diff --git "a/no_vowels_checkpoint/checkpoint-12300/trainer_state.json" "b/no_vowels_checkpoint/checkpoint-12300/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/no_vowels_checkpoint/checkpoint-12300/trainer_state.json"
@@ -0,0 +1,17610 @@
+{
+  "best_metric": 0.08887978955727616,
+  "best_model_checkpoint": "./wav2vec2-bert-pa_4/checkpoint-12300",
+  "epoch": 4.632070497853431,
+  "eval_steps": 300,
+  "global_step": 12300,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 14.354777336120605,
+      "learning_rate": 6.725854183481302e-08,
+      "loss": 6.9083,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 15.007977485656738,
+      "learning_rate": 1.3451708366962605e-07,
+      "loss": 6.9743,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 14.319838523864746,
+      "learning_rate": 2.0177562550443905e-07,
+      "loss": 7.0948,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 13.708984375,
+      "learning_rate": 2.690341673392521e-07,
+      "loss": 6.8005,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 19.09122657775879,
+      "learning_rate": 3.362927091740651e-07,
+      "loss": 7.0658,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 18.646995544433594,
+      "learning_rate": 4.035512510088781e-07,
+      "loss": 6.9572,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 15.488593101501465,
+      "learning_rate": 4.5735808447672856e-07,
+      "loss": 7.3736,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 16.115514755249023,
+      "learning_rate": 5.246166263115416e-07,
+      "loss": 6.3474,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 18.77143669128418,
+      "learning_rate": 5.918751681463546e-07,
+      "loss": 6.4022,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 21.01273536682129,
+      "learning_rate": 6.591337099811677e-07,
+      "loss": 6.4888,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 20.970500946044922,
+      "learning_rate": 7.263922518159807e-07,
+      "loss": 5.9204,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 20.81243324279785,
+      "learning_rate": 7.936507936507937e-07,
+      "loss": 6.2952,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 24.2125301361084,
+      "learning_rate": 8.609093354856067e-07,
+      "loss": 5.5923,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 34.544246673583984,
+      "learning_rate": 9.281678773204198e-07,
+      "loss": 5.224,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 28.963096618652344,
+      "learning_rate": 9.954264191552328e-07,
+      "loss": 5.2079,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 17.316585540771484,
+      "learning_rate": 1.0626849609900459e-06,
+      "loss": 4.5491,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 16.088727951049805,
+      "learning_rate": 1.129943502824859e-06,
+      "loss": 4.193,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.064581871032715,
+      "learning_rate": 1.1972020446596718e-06,
+      "loss": 3.9464,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 8.114545822143555,
+      "learning_rate": 1.2644605864944848e-06,
+      "loss": 3.903,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.04863977432251,
+      "learning_rate": 1.3317191283292979e-06,
+      "loss": 3.8494,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.059878826141357,
+      "learning_rate": 1.398977670164111e-06,
+      "loss": 3.7306,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.889921188354492,
+      "learning_rate": 1.466236211998924e-06,
+      "loss": 3.6729,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.0156025886535645,
+      "learning_rate": 1.533494753833737e-06,
+      "loss": 3.6025,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 8.51967716217041,
+      "learning_rate": 1.60075329566855e-06,
+      "loss": 3.5673,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.293105363845825,
+      "learning_rate": 1.668011837503363e-06,
+      "loss": 3.497,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.8694326877593994,
+      "learning_rate": 1.7352703793381762e-06,
+      "loss": 3.4438,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 7.301655292510986,
+      "learning_rate": 1.8025289211729892e-06,
+      "loss": 3.4319,
+      "step": 135
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 10.716564178466797,
+      "learning_rate": 1.8697874630078023e-06,
+      "loss": 3.4002,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 7.337723255157471,
+      "learning_rate": 1.937046004842615e-06,
+      "loss": 3.3665,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 10.341432571411133,
+      "learning_rate": 2.0043045466774284e-06,
+      "loss": 3.3772,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.8016531467437744,
+      "learning_rate": 2.071563088512241e-06,
+      "loss": 3.3814,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.7477502822875977,
+      "learning_rate": 2.1388216303470545e-06,
+      "loss": 3.3041,
+      "step": 160
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.7560316324234009,
+      "learning_rate": 2.2060801721818673e-06,
+      "loss": 3.3248,
+      "step": 165
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.820521593093872,
+      "learning_rate": 2.2733387140166806e-06,
+      "loss": 3.2615,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.081446409225464,
+      "learning_rate": 2.3405972558514934e-06,
+      "loss": 3.2126,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.2577123641967773,
+      "learning_rate": 2.4078557976863062e-06,
+      "loss": 3.1306,
+      "step": 180
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.315398693084717,
+      "learning_rate": 2.4751143395211195e-06,
+      "loss": 3.0733,
+      "step": 185
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.604109525680542,
+      "learning_rate": 2.5423728813559323e-06,
+      "loss": 3.0023,
+      "step": 190
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.508555889129639,
+      "learning_rate": 2.6096314231907456e-06,
+      "loss": 2.8853,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.1160340309143066,
+      "learning_rate": 2.676889965025558e-06,
+      "loss": 2.7973,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.671633243560791,
+      "learning_rate": 2.7441485068603713e-06,
+      "loss": 2.6951,
+      "step": 205
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.27539324760437,
+      "learning_rate": 2.811407048695184e-06,
+      "loss": 2.5681,
+      "step": 210
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.969864845275879,
+      "learning_rate": 2.8786655905299974e-06,
+      "loss": 2.4749,
+      "step": 215
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.089858055114746,
+      "learning_rate": 2.9459241323648102e-06,
+      "loss": 2.2944,
+      "step": 220
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.102409839630127,
+      "learning_rate": 3.0131826741996235e-06,
+      "loss": 2.238,
+      "step": 225
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.597651481628418,
+      "learning_rate": 3.0804412160344363e-06,
+      "loss": 2.0277,
+      "step": 230
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.943057537078857,
+      "learning_rate": 3.1476997578692496e-06,
+      "loss": 1.8801,
+      "step": 235
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.2581276893615723,
+      "learning_rate": 3.214958299704063e-06,
+      "loss": 1.7672,
+      "step": 240
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.6054794788360596,
+      "learning_rate": 3.2822168415388757e-06,
