.0635,
+      "step": 3252
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 5.668506514109887e-05,
+      "loss": 2.0779,
+      "step": 3255
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 5.6503679842872506e-05,
+      "loss": 2.0536,
+      "step": 3258
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 5.6322470871046825e-05,
+      "loss": 2.1569,
+      "step": 3261
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 5.6141438960211065e-05,
+      "loss": 2.1513,
+      "step": 3264
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 5.596058484423656e-05,
+      "loss": 2.1937,
+      "step": 3267
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 5.5779909256274035e-05,
+      "loss": 2.1962,
+      "step": 3270
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 5.559941292875035e-05,
+      "loss": 2.1182,
+      "step": 3273
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 5.5419096593365724e-05,
+      "loss": 2.1865,
+      "step": 3276
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 5.523896098109079e-05,
+      "loss": 2.158,
+      "step": 3279
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 5.505900682216354e-05,
+      "loss": 2.0896,
+      "step": 3282
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 5.487923484608629e-05,
+      "loss": 2.1242,
+      "step": 3285
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 5.469964578162288e-05,
+      "loss": 2.1423,
+      "step": 3288
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 5.4520240356795725e-05,
+      "loss": 2.114,
+      "step": 3291
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 5.4341019298882656e-05,
+      "loss": 2.1531,
+      "step": 3294
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 5.416198333441423e-05,
+      "loss": 2.1431,
+      "step": 3297
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 5.3983133189170686e-05,
+      "loss": 2.0837,
+      "step": 3300
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 5.380446958817888e-05,
+      "loss": 2.1647,
+      "step": 3303
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 5.362599325570945e-05,
+      "loss": 2.1104,
+      "step": 3306
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 5.344770491527402e-05,
+      "loss": 2.11,
+      "step": 3309
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 5.3269605289621947e-05,
+      "loss": 2.1962,
+      "step": 3312
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 5.309169510073777e-05,
+      "loss": 2.213,
+      "step": 3315
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 5.291397506983786e-05,
+      "loss": 2.0556,
+      "step": 3318
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 5.273644591736793e-05,
+      "loss": 2.2094,
+      "step": 3321
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 5.2559108362999796e-05,
+      "loss": 2.0953,
+      "step": 3324
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 5.238196312562851e-05,
+      "loss": 2.2436,
+      "step": 3327
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 5.220501092336966e-05,
+      "loss": 2.1752,
+      "step": 3330
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 5.2028252473556226e-05,
+      "loss": 2.1757,
+      "step": 3333
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 5.1851688492735705e-05,
+      "loss": 2.193,
+      "step": 3336
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 5.167531969666735e-05,
+      "loss": 2.1306,
+      "step": 3339
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 5.149914680031909e-05,
+      "loss": 2.1043,
+      "step": 3342
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 5.132317051786468e-05,
+      "loss": 2.1032,
+      "step": 3345
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 5.114739156268094e-05,
+      "loss": 2.1205,
+      "step": 3348
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 5.097181064734475e-05,
+      "loss": 2.1494,
+      "step": 3351
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 5.0796428483630074e-05,
+      "loss": 2.1382,
+      "step": 3354
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 5.062124578250529e-05,
+      "loss": 2.0946,
+      "step": 3357
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 5.04462632541301e-05,
+      "loss": 2.0986,
+      "step": 3360
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 5.027148160785273e-05,
+      "loss": 2.1342,
+      "step": 3363
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 5.009690155220715e-05,
+      "loss": 2.1049,
+      "step": 3366
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.992252379491012e-05,
+      "loss": 2.2269,
+      "step": 3369
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.974834904285822e-05,
+      "loss": 2.1275,
+      "step": 3372
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.957437800212512e-05,
+      "loss": 2.0984,
+      "step": 3375
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.940061137795876e-05,
+      "loss": 2.1488,
+      "step": 3378
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.9227049874778306e-05,
+      "loss": 2.1061,
+      "step": 3381
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.905369419617137e-05,
+      "loss": 2.2105,
+      "step": 3384
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.888054504489142e-05,
+      "loss": 2.1669,
+      "step": 3387
+    },
+    {
+      "epoch": 11.0,
+      "eval_accuracy": 0.6398645281002108,
+      "eval_loss": 2.068744421005249,
+      "eval_runtime": 16.3144,
+      "eval_samples_per_second": 134.666,
+      "eval_steps_per_second": 67.364,
+      "step": 3388
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.870760312285445e-05,
+      "loss": 2.1265,
+      "step": 3390
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.853486913113644e-05,
+      "loss": 2.2385,
+      "step": 3393
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.836234376997056e-05,
+      "loss": 2.0393,
+      "step": 3396
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.8190027738744134e-05,
+      "loss": 2.0961,
+      "step": 3399
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.801792173599586e-05,
+      "loss": 2.0916,
+      "step": 3402
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.784602645941314e-05,
+      "loss": 2.1191,
+      "step": 3405
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.7674342605829094e-05,
+      "loss": 2.0921,
+      "step": 3408
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.7502870871219675e-05,
+      "loss": 2.0888,
+      "step": 3411
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.7331611950701096e-05,
+      "loss": 2.1485,
+      "step": 3414
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.716056653852672e-05,
+      "loss": 2.1266,
+      "step": 3417
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.698973532808443e-05,
+      "loss": 2.1702,
+      "step": 3420
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.6819119011893805e-05,
+      "loss": 2.1296,
+      "step": 3423
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.664871828160331e-05,
+      "loss": 2.0754,
+      "step": 3426
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.647853382798736e-05,
+      "loss": 2.1102,
+      "step": 3429
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.630856634094366e-05,
+      "loss": 2.1222,
+      "step": 3432
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.613881650949044e-05,
+      "loss": 2.1703,
+      "step": 3435
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.596928502176349e-05,
+      "loss": 2.1288,
+      "step": 3438
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.579997256501355e-05,
+      "loss": 2.0522,
+      "step": 3441
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.563087982560345e-05,
+      "loss": 2.1225,
+      "step": 3444
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.546200748900525e-05,
+      "loss": 2.068,
+      "step": 3447
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.529335623979757e-05,
+      "loss": 2.0433,
+      "step": 3450
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.512492676166283e-05,
+      "loss": 2.0679,
+      "step": 3453
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.49567197373844e-05,
+      "loss": 2.1568,
+      "step": 3456
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.478873584884378e-05,
+      "loss": 2.0446,
+      "step": 3459
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.4620975777018034e-05,
+      "loss": 2.1532,
+      "step": 3462
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.44534402019769e-05,
+      "loss": 2.1659,
+      "step": 3465
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.428612980287996e-05,
+      "loss": 2.1492,
+      "step": 3468
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.411904525797408e-05,
+      "loss": 2.1019,
+      "step": 3471
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.395218724459047e-05,
+      "loss": 2.0283,
+      "step": 3474
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.3785556439142005e-05,
+      "loss": 2.1538,
+      "step": 3477
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.361915351712059e-05,
+      "loss": 2.1303,
+      "step": 3480
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.345297915309432e-05,
+      "loss": 2.0312,
+      "step": 3483
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.3287034020704684e-05,
+      "loss": 2.1296,
+      "step": 3486
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.3121318792663914e-05,
+      "loss": 2.1551,
+      "step": 3489
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.295583414075234e-05,
+      "loss": 2.1197,
+      "step": 3492
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.279058073581544e-05,
+      "loss": 2.2175,
+      "step": 3495
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.2625559247761394e-05,
+      "loss": 2.1445,
+      "step": 3498
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.246077034555819e-05,
+      "loss": 2.1581,
+      "step": 3501
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.229621469723091e-05,
+      "loss": 2.0796,
+      "step": 3504
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.2131892969859054e-05,
+      "loss": 2.1374,
+      "step": 3507
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.196780582957396e-05,
+      "loss": 2.1188,
+      "step": 3510
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.180395394155584e-05,
+      "loss": 2.0809,
+      "step": 3513
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.1640337970031384e-05,
+      "loss": 2.1211,
+      "step": 3516
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.1476958578270783e-05,
+      "loss": 2.1566,
+      "step": 3519
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.1313816428585316e-05,
+      "loss": 2.1824,
+      "step": 3522
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.1150912182324396e-05,
+      "loss": 2.1873,
+      "step": 3525
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.098824649987304e-05,
+      "loss": 2.1329,
+      "step": 3528
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.0825820040649246e-05,
+      "loss": 2.1283,
+      "step": 3531
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.06636334631012e-05,
+      "loss": 2.1378,
+      "step": 3534
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.0501687424704613e-05,
+      "loss": 2.1189,
+      "step": 3537
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.033998258196019e-05,
+      "loss": 2.1541,
+      "step": 3540
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.017851959039075e-05,
+      "loss": 2.1264,
+      "step": 3543
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.001729910453872e-05,
+      "loss": 2.0864,
+      "step": 3546
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 3.985632177796353e-05,
+      "loss": 2.0457,
+      "step": 3549
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 3.9695588263238847e-05,
+      "loss": 2.1076,
+      "step": 3552
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 3.953509921194991e-05,
+      "loss": 2.1368,
+      "step": 3555
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3.9374855274691035e-05,
+      "loss": 2.1462,
+      "step": 3558
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3.921485710106283e-05,
+      "loss": 2.1003,
+      "step": 3561
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3.9055105339669595e-05,
+      "loss": 2.1287,
+      "step": 3564
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 3.889560063811679e-05,
+      "loss": 2.144,
+      "step": 3567
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3.873634364300835e-05,
+      "loss": 2.1544,
+      "step": 3570
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3.857733499994397e-05,
+      "loss": 2.0732,
+      "step": 3573
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3.841857535351657e-05,
+      "loss": 2.1219,
+      "step": 3576
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3.82600653473098e-05,
+      "loss": 2.1629,
+      "step": 3579
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3.810180562389519e-05,
+      "loss": 2.1042,
+      "step": 3582
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3.794379682482965e-05,
+      "loss": 2.1244,
+      "step": 3585
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3.7786039590653076e-05,
+      "loss": 2.18,
+      "step": 3588
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3.762853456088538e-05,
+      "loss": 2.1475,
+      "step": 3591
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 3.747128237402409e-05,
+      "loss": 2.0909,
+      "step": 3594
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.7314283667541885e-05,
+      "loss": 2.123,
+      "step": 3597
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3.715753907788374e-05,
+      "loss": 2.121,
+      "step": 3600
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3.700104924046452e-05,
+      "loss": 2.0345,
+      "step": 3603
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3.6844814789666436e-05,
+      "loss": 2.0413,
+      "step": 3606
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3.6688836358836386e-05,
+      "loss": 2.1818,
+      "step": 3609
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3.6533114580283315e-05,
+      "loss": 2.1043,
+      "step": 3612
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3.6377650085275874e-05,
+      "loss": 2.0476,
+      "step": 3615
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3.622244350403965e-05,
+      "loss": 2.124,
+      "step": 3618
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3.6067495465754666e-05,
+      "loss": 2.0856,
+      "step": 3621
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 3.591280659855296e-05,
+      "loss": 2.1257,
+      "step": 3624
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3.575837752951591e-05,
+      "loss": 2.1757,
+      "step": 3627
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3.5604208884671645e-05,
+      "loss": 2.0124,
+      "step": 3630
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.5450301288992596e-05,
+      "loss": 2.0324,
+      "step": 3633
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.529665536639305e-05,
+      "loss": 2.1634,
+      "step": 3636
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.514327173972638e-05,
+      "loss": 2.1465,
+      "step": 3639
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3.4990151030782744e-05,
+      "loss": 2.0668,
+      "step": 3642
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 3.483729386028651e-05,
+      "loss": 2.1991,
+      "step": 3645
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3.468470084789359e-05,
+      "loss": 2.0814,
+      "step": 3648
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3.4532372612189104e-05,
+      "loss": 2.1976,
+      "step": 3651
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3.438030977068487e-05,
+      "loss": 2.1935,
+      "step": 3654
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 3.422851293981676e-05,
+      "loss": 2.1086,
+      "step": 3657
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3.4076982734942296e-05,
+      "loss": 2.1479,
+      "step": 3660
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.392571977033819e-05,
+      "loss": 2.1281,
+      "step": 3663
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3.377472465919784e-05,
+      "loss": 2.1517,
+      "step": 3666
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.3623998013628675e-05,
+      "loss": 2.1178,
+      "step": 3669
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3.347354044464997e-05,
+      "loss": 2.0988,
+      "step": 3672
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3.332335256219012e-05,
+      "loss": 2.2034,
+      "step": 3675
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.317343497508424e-05,
+      "loss": 2.1123,
+      "step": 3678