+      "loss": 1.6089,
+      "step": 245
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.729863166809082,
+      "learning_rate": 3.349475383373689e-06,
+      "loss": 1.5185,
+      "step": 250
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.080470561981201,
+      "learning_rate": 3.4167339252085018e-06,
+      "loss": 1.4243,
+      "step": 255
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.782588481903076,
+      "learning_rate": 3.483992467043315e-06,
+      "loss": 1.327,
+      "step": 260
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.053415298461914,
+      "learning_rate": 3.551251008878128e-06,
+      "loss": 1.2965,
+      "step": 265
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.398200273513794,
+      "learning_rate": 3.618509550712941e-06,
+      "loss": 1.1947,
+      "step": 270
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.759004592895508,
+      "learning_rate": 3.685768092547754e-06,
+      "loss": 1.1178,
+      "step": 275
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.8202435970306396,
+      "learning_rate": 3.7530266343825673e-06,
+      "loss": 1.0863,
+      "step": 280
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.264969825744629,
+      "learning_rate": 3.82028517621738e-06,
+      "loss": 1.0379,
+      "step": 285
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.7766692638397217,
+      "learning_rate": 3.887543718052193e-06,
+      "loss": 0.9759,
+      "step": 290
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.1840648651123047,
+      "learning_rate": 3.954802259887006e-06,
+      "loss": 0.9491,
+      "step": 295
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.854584217071533,
+      "learning_rate": 4.022060801721819e-06,
+      "loss": 0.9268,
+      "step": 300
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 0.7659996747970581,
+      "eval_runtime": 407.2258,
+      "eval_samples_per_second": 55.966,
+      "eval_steps_per_second": 6.996,
+      "eval_wer": 0.5487413969955661,
+      "step": 300
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.5880093574523926,
+      "learning_rate": 4.0893193435566314e-06,
+      "loss": 0.9178,
+      "step": 305
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.8778300285339355,
+      "learning_rate": 4.156577885391445e-06,
+      "loss": 0.8681,
+      "step": 310
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.6787381172180176,
+      "learning_rate": 4.223836427226258e-06,
+      "loss": 0.8382,
+      "step": 315
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.435033082962036,
+      "learning_rate": 4.29109496906107e-06,
+      "loss": 0.7988,
+      "step": 320
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.8749780654907227,
+      "learning_rate": 4.358353510895884e-06,
+      "loss": 0.7857,
+      "step": 325
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.7846786975860596,
+      "learning_rate": 4.425612052730697e-06,
+      "loss": 0.7397,
+      "step": 330
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.94502067565918,
+      "learning_rate": 4.479418886198548e-06,
+      "loss": 0.7387,
+      "step": 335
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.27470064163208,
+      "learning_rate": 4.546677428033361e-06,
+      "loss": 0.7405,
+      "step": 340
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.204165458679199,
+      "learning_rate": 4.6139359698681735e-06,
+      "loss": 0.7275,
+      "step": 345
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.4391987323760986,
+      "learning_rate": 4.681194511702987e-06,
+      "loss": 0.699,
+      "step": 350
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.0003039836883545,
+      "learning_rate": 4.7484530535378e-06,
+      "loss": 0.6758,
+      "step": 355
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.514505386352539,
+      "learning_rate": 4.8157115953726125e-06,
+      "loss": 0.6108,
+      "step": 360
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.828533172607422,
+      "learning_rate": 4.882970137207426e-06,
+      "loss": 0.6369,
+      "step": 365
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.272637844085693,
+      "learning_rate": 4.950228679042239e-06,
+      "loss": 0.6249,
+      "step": 370
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.553176403045654,
+      "learning_rate": 5.017487220877052e-06,
+      "loss": 0.6656,
+      "step": 375
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.294391393661499,
+      "learning_rate": 5.084745762711865e-06,
+      "loss": 0.6325,
+      "step": 380
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.417149066925049,
+      "learning_rate": 5.152004304546678e-06,
+      "loss": 0.6315,
+      "step": 385
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.738922119140625,
+      "learning_rate": 5.219262846381491e-06,
+      "loss": 0.6025,
+      "step": 390
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.0851354598999023,
+      "learning_rate": 5.286521388216304e-06,
+      "loss": 0.6029,
+      "step": 395
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.0445644855499268,
+      "learning_rate": 5.353779930051116e-06,
+      "loss": 0.5498,
+      "step": 400
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 7.218364238739014,
+      "learning_rate": 5.421038471885929e-06,
+      "loss": 0.6104,
+      "step": 405
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 9.436169624328613,
+      "learning_rate": 5.4882970137207426e-06,
+      "loss": 0.5648,
+      "step": 410
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.031554222106934,
+      "learning_rate": 5.555555555555556e-06,
+      "loss": 0.582,
+      "step": 415
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.7603867053985596,
+      "learning_rate": 5.622814097390368e-06,
+      "loss": 0.5399,
+      "step": 420
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.728298187255859,
+      "learning_rate": 5.6900726392251815e-06,
+      "loss": 0.583,
+      "step": 425
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.384444713592529,
+      "learning_rate": 5.757331181059995e-06,
+      "loss": 0.5407,
+      "step": 430
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.461092472076416,
+      "learning_rate": 5.824589722894808e-06,
+      "loss": 0.6036,
+      "step": 435
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.6895875930786133,
+      "learning_rate": 5.8918482647296204e-06,
+      "loss": 0.5211,
+      "step": 440
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 34.76512908935547,
+      "learning_rate": 5.959106806564434e-06,
+      "loss": 0.4864,
+      "step": 445
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.9635069370269775,
+      "learning_rate": 6.026365348399247e-06,
+      "loss": 0.5261,
+      "step": 450
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.241861343383789,
+      "learning_rate": 6.093623890234059e-06,
+      "loss": 0.5026,
+      "step": 455
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.966274261474609,
+      "learning_rate": 6.160882432068873e-06,
+      "loss": 0.4756,
+      "step": 460
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.7707536220550537,
+      "learning_rate": 6.228140973903686e-06,
+      "loss": 0.5353,
+      "step": 465
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.399672031402588,
+      "learning_rate": 6.295399515738499e-06,
+      "loss": 0.4679,
+      "step": 470
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.132097244262695,
+      "learning_rate": 6.362658057573312e-06,
+      "loss": 0.4794,
+      "step": 475
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.423065185546875,
+      "learning_rate": 6.429916599408126e-06,
+      "loss": 0.4761,
+      "step": 480
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.865494251251221,
+      "learning_rate": 6.497175141242938e-06,
+      "loss": 0.4976,
+      "step": 485
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 10.4710054397583,
+      "learning_rate": 6.564433683077751e-06,
+      "loss": 0.4876,
+      "step": 490
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.027119159698486,
+      "learning_rate": 6.631692224912564e-06,
+      "loss": 0.4882,
+      "step": 495
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.406029462814331,
+      "learning_rate": 6.698950766747378e-06,
+      "loss": 0.4704,
+      "step": 500
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.1524789333343506,
+      "learning_rate": 6.76620930858219e-06,
+      "loss": 0.4858,
+      "step": 505
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8089348077774048,
+      "learning_rate": 6.8334678504170036e-06,
+      "loss": 0.4677,
+      "step": 510
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.8395497798919678,
+      "learning_rate": 6.900726392251816e-06,
+      "loss": 0.4422,
+      "step": 515
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.593482494354248,
+      "learning_rate": 6.96798493408663e-06,
+      "loss": 0.454,
+      "step": 520
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.468766212463379,
+      "learning_rate": 7.0352434759214425e-06,
+      "loss": 0.4537,
+      "step": 525
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.4911346435546875,
+      "learning_rate": 7.102502017756256e-06,
+      "loss": 0.4495,
+      "step": 530
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.3779380321502686,
+      "learning_rate": 7.169760559591068e-06,
+      "loss": 0.459,
+      "step": 535
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.107753038406372,
+      "learning_rate": 7.237019101425882e-06,
+      "loss": 0.3958,
+      "step": 540
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.1042022705078125,
+      "learning_rate": 7.304277643260695e-06,
+      "loss": 0.4734,
+      "step": 545
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.1506924629211426,
+      "learning_rate": 7.371536185095508e-06,
+      "loss": 0.435,
+      "step": 550
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.979548454284668,
+      "learning_rate": 7.43879472693032e-06,
+      "loss": 0.429,
+      "step": 555
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.7409842014312744,
+      "learning_rate": 7.5060532687651345e-06,
+      "loss": 0.4423,
+      "step": 560
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5861988067626953,
+      "learning_rate": 7.573311810599946e-06,
+      "loss": 0.4421,
+      "step": 565
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.1968331336975098,
+      "learning_rate": 7.64057035243476e-06,
+      "loss": 0.4644,
+      "step": 570
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.536830425262451,
+      "learning_rate": 7.707828894269572e-06,
+      "loss": 0.3952,
+      "step": 575
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.2695887088775635,
+      "learning_rate": 7.775087436104387e-06,
+      "loss": 0.3959,
+      "step": 580
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 7.8040313720703125,
+      "learning_rate": 7.842345977939198e-06,
+      "loss": 0.4161,
+      "step": 585
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 8.609118461608887,
+      "learning_rate": 7.909604519774012e-06,
+      "loss": 0.409,
+      "step": 590
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.3660991191864014,
+      "learning_rate": 7.976863061608825e-06,
+      "loss": 0.3958,
+      "step": 595
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.783254861831665,
+      "learning_rate": 8.044121603443638e-06,
+      "loss": 0.3867,
+      "step": 600
+    },
+    {
+      "epoch": 0.23,
+      "eval_loss": 0.3323296904563904,
+      "eval_runtime": 347.9329,
+      "eval_samples_per_second": 65.504,
+      "eval_steps_per_second": 8.188,
+      "eval_wer": 0.2807660810005956,
+      "step": 600
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.8303325176239014,
+      "learning_rate": 8.111380145278451e-06,
+      "loss": 0.3982,
+      "step": 605
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.565206527709961,
+      "learning_rate": 8.178638687113263e-06,
+      "loss": 0.4294,
+      "step": 610
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.9662604331970215,
+      "learning_rate": 8.245897228948076e-06,
+      "loss": 0.3745,
+      "step": 615
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.9636054039001465,
+      "learning_rate": 8.31315577078289e-06,
+      "loss": 0.3753,
+      "step": 620
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.9991724491119385,
+      "learning_rate": 8.380414312617703e-06,
+      "loss": 0.3778,
+      "step": 625
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.218487024307251,
+      "learning_rate": 8.447672854452516e-06,
+      "loss": 0.3555,
+      "step": 630
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.5706021785736084,
+      "learning_rate": 8.51493139628733e-06,
+      "loss": 0.3784,
+      "step": 635
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.293356418609619,
+      "learning_rate": 8.58218993812214e-06,
+      "loss": 0.3582,
+      "step": 640
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.760920524597168,
+      "learning_rate": 8.649448479956956e-06,
+      "loss": 0.3459,
+      "step": 645
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9550977945327759,
+      "learning_rate": 8.716707021791767e-06,
+      "loss": 0.3906,
+      "step": 650
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.899873971939087,
+      "learning_rate": 8.78396556362658e-06,
+      "loss": 0.3362,
+      "step": 655
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.135042667388916,
+      "learning_rate": 8.851224105461394e-06,
+      "loss": 0.3641,
+      "step": 660
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.928522109985352,
+      "learning_rate": 8.918482647296207e-06,
+      "loss": 0.3751,
+      "step": 665
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3457231521606445,
+      "learning_rate": 8.985741189131019e-06,
+      "loss": 0.3446,
+      "step": 670
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.5795159339904785,
+      "learning_rate": 9.052999730965834e-06,
+      "loss": 0.3657,
+      "step": 675
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.7840797901153564,
+      "learning_rate": 9.120258272800645e-06,
+      "loss": 0.3751,
+      "step": 680
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1620266437530518,
+      "learning_rate": 9.187516814635458e-06,
+      "loss": 0.3158,
+      "step": 685
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.75873064994812,
+      "learning_rate": 9.254775356470272e-06,
+      "loss": 0.3419,
+      "step": 690
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.593353271484375,
+      "learning_rate": 9.322033898305085e-06,
+      "loss": 0.3167,
+      "step": 695
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.5261001586914062,
+      "learning_rate": 9.389292440139898e-06,
+      "loss": 0.4023,
+      "step": 700
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.605105400085449,
+      "learning_rate": 9.456550981974711e-06,
+      "loss": 0.3529,