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3.302378829107178e-05,
+      "loss": 2.1258,
+      "step": 3681
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3.2874413116794e-05,
+      "loss": 2.1263,
+      "step": 3684
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 3.2725310057791456e-05,
+      "loss": 2.0565,
+      "step": 3687
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3.2576479718501584e-05,
+      "loss": 2.0825,
+      "step": 3690
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3.242792270225635e-05,
+      "loss": 2.1262,
+      "step": 3693
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3.227963961127961e-05,
+      "loss": 2.0983,
+      "step": 3696
+    },
+    {
+      "epoch": 12.0,
+      "eval_accuracy": 0.6422703241176929,
+      "eval_loss": 2.062872886657715,
+      "eval_runtime": 16.2609,
+      "eval_samples_per_second": 135.109,
+      "eval_steps_per_second": 67.585,
+      "step": 3696
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 3.213163104668485e-05,
+      "loss": 2.0955,
+      "step": 3699
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 3.19838976084727e-05,
+      "loss": 2.1043,
+      "step": 3702
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 3.18364398955284e-05,
+      "loss": 2.1251,
+      "step": 3705
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 3.168925850561943e-05,
+      "loss": 2.1206,
+      "step": 3708
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 3.154235403539323e-05,
+      "loss": 2.0734,
+      "step": 3711
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 3.1395727080374505e-05,
+      "loss": 2.178,
+      "step": 3714
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 3.12493782349631e-05,
+      "loss": 2.0629,
+      "step": 3717
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 3.110330809243134e-05,
+      "loss": 2.1535,
+      "step": 3720
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 3.095751724492185e-05,
+      "loss": 2.1331,
+      "step": 3723
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 3.081200628344494e-05,
+      "loss": 2.0902,
+      "step": 3726
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 3.066677579787631e-05,
+      "loss": 2.0434,
+      "step": 3729
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 3.0521826376954755e-05,
+      "loss": 2.1005,
+      "step": 3732
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 3.0377158608279655e-05,
+      "loss": 2.1764,
+      "step": 3735
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 3.0232773078308517e-05,
+      "loss": 2.1099,
+      "step": 3738
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 3.0088670372354877e-05,
+      "loss": 2.1211,
+      "step": 3741
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.99448510745856e-05,
+      "loss": 2.1546,
+      "step": 3744
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 2.9801315768018688e-05,
+      "loss": 2.1664,
+      "step": 3747
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.9658065034520978e-05,
+      "loss": 2.0983,
+      "step": 3750
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.9515099454805663e-05,
+      "loss": 2.0519,
+      "step": 3753
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.93724196084299e-05,
+      "loss": 2.1333,
+      "step": 3756
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 2.923002607379265e-05,
+      "loss": 2.0304,
+      "step": 3759
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.9087919428132114e-05,
+      "loss": 2.1549,
+      "step": 3762
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 2.8946100247523533e-05,
+      "loss": 2.1191,
+      "step": 3765
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.8804569106876832e-05,
+      "loss": 2.1154,
+      "step": 3768
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 2.8663326579934292e-05,
+      "loss": 2.175,
+      "step": 3771
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.8522373239268152e-05,
+      "loss": 2.1378,
+      "step": 3774
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 2.8381709656278333e-05,
+      "loss": 2.0927,
+      "step": 3777
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.8241336401190222e-05,
+      "loss": 2.1146,
+      "step": 3780
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 2.810125404305216e-05,
+      "loss": 2.0147,
+      "step": 3783
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.796146314973325e-05,
+      "loss": 2.1068,
+      "step": 3786
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.7821964287921197e-05,
+      "loss": 2.1693,
+      "step": 3789
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 2.7682758023119694e-05,
+      "loss": 2.1336,
+      "step": 3792
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.7543844919646323e-05,
+      "loss": 2.0793,
+      "step": 3795
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 2.740522554063033e-05,
+      "loss": 2.0712,
+      "step": 3798
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.726690044801018e-05,
+      "loss": 2.0706,
+      "step": 3801
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 2.7128870202531343e-05,
+      "loss": 2.0728,
+      "step": 3804
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.6991135363744068e-05,
+      "loss": 2.1108,
+      "step": 3807
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 2.6853696490001112e-05,
+      "loss": 2.104,
+      "step": 3810
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.6716554138455353e-05,
+      "loss": 2.0752,
+      "step": 3813
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 2.6579708865057694e-05,
+      "loss": 2.154,
+      "step": 3816
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.6443161224554704e-05,
+      "loss": 2.0717,
+      "step": 3819
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 2.6306911770486353e-05,
+      "loss": 2.1225,
+      "step": 3822
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 2.6170961055183906e-05,
+      "loss": 2.1377,
+      "step": 3825
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.6035309629767603e-05,
+      "loss": 2.1614,
+      "step": 3828
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 2.5899958044144302e-05,
+      "loss": 2.1486,
+      "step": 3831
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.576490684700542e-05,
+      "loss": 2.1206,
+      "step": 3834
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 2.5630156585824727e-05,
+      "loss": 2.0882,
+      "step": 3837
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 2.5495707806855938e-05,
+      "loss": 2.1787,
+      "step": 3840
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 2.536156105513062e-05,
+      "loss": 2.0932,
+      "step": 3843
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.522771687445612e-05,
+      "loss": 2.1471,
+      "step": 3846
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 2.5094175807413055e-05,
+      "loss": 2.1226,
+      "step": 3849
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 2.4960938395353296e-05,
+      "loss": 2.1666,
+      "step": 3852
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.4828005178397838e-05,
+      "loss": 2.0437,
+      "step": 3855
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.4695376695434448e-05,
+      "loss": 2.0396,
+      "step": 3858
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.456305348411554e-05,
+      "loss": 2.09,
+      "step": 3861
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 2.4431036080856073e-05,
+      "loss": 2.0419,
+      "step": 3864
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.429932502083132e-05,
+      "loss": 2.0626,
+      "step": 3867
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.41679208379746e-05,
+      "loss": 2.1798,
+      "step": 3870
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 2.4036824064975317e-05,
+      "loss": 2.1082,
+      "step": 3873
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.3906035233276614e-05,
+      "loss": 2.0504,
+      "step": 3876
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 2.3775554873073292e-05,
+      "loss": 2.0439,
+      "step": 3879
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 2.3645383513309704e-05,
+      "loss": 2.1104,
+      "step": 3882
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.351552168167761e-05,
+      "loss": 2.088,
+      "step": 3885
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.338596990461388e-05,
+      "loss": 2.0038,
+      "step": 3888
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 2.3256728707298546e-05,
+      "loss": 2.043,
+      "step": 3891
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 2.312779861365263e-05,
+      "loss": 2.0785,
+      "step": 3894
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.299918014633592e-05,
+      "loss": 2.1406,
+      "step": 3897
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 2.2870873826744988e-05,
+      "loss": 2.1155,
+      "step": 3900
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 2.2742880175011028e-05,
+      "loss": 2.1258,
+      "step": 3903
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 2.261519970999768e-05,
+      "loss": 2.1664,
+      "step": 3906
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.248783294929897e-05,
+      "loss": 2.0733,
+      "step": 3909
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.2360780409237294e-05,
+      "loss": 2.135,
+      "step": 3912
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.2234042604861182e-05,
+      "loss": 2.1826,
+      "step": 3915
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 2.2107620049943346e-05,
+      "loss": 2.0611,
+      "step": 3918
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 2.1981513256978458e-05,
+      "loss": 2.0883,
+      "step": 3921
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.185572273718124e-05,
+      "loss": 2.0715,
+      "step": 3924
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 2.1730249000484203e-05,
+      "loss": 2.0608,
+      "step": 3927
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.1605092555535712e-05,
+      "loss": 2.0845,
+      "step": 3930
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 2.14802539096979e-05,
+      "loss": 2.0893,
+      "step": 3933
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.1355733569044635e-05,
+      "loss": 2.0895,
+      "step": 3936
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.1231532038359326e-05,
+      "loss": 2.1653,
+      "step": 3939
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 2.11076498211331e-05,
+      "loss": 2.0523,
+      "step": 3942
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.098408741956256e-05,
+      "loss": 2.056,
+      "step": 3945
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.086084533454784e-05,
+      "loss": 2.1268,
+      "step": 3948
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.0737924065690606e-05,
+      "loss": 2.0818,
+      "step": 3951
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 2.0615324111292013e-05,
+      "loss": 2.038,
+      "step": 3954
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.0493045968350567e-05,
+      "loss": 2.124,
+      "step": 3957
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.0371090132560322e-05,
+      "loss": 2.0462,
+      "step": 3960
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.0249457098308665e-05,
+      "loss": 2.1401,
+      "step": 3963
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.012814735867442e-05,
+      "loss": 2.1156,
+      "step": 3966
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.0007161405425866e-05,
+      "loss": 2.1056,
+      "step": 3969
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 1.9886499729018737e-05,
+      "loss": 2.093,
+      "step": 3972
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 1.9766162818594114e-05,
+      "loss": 2.1043,
+      "step": 3975
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 1.9646151161976556e-05,
+      "loss": 2.0714,
+      "step": 3978
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 1.9526465245672187e-05,
+      "loss": 2.1165,
+      "step": 3981
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 1.9407105554866557e-05,
+      "loss": 2.0959,
+      "step": 3984
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 1.92880725734227e-05,
+      "loss": 2.1551,
+      "step": 3987
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 1.9169366783879428e-05,
+      "loss": 2.0905,
+      "step": 3990
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 1.9050988667448977e-05,
+      "loss": 2.1042,
+      "step": 3993
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 1.8932938704015314e-05,
+      "loss": 2.1607,
+      "step": 3996
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 1.8815217372132198e-05,
+      "loss": 2.0262,
+      "step": 3999
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 1.8697825149021086e-05,
+      "loss": 2.1215,
+      "step": 4002
+    },
+    {
+      "epoch": 13.0,
+      "eval_accuracy": 0.6475752433152033,
+      "eval_loss": 2.025885820388794,
+      "eval_runtime": 16.3422,
+      "eval_samples_per_second": 134.437,
+      "eval_steps_per_second": 67.249,
+      "step": 4004
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 1.8580762510569295e-05,
+      "loss": 2.1667,
+      "step": 4005
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 1.846402993132811e-05,
+      "loss": 2.1035,
+      "step": 4008
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 1.8347627884510832e-05,
+      "loss": 2.1239,
+      "step": 4011
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 1.823155684199074e-05,
+      "loss": 2.1166,
+      "step": 4014
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 1.8115817274299396e-05,
+      "loss": 2.1101,
+      "step": 4017
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 1.800040965062455e-05,
+      "loss": 2.0607,
+      "step": 4020
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 1.7885334438808287e-05,
+      "loss": 2.0964,
+      "step": 4023
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 1.777059210534524e-05,
+      "loss": 2.1132,
+      "step": 4026
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 1.7656183115380577e-05,
+      "loss": 2.0793,
+      "step": 4029
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 1.754210793270812e-05,
+      "loss": 2.1091,
+      "step": 4032
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 1.742836701976849e-05,
+      "loss": 2.0793,
+      "step": 4035
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 1.7314960837647297e-05,
+      "loss": 2.0164,
+      "step": 4038
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 1.7201889846073183e-05,
+      "loss": 2.0697,
+      "step": 4041
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 1.7089154503415895e-05,
+      "loss": 2.0631,
+      "step": 4044
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 1.697675526668473e-05,
+      "loss": 2.0714,
+      "step": 4047
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 1.6864692591526278e-05,
+      "loss": 2.1084,
+      "step": 4050
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 1.6752966932222826e-05,
+      "loss": 2.1333,
+      "step": 4053
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 1.664157874169049e-05,
+      "loss": 2.1206,
+      "step": 4056
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 1.6530528471477326e-05,
+      "loss": 2.1727,
+      "step": 4059
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 1.6419816571761482e-05,
+      "loss": 2.1396,
+      "step": 4062
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 1.6309443491349475e-05,
+      "loss": 2.0632,
+      "step": 4065
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 1.6199409677674314e-05,
+      "loss": 2.0968,
+      "step": 4068
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 1.6089715576793584e-05,
+      "loss": 2.0538,
+      "step": 4071
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 1.5980361633387853e-05,
+      "loss": 2.1114,
+      "step": 4074
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 1.587134829075867e-05,
+      "loss": 2.1656,
+      "step": 4077
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 1.576267599082686e-05,
+      "loss": 2.0781,
+      "step": 4080
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 1.5654345174130756e-05,
+      "loss": 2.1749,
+      "step": 4083
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 1.5546356279824382e-05,
+      "loss": 2.0654,
+      "step": 4086
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 1.5438709745675606e-05,
+      "loss": 2.0904,
+      "step": 4089
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 1.5331406008064475e-05,
+      "loss": 2.0368,
+      "step": 4092
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 1.522444550198141e-05,
+      "loss": 2.0759,
+      "step": 4095
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 1.511782866102539e-05,