+      "step": 705
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.380168914794922,
+      "learning_rate": 9.523809523809523e-06,
+      "loss": 0.3295,
+      "step": 710
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1508231163024902,
+      "learning_rate": 9.591068065644338e-06,
+      "loss": 0.334,
+      "step": 715
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.075620412826538,
+      "learning_rate": 9.65832660747915e-06,
+      "loss": 0.3402,
+      "step": 720
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.256362199783325,
+      "learning_rate": 9.725585149313963e-06,
+      "loss": 0.3136,
+      "step": 725
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.005175828933716,
+      "learning_rate": 9.792843691148776e-06,
+      "loss": 0.3004,
+      "step": 730
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.374253273010254,
+      "learning_rate": 9.86010223298359e-06,
+      "loss": 0.3275,
+      "step": 735
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0386574268341064,
+      "learning_rate": 9.927360774818403e-06,
+      "loss": 0.3155,
+      "step": 740
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.085981845855713,
+      "learning_rate": 9.994619316653216e-06,
+      "loss": 0.2813,
+      "step": 745
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.3343632221221924,
+      "learning_rate": 1.0061877858488027e-05,
+      "loss": 0.3234,
+      "step": 750
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.110535144805908,
+      "learning_rate": 1.0129136400322842e-05,
+      "loss": 0.3326,
+      "step": 755
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.686037302017212,
+      "learning_rate": 1.0196394942157654e-05,
+      "loss": 0.3282,
+      "step": 760
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.040796995162964,
+      "learning_rate": 1.0263653483992467e-05,
+      "loss": 0.3119,
+      "step": 765
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6064058542251587,
+      "learning_rate": 1.033091202582728e-05,
+      "loss": 0.2791,
+      "step": 770
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.2085189819335938,
+      "learning_rate": 1.0398170567662094e-05,
+      "loss": 0.3039,
+      "step": 775
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.689975619316101,
+      "learning_rate": 1.0465429109496907e-05,
+      "loss": 0.3066,
+      "step": 780
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.6196117401123047,
+      "learning_rate": 1.053268765133172e-05,
+      "loss": 0.322,
+      "step": 785
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.840836763381958,
+      "learning_rate": 1.0599946193166532e-05,
+      "loss": 0.3248,
+      "step": 790
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5852344036102295,
+      "learning_rate": 1.0667204735001347e-05,
+      "loss": 0.3351,
+      "step": 795
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.5725057125091553,
+      "learning_rate": 1.0734463276836158e-05,
+      "loss": 0.2955,
+      "step": 800
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8865330219268799,
+      "learning_rate": 1.0801721818670972e-05,
+      "loss": 0.3195,
+      "step": 805
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.9064290523529053,
+      "learning_rate": 1.0868980360505785e-05,
+      "loss": 0.3107,
+      "step": 810
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.041391134262085,
+      "learning_rate": 1.0936238902340598e-05,
+      "loss": 0.2906,
+      "step": 815
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.1083896160125732,
+      "learning_rate": 1.100349744417541e-05,
+      "loss": 0.3126,
+      "step": 820
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0025436878204346,
+      "learning_rate": 1.1070755986010225e-05,
+      "loss": 0.2834,
+      "step": 825
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.2463297843933105,
+      "learning_rate": 1.1138014527845036e-05,
+      "loss": 0.2711,
+      "step": 830
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7511409521102905,
+      "learning_rate": 1.120527306967985e-05,
+      "loss": 0.2773,
+      "step": 835
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.6397161483764648,
+      "learning_rate": 1.1272531611514663e-05,
+      "loss": 0.3018,
+      "step": 840
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.5769562721252441,
+      "learning_rate": 1.1339790153349476e-05,
+      "loss": 0.2836,
+      "step": 845
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8413454294204712,
+      "learning_rate": 1.140704869518429e-05,
+      "loss": 0.3065,
+      "step": 850
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1670055389404297,
+      "learning_rate": 1.1474307237019103e-05,
+      "loss": 0.3314,
+      "step": 855
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8437851667404175,
+      "learning_rate": 1.1541565778853914e-05,
+      "loss": 0.3004,
+      "step": 860
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.2325055599212646,
+      "learning_rate": 1.1608824320688729e-05,
+      "loss": 0.3009,
+      "step": 865
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9645568132400513,
+      "learning_rate": 1.167608286252354e-05,
+      "loss": 0.302,
+      "step": 870
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.3938262462615967,
+      "learning_rate": 1.1743341404358354e-05,
+      "loss": 0.2552,
+      "step": 875
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.432025909423828,
+      "learning_rate": 1.1810599946193167e-05,
+      "loss": 0.301,
+      "step": 880
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7212738990783691,
+      "learning_rate": 1.187785848802798e-05,
+      "loss": 0.2807,
+      "step": 885
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.442462682723999,
+      "learning_rate": 1.1945117029862794e-05,
+      "loss": 0.3038,
+      "step": 890
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.5379706621170044,
+      "learning_rate": 1.2012375571697607e-05,
+      "loss": 0.2905,
+      "step": 895
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8098061084747314,
+      "learning_rate": 1.2079634113532419e-05,
+      "loss": 0.2881,
+      "step": 900
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 0.24923652410507202,
+      "eval_runtime": 345.5157,
+      "eval_samples_per_second": 65.962,
+      "eval_steps_per_second": 8.246,
+      "eval_wer": 0.21471113758189397,
+      "step": 900
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.5291564464569092,
+      "learning_rate": 1.2146892655367234e-05,
+      "loss": 0.3034,
+      "step": 905
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8100793361663818,
+      "learning_rate": 1.2214151197202045e-05,
+      "loss": 0.2871,
+      "step": 910
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.001034259796143,
+      "learning_rate": 1.2281409739036858e-05,
+      "loss": 0.3243,
+      "step": 915
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8379939794540405,
+      "learning_rate": 1.2348668280871672e-05,
+      "loss": 0.2831,
+      "step": 920
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.3951992988586426,
+      "learning_rate": 1.2415926822706485e-05,
+      "loss": 0.2767,
+      "step": 925
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9685766696929932,
+      "learning_rate": 1.2483185364541296e-05,
+      "loss": 0.2658,
+      "step": 930
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 12.722960472106934,
+      "learning_rate": 1.2550443906376111e-05,