+      "loss": 2.0462,
+      "step": 4098
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 1.5011555917402265e-05,
+      "loss": 2.0873,
+      "step": 4101
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 1.4905627701923009e-05,
+      "loss": 2.0913,
+      "step": 4104
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 1.480004444400187e-05,
+      "loss": 2.0516,
+      "step": 4107
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 1.4694806571654696e-05,
+      "loss": 2.1136,
+      "step": 4110
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 1.4589914511497305e-05,
+      "loss": 2.1294,
+      "step": 4113
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 1.4485368688743527e-05,
+      "loss": 2.068,
+      "step": 4116
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 1.4381169527203719e-05,
+      "loss": 2.0402,
+      "step": 4119
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 1.4277317449282834e-05,
+      "loss": 2.048,
+      "step": 4122
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 1.4173812875978886e-05,
+      "loss": 2.0875,
+      "step": 4125
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 1.407065622688113e-05,
+      "loss": 2.1008,
+      "step": 4128
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 1.3967847920168386e-05,
+      "loss": 2.1113,
+      "step": 4131
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 1.386538837260738e-05,
+      "loss": 2.0277,
+      "step": 4134
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 1.376327799955105e-05,
+      "loss": 2.1696,
+      "step": 4137
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 1.3661517214936782e-05,
+      "loss": 2.1531,
+      "step": 4140
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 1.356010643128487e-05,
+      "loss": 2.1222,
+      "step": 4143
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 1.345904605969669e-05,
+      "loss": 2.0299,
+      "step": 4146
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 1.3358336509853131e-05,
+      "loss": 2.1065,
+      "step": 4149
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 1.3257978190012931e-05,
+      "loss": 2.0989,
+      "step": 4152
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 1.3157971507011036e-05,
+      "loss": 2.0679,
+      "step": 4155
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 1.3058316866256826e-05,
+      "loss": 2.1828,
+      "step": 4158
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 1.295901467173265e-05,
+      "loss": 2.057,
+      "step": 4161
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 1.2860065325992066e-05,
+      "loss": 2.0964,
+      "step": 4164
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 1.2761469230158208e-05,
+      "loss": 2.1366,
+      "step": 4167
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 1.2663226783922266e-05,
+      "loss": 2.0889,
+      "step": 4170
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 1.2565338385541792e-05,
+      "loss": 2.0918,
+      "step": 4173
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 1.2467804431839037e-05,
+      "loss": 2.1852,
+      "step": 4176
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 1.2370625318199414e-05,
+      "loss": 2.0561,
+      "step": 4179
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 1.2273801438569932e-05,
+      "loss": 2.0864,
+      "step": 4182
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 1.2177333185457474e-05,
+      "loss": 2.125,
+      "step": 4185
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 1.2081220949927252e-05,
+      "loss": 2.0829,
+      "step": 4188
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 1.1985465121601392e-05,
+      "loss": 2.0291,
+      "step": 4191
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 1.189006608865707e-05,
+      "loss": 2.1237,
+      "step": 4194
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 1.1795024237825092e-05,
+      "loss": 2.1423,
+      "step": 4197
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 1.1700339954388384e-05,
+      "loss": 2.1,
+      "step": 4200
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 1.1606013622180278e-05,
+      "loss": 2.0367,
+      "step": 4203
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 1.1512045623583068e-05,
+      "loss": 2.0967,
+      "step": 4206
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 1.1418436339526429e-05,
+      "loss": 2.0585,
+      "step": 4209
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 1.1325186149485889e-05,
+      "loss": 2.1754,
+      "step": 4212
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 1.1232295431481222e-05,
+      "loss": 2.0563,
+      "step": 4215
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 1.1139764562075017e-05,
+      "loss": 2.1228,
+      "step": 4218
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 1.104759391637108e-05,
+      "loss": 2.013,
+      "step": 4221
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 1.0955783868012892e-05,
+      "loss": 2.1053,
+      "step": 4224
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 1.0864334789182218e-05,
+      "loss": 2.0723,
+      "step": 4227
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 1.0773247050597468e-05,
+      "loss": 2.142,
+      "step": 4230
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 1.0682521021512249e-05,
+      "loss": 2.0928,
+      "step": 4233
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 1.0592157069713826e-05,
+      "loss": 2.0371,
+      "step": 4236
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 1.0502155561521766e-05,
+      "loss": 2.1179,
+      "step": 4239
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 1.0412516861786236e-05,
+      "loss": 2.0816,
+      "step": 4242
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 1.032324133388668e-05,
+      "loss": 2.0207,
+      "step": 4245
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 1.0234329339730398e-05,
+      "loss": 2.0805,
+      "step": 4248
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 1.0145781239750863e-05,
+      "loss": 2.1022,
+      "step": 4251
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 1.0057597392906414e-05,
+      "loss": 2.1438,
+      "step": 4254
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 9.969778156678854e-06,
+      "loss": 2.097,
+      "step": 4257
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 9.88232388707182e-06,
+      "loss": 2.0942,
+      "step": 4260
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 9.795234938609466e-06,
+      "loss": 2.0325,
+      "step": 4263
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 9.708511664335029e-06,
+      "loss": 2.1505,
+      "step": 4266
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 9.62215441580936e-06,
+      "loss": 2.05,
+      "step": 4269
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 9.536163543109488e-06,
+      "loss": 2.0526,
+      "step": 4272
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 9.450539394827185e-06,
+      "loss": 2.0956,
+      "step": 4275
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 9.365282318067681e-06,
+      "loss": 2.0701,
+      "step": 4278
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 9.280392658448078e-06,
+      "loss": 2.1114,
+      "step": 4281
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 9.19587076009607e-06,
+      "loss": 2.0833,
+      "step": 4284
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 9.11171696564853e-06,
+      "loss": 2.1273,
+      "step": 4287
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 9.027931616250063e-06,
+      "loss": 2.0479,
+      "step": 4290
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 8.94451505155165e-06,
+      "loss": 2.0574,
+      "step": 4293
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 8.861467609709373e-06,
+      "loss": 2.0643,
+      "step": 4296
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 8.778789627382833e-06,
+      "loss": 2.1623,
+      "step": 4299
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 8.696481439734017e-06,
+      "loss": 2.0858,
+      "step": 4302
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 8.614543380425766e-06,
+      "loss": 2.1034,
+      "step": 4305
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 8.532975781620512e-06,
+      "loss": 2.1097,
+      "step": 4308
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 8.451778973978874e-06,
+      "loss": 2.1255,