+      "loss": 0.292,
+      "step": 935
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7162452936172485,
+      "learning_rate": 1.2617702448210925e-05,
+      "loss": 0.2758,
+      "step": 940
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9299818277359009,
+      "learning_rate": 1.2684960990045736e-05,
+      "loss": 0.3046,
+      "step": 945
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9780501127243042,
+      "learning_rate": 1.275221953188055e-05,
+      "loss": 0.2825,
+      "step": 950
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.6639394760131836,
+      "learning_rate": 1.2819478073715363e-05,
+      "loss": 0.2774,
+      "step": 955
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.709362745285034,
+      "learning_rate": 1.2886736615550176e-05,
+      "loss": 0.2906,
+      "step": 960
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7930606603622437,
+      "learning_rate": 1.2953995157384988e-05,
+      "loss": 0.2877,
+      "step": 965
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8784518241882324,
+      "learning_rate": 1.30212536992198e-05,
+      "loss": 0.2961,
+      "step": 970
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.2258481979370117,
+      "learning_rate": 1.3088512241054616e-05,
+      "loss": 0.2654,
+      "step": 975
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.3855760097503662,
+      "learning_rate": 1.3155770782889429e-05,
+      "loss": 0.3057,
+      "step": 980
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.848044991493225,
+      "learning_rate": 1.322302932472424e-05,
+      "loss": 0.2902,
+      "step": 985
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.671026587486267,
+      "learning_rate": 1.3290287866559054e-05,
+      "loss": 0.2676,
+      "step": 990
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8223345279693604,
+      "learning_rate": 1.3357546408393867e-05,
+      "loss": 0.2754,
+      "step": 995
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.218132495880127,
+      "learning_rate": 1.342480495022868e-05,
+      "loss": 0.2735,
+      "step": 1000
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.459094762802124,
+      "learning_rate": 1.3492063492063492e-05,
+      "loss": 0.2777,
+      "step": 1005
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.807625412940979,
+      "learning_rate": 1.3559322033898305e-05,
+      "loss": 0.2677,
+      "step": 1010
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7342606782913208,
+      "learning_rate": 1.362658057573312e-05,
+      "loss": 0.2805,
+      "step": 1015
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.3494089841842651,
+      "learning_rate": 1.3693839117567933e-05,
+      "loss": 0.2709,
+      "step": 1020
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.787153720855713,
+      "learning_rate": 1.3761097659402743e-05,
+      "loss": 0.2891,
+      "step": 1025
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.547590970993042,
+      "learning_rate": 1.3828356201237558e-05,
+      "loss": 0.3514,
+      "step": 1030
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.545663595199585,
+      "learning_rate": 1.3895614743072372e-05,
+      "loss": 0.2332,
+      "step": 1035
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7619597911834717,
+      "learning_rate": 1.3962873284907183e-05,
+      "loss": 0.2825,
+      "step": 1040
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.3518449068069458,
+      "learning_rate": 1.4030131826741996e-05,
+      "loss": 0.2841,
+      "step": 1045
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.708491325378418,
+      "learning_rate": 1.409739036857681e-05,
+      "loss": 0.2999,
+      "step": 1050
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.1553268432617188,
+      "learning_rate": 1.4164648910411623e-05,
+      "loss": 0.3142,
+      "step": 1055
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.2359050512313843,
+      "learning_rate": 1.4231907452246434e-05,
+      "loss": 0.2721,
+      "step": 1060
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.04133939743042,
+      "learning_rate": 1.4299165994081248e-05,
+      "loss": 0.2796,
+      "step": 1065
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8437272310256958,
+      "learning_rate": 1.4366424535916063e-05,
+      "loss": 0.2638,
+      "step": 1070
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.6330442428588867,
+      "learning_rate": 1.4433683077750876e-05,
+      "loss": 0.2696,
+      "step": 1075
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6921255588531494,
+      "learning_rate": 1.4500941619585688e-05,
+      "loss": 0.2539,
+      "step": 1080
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 11.057169914245605,
+      "learning_rate": 1.45682001614205e-05,
+      "loss": 0.2811,
+      "step": 1085
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.4175479412078857,
+      "learning_rate": 1.4635458703255314e-05,
+      "loss": 0.2744,
+      "step": 1090
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.466901183128357,
+      "learning_rate": 1.4702717245090127e-05,
+      "loss": 0.2649,
+      "step": 1095
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7737650871276855,
+      "learning_rate": 1.4756524078557978e-05,
+      "loss": 0.3143,
+      "step": 1100
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.2421361207962036,
+      "learning_rate": 1.482378262039279e-05,
+      "loss": 0.2794,
+      "step": 1105
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.51334810256958,
+      "learning_rate": 1.4891041162227603e-05,
+      "loss": 0.2631,
+      "step": 1110
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9589332342147827,
+      "learning_rate": 1.4958299704062418e-05,
+      "loss": 0.2679,
+      "step": 1115
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.2889121770858765,
+      "learning_rate": 1.5025558245897231e-05,
+      "loss": 0.2507,
+      "step": 1120
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 7.654943943023682,
+      "learning_rate": 1.5092816787732041e-05,
+      "loss": 0.2706,
+      "step": 1125
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 6.227551460266113,
+      "learning_rate": 1.5160075329566856e-05,
+      "loss": 0.2508,
+      "step": 1130
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.388632893562317,
+      "learning_rate": 1.522733387140167e-05,
+      "loss": 0.2552,
+      "step": 1135
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.1430631875991821,
+      "learning_rate": 1.529459241323648e-05,
+      "loss": 0.2838,
+      "step": 1140
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8639068603515625,
+      "learning_rate": 1.5361850955071294e-05,
+      "loss": 0.2446,
+      "step": 1145
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.6784844398498535,
+      "learning_rate": 1.5429109496906108e-05,
+      "loss": 0.2636,
+      "step": 1150
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.447296619415283,
+      "learning_rate": 1.549636803874092e-05,
+      "loss": 0.2574,
+      "step": 1155
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7050584554672241,
+      "learning_rate": 1.5563626580575734e-05,
+      "loss": 0.2598,
+      "step": 1160
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.4974327087402344,
+      "learning_rate": 1.5630885122410547e-05,
+      "loss": 0.2306,
+      "step": 1165
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 7.473500728607178,
+      "learning_rate": 1.569814366424536e-05,
+      "loss": 0.2365,