+      "step": 4311
+    },
+    {
+      "epoch": 14.0,
+      "eval_accuracy": 0.6460723583804651,
+      "eval_loss": 2.037827730178833,
+      "eval_runtime": 16.2998,
+      "eval_samples_per_second": 134.787,
+      "eval_steps_per_second": 67.424,
+      "step": 4312
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 8.370953286658389e-06,
+      "loss": 2.0412,
+      "step": 4314
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 8.290499047312106e-06,
+      "loss": 2.1136,
+      "step": 4317
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 8.210416582087332e-06,
+      "loss": 2.1369,
+      "step": 4320
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 8.130706215624195e-06,
+      "loss": 2.0917,
+      "step": 4323
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 8.051368271054493e-06,
+      "loss": 2.1272,
+      "step": 4326
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 7.972403070000222e-06,
+      "loss": 2.1439,
+      "step": 4329
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 7.893810932572333e-06,
+      "loss": 2.0715,
+      "step": 4332
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 7.815592177369502e-06,
+      "loss": 2.0255,
+      "step": 4335
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 7.737747121476757e-06,
+      "loss": 2.0631,
+      "step": 4338
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 7.66027608046419e-06,
+      "loss": 2.1339,
+      "step": 4341
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 7.58317936838574e-06,
+      "loss": 2.0685,
+      "step": 4344
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 7.506457297777847e-06,
+      "loss": 2.0141,
+      "step": 4347
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 7.4301101796582225e-06,
+      "loss": 2.0722,
+      "step": 4350
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 7.354138323524617e-06,
+      "loss": 2.0758,
+      "step": 4353
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 7.278542037353542e-06,
+      "loss": 2.0993,
+      "step": 4356
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 7.203321627598947e-06,
+      "loss": 2.114,
+      "step": 4359
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 7.128477399191136e-06,
+      "loss": 2.114,
+      "step": 4362
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 7.054009655535354e-06,
+      "loss": 2.0214,
+      "step": 4365
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 6.979918698510701e-06,
+      "loss": 2.0729,
+      "step": 4368
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 6.906204828468821e-06,
+      "loss": 1.9927,
+      "step": 4371
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 6.832868344232757e-06,
+      "loss": 2.0514,
+      "step": 4374
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 6.759909543095632e-06,
+      "loss": 2.1031,
+      "step": 4377
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 6.687328720819552e-06,
+      "loss": 2.0984,
+      "step": 4380
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 6.615126171634367e-06,
+      "loss": 2.0636,
+      "step": 4383
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 6.543302188236445e-06,
+      "loss": 2.0285,
+      "step": 4386
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 6.471857061787501e-06,
+      "loss": 2.0266,
+      "step": 4389
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 6.400791081913538e-06,
+      "loss": 2.1057,
+      "step": 4392
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 6.33010453670343e-06,
+      "loss": 2.0328,
+      "step": 4395
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 6.25979771270796e-06,
+      "loss": 2.0908,
+      "step": 4398
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 6.189870894938587e-06,
+      "loss": 2.067,
+      "step": 4401
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 6.120324366866281e-06,
+      "loss": 2.0099,
+      "step": 4404
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 6.051158410420355e-06,
+      "loss": 2.1048,
+      "step": 4407
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 5.98237330598741e-06,
+      "loss": 2.0286,
+      "step": 4410
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 5.91396933241013e-06,
+      "loss": 2.0796,
+      "step": 4413
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 5.845946766986099e-06,
+      "loss": 2.0391,
+      "step": 4416
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 5.778305885466828e-06,
+      "loss": 2.0937,
+      "step": 4419
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 5.711046962056488e-06,
+      "loss": 2.1198,
+      "step": 4422
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 5.644170269410853e-06,
+      "loss": 2.1277,
+      "step": 4425
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 5.577676078636251e-06,
+      "loss": 2.1039,
+      "step": 4428
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 5.511564659288404e-06,
+      "loss": 2.0803,
+      "step": 4431
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 5.445836279371308e-06,
+      "loss": 2.0594,
+      "step": 4434
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 5.380491205336202e-06,
+      "loss": 2.131,
+      "step": 4437
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 5.315529702080491e-06,
+      "loss": 2.0799,
+      "step": 4440
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 5.250952032946643e-06,
+      "loss": 2.0678,
+      "step": 4443
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 5.186758459721075e-06,
+      "loss": 2.115,
+      "step": 4446
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 5.122949242633279e-06,
+      "loss": 2.0485,
+      "step": 4449
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 5.059524640354496e-06,
+      "loss": 2.1301,
+      "step": 4452
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 4.996484909996868e-06,
+      "loss": 2.0767,
+      "step": 4455
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 4.933830307112353e-06,
+      "loss": 2.1214,
+      "step": 4458
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 4.871561085691634e-06,
+      "loss": 2.0206,
+      "step": 4461
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 4.8096774981631235e-06,
+      "loss": 2.1629,
+      "step": 4464
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 4.7481797953919605e-06,
+      "loss": 2.0928,
+      "step": 4467
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 4.687068226679004e-06,
+      "loss": 2.1403,
+      "step": 4470
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 4.6263430397597395e-06,
+      "loss": 2.0604,
+      "step": 4473
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 4.566004480803332e-06,
+      "loss": 2.1267,
+      "step": 4476
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 4.5060527944116856e-06,
+      "loss": 2.1578,
+      "step": 4479
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 4.446488223618306e-06,
+      "loss": 2.1332,
+      "step": 4482
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 4.387311009887463e-06,
+      "loss": 1.9903,
+      "step": 4485
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 4.328521393113149e-06,
+      "loss": 2.2028,
+      "step": 4488
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 4.270119611618073e-06,
+      "loss": 2.0242,
+      "step": 4491
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 4.21210590215273e-06,
+      "loss": 2.0528,
+      "step": 4494
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 4.1544804998944756e-06,
+      "loss": 2.1233,
+      "step": 4497
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 4.097243638446502e-06,
+      "loss": 2.1108,
+      "step": 4500
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 4.040395549836928e-06,
+      "loss": 2.0646,
+      "step": 4503
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.983936464517901e-06,
+      "loss": 2.1225,
+      "step": 4506
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.9278666113645615e-06,
+      "loss": 2.0553,
+      "step": 4509
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.872186217674167e-06,
+      "loss": 2.1065,
+      "step": 4512
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.816895509165252e-06,
+      "loss": 2.0563,
+      "step": 4515
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.7619947099765353e-06,
+      "loss": 2.1165,
+      "step": 4518
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.707484042666198e-06,
+      "loss": 2.0359,
+      "step": 4521
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.6533637282108347e-06,
+      "loss": 2.0946,
+      "step": 4524
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.599633986004669e-06,
+      "loss": 2.0738,
+      "step": 4527
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.5462950338585597e-06,
+      "loss": 2.0815,
+      "step": 4530
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.4933470879992104e-06,
+      "loss": 2.05,
+      "step": 4533
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.440790363068247e-06,
+      "loss": 2.0606,
+      "step": 4536
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.3886250721213544e-06,
+      "loss": 2.1675,
+      "step": 4539
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.3368514266273964e-06,
+      "loss": 2.0499,
+      "step": 4542
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.2854696364675974e-06,
+      "loss": 2.1578,
+      "step": 4545
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.2344799099346733e-06,
+      "loss": 2.0859,
+      "step": 4548
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.1838824537319456e-06,
+      "loss": 2.1324,
+      "step": 4551
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.1336774729725736e-06,
+      "loss": 2.117,
+      "step": 4554
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.0838651711787013e-06,
+      "loss": 2.0503,
+      "step": 4557
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.034445750280579e-06,
+      "loss": 2.0449,
+      "step": 4560
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 2.985419410615831e-06,
+      "loss": 2.1285,
+      "step": 4563
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 2.9367863509285775e-06,
+      "loss": 2.0391,
+      "step": 4566
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 2.8885467683686497e-06,
+      "loss": 2.0469,
+      "step": 4569
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 2.840700858490786e-06,
+      "loss": 2.1386,
+      "step": 4572
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 2.7932488152538794e-06,
+      "loss": 2.1428,
+      "step": 4575
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 2.7461908310201123e-06,
+      "loss": 2.0901,
+      "step": 4578
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 2.6995270965542554e-06,
+      "loss": 2.0583,
+      "step": 4581
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 2.653257801022835e-06,
+      "loss": 2.0573,
+      "step": 4584
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 2.607383131993424e-06,
+      "loss": 2.035,
+      "step": 4587
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 2.561903275433797e-06,
+      "loss": 2.1873,
+      "step": 4590
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 2.5168184157113084e-06,
+      "loss": 2.0505,
+      "step": 4593
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 2.472128735591983e-06,
+      "loss": 2.0236,
+      "step": 4596
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 2.4278344162398935e-06,
+      "loss": 2.022,
+      "step": 4599
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 2.3839356372164056e-06,
+      "loss": 1.9994,
+      "step": 4602
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 2.3404325764794012e-06,
+      "loss": 1.9757,
+      "step": 4605
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 2.2973254103826e-06,
+      "loss": 2.0497,
+      "step": 4608
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 2.254614313674863e-06,
+      "loss": 2.1178,
+      "step": 4611
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 2.2122994594994227e-06,
+      "loss": 2.1794,
+      "step": 4614
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 2.1703810193932307e-06,
+      "loss": 2.0673,
+      "step": 4617
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 2.1288591632862343e-06,
+      "loss": 2.1751,
+      "step": 4620
+    },
+    {
+      "epoch": 15.0,
+      "eval_accuracy": 0.6458123953098828,
+      "eval_loss": 2.0256659984588623,
+      "eval_runtime": 16.3182,
+      "eval_samples_per_second": 134.635,
+      "eval_steps_per_second": 67.348,
+      "step": 4620
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 2.087734059500712e-06,
+      "loss": 2.0922,
+      "step": 4623
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 2.0470058747505516e-06,
+      "loss": 2.1413,
+      "step": 4626
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 2.006674774140638e-06,
+      "loss": 2.0214,
+      "step": 4629
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 1.9667409211661437e-06,
+      "loss": 2.1027,
+      "step": 4632
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 1.9272044777118524e-06,
+      "loss": 2.0475,
+      "step": 4635
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 1.8880656040514921e-06,
+      "loss": 2.0842,
+      "step": 4638
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 1.8493244588471793e-06,
+      "loss": 2.0245,
+      "step": 4641
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 1.8109811991486646e-06,
+      "loss": 2.0969,
+      "step": 4644
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 1.7730359803927343e-06,
+      "loss": 2.1304,
+      "step": 4647
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 1.735488956402631e-06,
+      "loss": 2.001,
+      "step": 4650
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 1.698340279387356e-06,
+      "loss": 2.1577,
+      "step": 4653
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 1.6615900999410683e-06,
+      "loss": 2.058,
+      "step": 4656
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 1.6252385670425307e-06,
+      "loss": 2.0714,
+      "step": 4659
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 1.589285828054421e-06,
+      "loss": 2.0709,
+      "step": 4662
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 1.5537320287227764e-06,
+      "loss": 2.0754,
+      "step": 4665
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 1.5185773131764502e-06,
+      "loss": 2.1037,
+      "step": 4668
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 1.4838218239264456e-06,
+      "loss": 2.1344,
+      "step": 4671
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 1.4494657018653823e-06,
+      "loss": 2.0933,
+      "step": 4674
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 1.4155090862668863e-06,
+      "loss": 2.0771,
+      "step": 4677
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 1.3819521147851123e-06,
+      "loss": 2.0867,
+      "step": 4680
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 1.3487949234540664e-06,
+      "loss": 2.1519,
+      "step": 4683
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 1.3160376466871739e-06,
+      "loss": 2.0861,
+      "step": 4686
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 1.2836804172766449e-06,
+      "loss": 2.0644,
+      "step": 4689
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 1.2517233663929651e-06,
+      "loss": 2.1133,
+      "step": 4692
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 1.2201666235843735e-06,
+      "loss": 2.0873,
+      "step": 4695
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 1.18901031677634e-06,
+      "loss": 2.1127,
+      "step": 4698
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 1.1582545722710225e-06,
+      "loss": 2.0917,
+      "step": 4701
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 1.1278995147467885e-06,
+      "loss": 2.0339,
+      "step": 4704
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 1.0979452672576718e-06,
+      "loss": 2.0649,
+      "step": 4707
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 1.0683919512329166e-06,
+      "loss": 2.0376,
+      "step": 4710
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 1.0392396864764231e-06,
+      "loss": 2.0764,
+      "step": 4713
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 1.0104885911663474e-06,
+      "loss": 2.1247,
+      "step": 4716
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 9.821387818545358e-07,
+      "loss": 2.1067,
+      "step": 4719
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 9.54190373466113e-07,
+      "loss": 2.0267,
+      "step": 4722
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 9.266434792989942e-07,
+      "loss": 2.0377,
+      "step": 4725
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 8.994982110234307e-07,
+      "loss": 2.0895,