+      "step": 1170
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.3481996059417725,
+      "learning_rate": 1.5765402206080174e-05,
+      "loss": 0.2283,
+      "step": 1175
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.0838769674301147,
+      "learning_rate": 1.5832660747914984e-05,
+      "loss": 0.2359,
+      "step": 1180
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.5355899333953857,
+      "learning_rate": 1.5899919289749797e-05,
+      "loss": 0.2489,
+      "step": 1185
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.202460765838623,
+      "learning_rate": 1.5967177831584614e-05,
+      "loss": 0.2359,
+      "step": 1190
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.3273018598556519,
+      "learning_rate": 1.6034436373419427e-05,
+      "loss": 0.2294,
+      "step": 1195
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.762060284614563,
+      "learning_rate": 1.6101694915254237e-05,
+      "loss": 0.2333,
+      "step": 1200
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 0.21728818118572235,
+      "eval_runtime": 359.1091,
+      "eval_samples_per_second": 63.465,
+      "eval_steps_per_second": 7.934,
+      "eval_wer": 0.18808938852491563,
+      "step": 1200
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.193408489227295,
+      "learning_rate": 1.616895345708905e-05,
+      "loss": 0.2508,
+      "step": 1205
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.0018819570541382,
+      "learning_rate": 1.6236211998923863e-05,
+      "loss": 0.2425,
+      "step": 1210
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.4503206014633179,
+      "learning_rate": 1.6303470540758677e-05,
+      "loss": 0.213,
+      "step": 1215
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.504359245300293,
+      "learning_rate": 1.637072908259349e-05,
+      "loss": 0.2338,
+      "step": 1220
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.3940457105636597,
+      "learning_rate": 1.6437987624428303e-05,
+      "loss": 0.2767,
+      "step": 1225
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.1934741735458374,
+      "learning_rate": 1.6505246166263116e-05,
+      "loss": 0.2248,
+      "step": 1230
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.621188163757324,
+      "learning_rate": 1.657250470809793e-05,
+      "loss": 0.2282,
+      "step": 1235
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.45343017578125,
+      "learning_rate": 1.663976324993274e-05,
+      "loss": 0.2715,
+      "step": 1240
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.2950636148452759,
+      "learning_rate": 1.6707021791767556e-05,
+      "loss": 0.2395,
+      "step": 1245
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.2662973403930664,
+      "learning_rate": 1.677428033360237e-05,
+      "loss": 0.2281,
+      "step": 1250
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.1355154514312744,
+      "learning_rate": 1.6841538875437183e-05,
+      "loss": 0.2126,
+      "step": 1255
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 7.395053863525391,
+      "learning_rate": 1.6908797417271992e-05,
+      "loss": 0.2295,
+      "step": 1260
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.5153887271881104,
+      "learning_rate": 1.6976055959106806e-05,
+      "loss": 0.2605,
+      "step": 1265
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.7468416690826416,
+      "learning_rate": 1.704331450094162e-05,
+      "loss": 0.2566,
+      "step": 1270
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.5864756107330322,
+      "learning_rate": 1.7110573042776436e-05,
+      "loss": 0.2604,
+      "step": 1275
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.574877977371216,
+      "learning_rate": 1.7177831584611246e-05,
+      "loss": 0.2369,
+      "step": 1280
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.776024580001831,
+      "learning_rate": 1.724509012644606e-05,
+      "loss": 0.2289,
+      "step": 1285
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9084045886993408,
+      "learning_rate": 1.7312348668280872e-05,
+      "loss": 0.2324,
+      "step": 1290
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7296181917190552,
+      "learning_rate": 1.7379607210115685e-05,
+      "loss": 0.2418,
+      "step": 1295
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.3596465587615967,
+      "learning_rate": 1.74468657519505e-05,
+      "loss": 0.2871,
+      "step": 1300
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.2359967231750488,
+      "learning_rate": 1.7514124293785312e-05,
+      "loss": 0.2268,
+      "step": 1305
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.5680181980133057,
+      "learning_rate": 1.7581382835620125e-05,
+      "loss": 0.2254,
+      "step": 1310
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.332973837852478,
+      "learning_rate": 1.764864137745494e-05,
+      "loss": 0.279,
+      "step": 1315
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.2815146446228027,
+      "learning_rate": 1.7715899919289748e-05,
+      "loss": 0.216,
+      "step": 1320
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.2612462043762207,
+      "learning_rate": 1.7783158461124565e-05,
+      "loss": 0.2499,
+      "step": 1325
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.5820578336715698,
+      "learning_rate": 1.7850417002959378e-05,
+      "loss": 0.2246,
+      "step": 1330
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9472665786743164,
+      "learning_rate": 1.791767554479419e-05,
+      "loss": 0.2475,
+      "step": 1335
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.0286521911621094,
+      "learning_rate": 1.7984934086629e-05,
+      "loss": 0.2264,
+      "step": 1340
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.8059351444244385,
+      "learning_rate": 1.8052192628463815e-05,
+      "loss": 0.2111,
+      "step": 1345
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.500752329826355,
+      "learning_rate": 1.8119451170298628e-05,
+      "loss": 0.2271,
+      "step": 1350
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.0558726787567139,
+      "learning_rate": 1.8186709712133444e-05,
+      "loss": 0.2403,
+      "step": 1355
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.0742583274841309,
+      "learning_rate": 1.8253968253968254e-05,
+      "loss": 0.2152,
+      "step": 1360
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.8488976955413818,
+      "learning_rate": 1.8321226795803068e-05,
+      "loss": 0.2368,
+      "step": 1365
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.3359456062316895,
+      "learning_rate": 1.838848533763788e-05,
+      "loss": 0.2092,
+      "step": 1370
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.1061750650405884,
+      "learning_rate": 1.8455743879472694e-05,
+      "loss": 0.29,
+      "step": 1375
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.412407159805298,
+      "learning_rate": 1.8523002421307507e-05,
+      "loss": 0.2204,
+      "step": 1380
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.2632677555084229,
+      "learning_rate": 1.859026096314232e-05,
+      "loss": 0.2201,
+      "step": 1385
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.971044898033142,
+      "learning_rate": 1.8657519504977134e-05,
+      "loss": 0.201,
+      "step": 1390
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.5854196548461914,
+      "learning_rate": 1.8711326338444983e-05,
+      "loss": 0.2651,
+      "step": 1395
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.2765356302261353,