+      "step": 4728
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 8.727546786815421e-07,
+      "loss": 2.027,
+      "step": 4731
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 8.464129906868734e-07,
+      "loss": 2.0593,
+      "step": 4734
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 8.204732538239835e-07,
+      "loss": 1.9569,
+      "step": 4737
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 7.949355732479902e-07,
+      "loss": 2.0478,
+      "step": 4740
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 7.698000524841376e-07,
+      "loss": 2.0132,
+      "step": 4743
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 7.450667934273958e-07,
+      "loss": 2.1406,
+      "step": 4746
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 7.207358963420063e-07,
+      "loss": 2.0989,
+      "step": 4749
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 6.968074598611484e-07,
+      "loss": 2.1065,
+      "step": 4752
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 6.732815809864734e-07,
+      "loss": 2.0832,
+      "step": 4755
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 6.501583550877488e-07,
+      "loss": 2.1096,
+      "step": 4758
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 6.274378759024257e-07,
+      "loss": 2.0831,
+      "step": 4761
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 6.051202355353392e-07,
+      "loss": 2.0592,
+      "step": 4764
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 5.832055244582524e-07,
+      "loss": 2.1122,
+      "step": 4767
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 5.616938315095243e-07,
+      "loss": 2.1232,
+      "step": 4770
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 5.405852438937764e-07,
+      "loss": 2.0721,
+      "step": 4773
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 5.198798471814814e-07,
+      "loss": 2.1295,
+      "step": 4776
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 4.995777253086753e-07,
+      "loss": 2.0643,
+      "step": 4779
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 4.796789605765573e-07,
+      "loss": 2.0741,
+      "step": 4782
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 4.601836336512233e-07,
+      "loss": 2.0587,
+      "step": 4785
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 4.4109182356327774e-07,
+      "loss": 2.1124,
+      "step": 4788
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 4.2240360770753327e-07,
+      "loss": 2.1365,
+      "step": 4791
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 4.0411906184273376e-07,
+      "loss": 2.1643,
+      "step": 4794
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.8623826009120955e-07,
+      "loss": 2.066,
+      "step": 4797
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.6876127493854495e-07,
+      "loss": 2.0562,
+      "step": 4800
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.516881772333669e-07,
+      "loss": 2.1947,
+      "step": 4803
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.35019036187012e-07,
+      "loss": 2.0512,
+      "step": 4806
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.187539193732048e-07,
+      "loss": 2.0424,
+      "step": 4809
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.028928927278685e-07,
+      "loss": 2.101,
+      "step": 4812
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 2.874360205488258e-07,
+      "loss": 2.1028,
+      "step": 4815
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 2.723833654954655e-07,
+      "loss": 2.0677,
+      "step": 4818
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 2.577349885886315e-07,
+      "loss": 2.0715,
+      "step": 4821
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 2.434909492102455e-07,
+      "loss": 2.106,
+      "step": 4824
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 2.2965130510310685e-07,
+      "loss": 2.0889,
+      "step": 4827
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 2.1621611237071516e-07,
+      "loss": 2.0497,
+      "step": 4830
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 2.031854254769594e-07,
+      "loss": 2.0797,
+      "step": 4833
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 1.9055929724595134e-07,
+      "loss": 2.0286,
+      "step": 4836
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 1.7833777886175907e-07,
+      "loss": 1.9975,
+      "step": 4839
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 1.66520919868296e-07,
+      "loss": 2.0781,
+      "step": 4842
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 1.5510876816898778e-07,
+      "loss": 2.1324,
+      "step": 4845
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 1.4410137002670575e-07,
+      "loss": 2.0367,
+      "step": 4848
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 1.334987700634893e-07,
+      "loss": 2.1207,
+      "step": 4851
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 1.233010112604016e-07,
+      "loss": 2.1331,
+      "step": 4854
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 1.1350813495737411e-07,
+      "loss": 2.1653,
+      "step": 4857
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 1.0412018085297348e-07,
+      "loss": 2.0604,
+      "step": 4860
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 9.513718700432384e-08,
+      "loss": 2.0798,
+      "step": 4863
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 8.655918982689581e-08,
+      "loss": 2.04,
+      "step": 4866
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 7.838622409436225e-08,
+      "loss": 2.017,
+      "step": 4869
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 7.061832293849823e-08,
+      "loss": 2.0456,
+      "step": 4872
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 6.325551784900352e-08,
+      "loss": 2.1483,
+      "step": 4875
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 5.629783867336924e-08,
+      "loss": 2.0091,
+      "step": 4878
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 4.9745313616822445e-08,
+      "loss": 2.1577,
+      "step": 4881
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 4.3597969242126225e-08,
+      "loss": 2.0558,
+      "step": 4884
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.7855830469535334e-08,
+      "loss": 2.0223,
+      "step": 4887
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.2518920576662945e-08,
+      "loss": 2.0815,
+      "step": 4890
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 2.7587261198414038e-08,
+      "loss": 2.0999,
+      "step": 4893
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 2.3060872326841066e-08,
+      "loss": 2.1187,
+      "step": 4896
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 1.8939772311143967e-08,
+      "loss": 2.0901,
+      "step": 4899
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 1.522397785752583e-08,
+      "loss": 2.1402,
+      "step": 4902
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 1.1913504029159583e-08,
+      "loss": 2.1058,
+      "step": 4905
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 9.008364246121393e-09,
+      "loss": 2.0656,
+      "step": 4908
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 6.508570285346238e-09,
+      "loss": 2.1138,
+      "step": 4911
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 4.414132280550209e-09,
+      "loss": 2.0542,
+      "step": 4914
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 2.7250587222082957e-09,
+      "loss": 1.972,
+      "step": 4917
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 1.4413564575432858e-09,
+      "loss": 2.069,
+      "step": 4920
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 5.630306904369498e-10,
+      "loss": 2.1005,
+      "step": 4923
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 9.008498147444755e-11,
+      "loss": 1.9516,
+      "step": 4926
+    },
+    {
+      "epoch": 16.0,
+      "eval_accuracy": 0.6450172750488208,
+      "eval_loss": 2.037081241607666,
+      "eval_runtime": 16.3187,
+      "eval_samples_per_second": 134.631,
+      "eval_steps_per_second": 67.346,
+      "step": 4928
+    },
+    {
+      "epoch": 16.0,
+      "step": 4928,
+      "total_flos": 1.0508396840353792e+16,
+      "train_loss": 2.4320973860366,
+      "train_runtime": 3195.9564,
+      "train_samples_per_second": 98.68,
+      "train_steps_per_second": 1.542
+    }
+  ],
+  "max_steps": 4928,
+  "num_train_epochs": 16,
+  "total_flos": 1.0508396840353792e+16,
+  "trial_name": null,
+  "trial_params": null
+}