+      "learning_rate": 1.8778584880279796e-05,
+      "loss": 0.2257,
+      "step": 1400
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.431352138519287,
+      "learning_rate": 1.884584342211461e-05,
+      "loss": 0.2191,
+      "step": 1405
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 10.63003921508789,
+      "learning_rate": 1.8913101963949423e-05,
+      "loss": 0.2484,
+      "step": 1410
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.0647368431091309,
+      "learning_rate": 1.8980360505784236e-05,
+      "loss": 0.2267,
+      "step": 1415
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8995018005371094,
+      "learning_rate": 1.9047619047619046e-05,
+      "loss": 0.2216,
+      "step": 1420
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.541861653327942,
+      "learning_rate": 1.9114877589453863e-05,
+      "loss": 0.2211,
+      "step": 1425
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.1336451768875122,
+      "learning_rate": 1.9182136131288676e-05,
+      "loss": 0.209,
+      "step": 1430
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.100672721862793,
+      "learning_rate": 1.924939467312349e-05,
+      "loss": 0.2197,
+      "step": 1435
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.4497236013412476,
+      "learning_rate": 1.93166532149583e-05,
+      "loss": 0.2343,
+      "step": 1440
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.0944918394088745,
+      "learning_rate": 1.9383911756793112e-05,
+      "loss": 0.2199,
+      "step": 1445
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.528102159500122,
+      "learning_rate": 1.9451170298627926e-05,
+      "loss": 0.2433,
+      "step": 1450
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8009004592895508,
+      "learning_rate": 1.9518428840462742e-05,
+      "loss": 0.2283,
+      "step": 1455
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.4100067615509033,
+      "learning_rate": 1.9585687382297552e-05,
+      "loss": 0.2289,
+      "step": 1460
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.2447171211242676,
+      "learning_rate": 1.9652945924132365e-05,
+      "loss": 0.211,
+      "step": 1465
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.7598168849945068,
+      "learning_rate": 1.972020446596718e-05,
+      "loss": 0.2398,
+      "step": 1470
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.7354618310928345,
+      "learning_rate": 1.9787463007801992e-05,
+      "loss": 0.225,
+      "step": 1475
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.9788336753845215,
+      "learning_rate": 1.9854721549636805e-05,
+      "loss": 0.2183,
+      "step": 1480
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.7741131782531738,
+      "learning_rate": 1.992198009147162e-05,
+      "loss": 0.2245,
+      "step": 1485
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9202806949615479,
+      "learning_rate": 1.9989238633306432e-05,
+      "loss": 0.2209,
+      "step": 1490
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.4790838956832886,
+      "learning_rate": 2.0056497175141245e-05,
+      "loss": 0.2431,
+      "step": 1495
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.5455936193466187,
+      "learning_rate": 2.0123755716976055e-05,
+      "loss": 0.2388,
+      "step": 1500
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 0.19693605601787567,
+      "eval_runtime": 357.5274,
+      "eval_samples_per_second": 63.746,
+      "eval_steps_per_second": 7.969,
+      "eval_wer": 0.17047597776454237,
+      "step": 1500
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.1404078006744385,
+      "learning_rate": 2.0191014258810868e-05,
+      "loss": 0.2275,
+      "step": 1505
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.519964575767517,
+      "learning_rate": 2.0258272800645685e-05,
+      "loss": 0.287,
+      "step": 1510
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.4172720909118652,
+      "learning_rate": 2.0325531342480498e-05,
+      "loss": 0.2172,
+      "step": 1515
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.513420820236206,
+      "learning_rate": 2.0392789884315308e-05,
+      "loss": 0.2272,
+      "step": 1520
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.2000703811645508,
+      "learning_rate": 2.046004842615012e-05,
+      "loss": 0.2276,
+      "step": 1525
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.4187487363815308,
+      "learning_rate": 2.0527306967984934e-05,
+      "loss": 0.2088,
+      "step": 1530
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.4195059537887573,
+      "learning_rate": 2.0594565509819748e-05,
+      "loss": 0.2267,
+      "step": 1535
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.3100816011428833,
+      "learning_rate": 2.066182405165456e-05,
+      "loss": 0.2485,
+      "step": 1540
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.7420188188552856,
+      "learning_rate": 2.0729082593489374e-05,
+      "loss": 0.2153,
+      "step": 1545
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.4291852712631226,
+      "learning_rate": 2.0796341135324188e-05,
+      "loss": 0.2076,
+      "step": 1550
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.208625316619873,
+      "learning_rate": 2.0863599677159e-05,
+      "loss": 0.2366,
+      "step": 1555
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0919506549835205,
+      "learning_rate": 2.0930858218993814e-05,
+      "loss": 0.2095,
+      "step": 1560
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.33111572265625,
+      "learning_rate": 2.0998116760828627e-05,
+      "loss": 0.1961,
+      "step": 1565
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.3000023365020752,
+      "learning_rate": 2.106537530266344e-05,
+      "loss": 0.2311,
+      "step": 1570
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.482287406921387,
+      "learning_rate": 2.1132633844498254e-05,
+      "loss": 0.2345,
+      "step": 1575
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9414831399917603,
+      "learning_rate": 2.1199892386333064e-05,
+      "loss": 0.1956,
+      "step": 1580
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.1252304315567017,
+      "learning_rate": 2.1267150928167877e-05,
+      "loss": 0.2126,
+      "step": 1585
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.282502293586731,
+      "learning_rate": 2.1334409470002694e-05,
+      "loss": 0.209,
+      "step": 1590
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.1397236585617065,
+      "learning_rate": 2.1401668011837507e-05,
+      "loss": 0.2104,
+      "step": 1595
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7864363193511963,
+      "learning_rate": 2.1468926553672317e-05,
+      "loss": 0.1863,
+      "step": 1600
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.307068109512329,
+      "learning_rate": 2.153618509550713e-05,
+      "loss": 0.2303,
+      "step": 1605
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.6946921348571777,
+      "learning_rate": 2.1603443637341943e-05,
+      "loss": 0.2412,
+      "step": 1610
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.924528956413269,
+      "learning_rate": 2.1670702179176757e-05,
+      "loss": 0.211,
+      "step": 1615
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.421558141708374,
+      "learning_rate": 2.173796072101157e-05,
+      "loss": 0.2448,
+      "step": 1620
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.9734997749328613,
+      "learning_rate": 2.1805219262846383e-05,
+      "loss": 0.2038,
+      "step": 1625
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8193857669830322,
+      "learning_rate": 2.1872477804681196e-05,
+      "loss": 0.2424,
+      "step": 1630
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.3094624280929565,
+      "learning_rate": 2.193973634651601e-05,
+      "loss": 0.1862,
+      "step": 1635
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.3237059116363525,
+      "learning_rate": 2.200699488835082e-05,
+      "loss": 0.2075,
+      "step": 1640
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6700096130371094,
+      "learning_rate": 2.2074253430185636e-05,
+      "loss": 0.198,
+      "step": 1645
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.8978589177131653,
+      "learning_rate": 2.214151197202045e-05,
+      "loss": 0.2236,
+      "step": 1650
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.2809327840805054,
+      "learning_rate": 2.220877051385526e-05,
+      "loss": 0.1855,
+      "step": 1655
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.0053337812423706,
+      "learning_rate": 2.2276029055690073e-05,
+      "loss": 0.2424,
+      "step": 1660
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.2768789529800415,
+      "learning_rate": 2.2343287597524886e-05,
+      "loss": 0.2044,
+      "step": 1665
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.9553895592689514,
+      "learning_rate": 2.24105461393597e-05,
+      "loss": 0.2022,
+      "step": 1670
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.951020240783691,
+      "learning_rate": 2.2477804681194512e-05,
+      "loss": 0.2014,
+      "step": 1675
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.2210066318511963,
+      "learning_rate": 2.2545063223029326e-05,
+      "loss": 0.2284,
+      "step": 1680
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.4670777320861816,
+      "learning_rate": 2.261232176486414e-05,
+      "loss": 0.2019,
+      "step": 1685
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.8476448059082031,
+      "learning_rate": 2.2679580306698952e-05,
+      "loss": 0.2084,
+      "step": 1690
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.0851925611495972,
+      "learning_rate": 2.2746838848533762e-05,
+      "loss": 0.2036,
+      "step": 1695
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.9431840777397156,
+      "learning_rate": 2.281409739036858e-05,
+      "loss": 0.1917,
+      "step": 1700
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.3302503824234009,
+      "learning_rate": 2.2881355932203392e-05,
+      "loss": 0.2319,
+      "step": 1705
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.1124733686447144,
+      "learning_rate": 2.2948614474038205e-05,
+      "loss": 0.2113,
+      "step": 1710
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.3600150346755981,
+      "learning_rate": 2.3015873015873015e-05,
+      "loss": 0.1841,
+      "step": 1715
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.6662566661834717,
+      "learning_rate": 2.3083131557707828e-05,
+      "loss": 0.2193,
+      "step": 1720
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.9137592911720276,
+      "learning_rate": 2.315039009954264e-05,
+      "loss": 0.2038,
+      "step": 1725
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.4198087453842163,
+      "learning_rate": 2.3217648641377458e-05,
+      "loss": 0.2012,
+      "step": 1730
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.1093159914016724,
+      "learning_rate": 2.3284907183212268e-05,
+      "loss": 0.2245,
+      "step": 1735
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.1159968376159668,
+      "learning_rate": 2.335216572504708e-05,
+      "loss": 0.1963,
+      "step": 1740
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 6.297104358673096,
+      "learning_rate": 2.3419424266881895e-05,
+      "loss": 0.2858,
+      "step": 1745
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.903351068496704,
+      "learning_rate": 2.3486682808716708e-05,
+      "loss": 0.2158,
+      "step": 1750
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 6.906310558319092,
+      "learning_rate": 2.355394135055152e-05,
+      "loss": 0.195,
+      "step": 1755
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.3247296810150146,
+      "learning_rate": 2.3621199892386334e-05,
+      "loss": 0.193,
+      "step": 1760
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.4418847560882568,
+      "learning_rate": 2.3688458434221148e-05,
+      "loss": 0.2591,
+      "step": 1765
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.7647428512573242,
+      "learning_rate": 2.375571697605596e-05,
+      "loss": 0.1778,
+      "step": 1770
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.348227024078369,
+      "learning_rate": 2.382297551789077e-05,
+      "loss": 0.2079,
+      "step": 1775
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.9404179453849792,
+      "learning_rate": 2.3890234059725587e-05,
+      "loss": 0.1879,
+      "step": 1780
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.8146488666534424,
+      "learning_rate": 2.39574926015604e-05,
+      "loss": 0.1901,
+      "step": 1785
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.1626732349395752,
+      "learning_rate": 2.4024751143395214e-05,
+      "loss": 0.2104,
+      "step": 1790
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.13757061958313,
+      "learning_rate": 2.4092009685230024e-05,
+      "loss": 0.206,
+      "step": 1795
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.9411909580230713,
+      "learning_rate": 2.4159268227064837e-05,
+      "loss": 0.2188,
+      "step": 1800
+    },
+    {
+      "epoch": 0.68,
+      "eval_loss": 0.18140146136283875,
+      "eval_runtime": 357.1907,
+      "eval_samples_per_second": 63.806,
+      "eval_steps_per_second": 7.976,
+      "eval_wer": 0.160156508503739,
+      "step": 1800
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.159300446510315,
+      "learning_rate": 2.422652676889965e-05,
+      "loss": 0.175,
+      "step": 1805
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.4202321767807007,
+      "learning_rate": 2.4293785310734467e-05,
+      "loss": 0.2129,
+      "step": 1810
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.9339726567268372,
+      "learning_rate": 2.4361043852569277e-05,
+      "loss": 0.2122,
+      "step": 1815
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.4866915941238403,
+      "learning_rate": 2.442830239440409e-05,
+      "loss": 0.2287,
+      "step": 1820
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.9092835783958435,
+      "learning_rate": 2.4495560936238903e-05,
+      "loss": 0.1917,
+      "step": 1825
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1197502613067627,
+      "learning_rate": 2.4562819478073717e-05,
+      "loss": 0.1958,
+      "step": 1830
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9912203550338745,
+      "learning_rate": 2.463007801990853e-05,
+      "loss": 0.2005,
+      "step": 1835
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 6.548855781555176,
+      "learning_rate": 2.4697336561743343e-05,
+      "loss": 0.2213,
+      "step": 1840
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.2119795083999634,
+      "learning_rate": 2.4764595103578156e-05,
+      "loss": 0.2205,
+      "step": 1845
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.7886515855789185,
+      "learning_rate": 2.4818401937046006e-05,
+      "loss": 0.2342,
+      "step": 1850
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.2907798290252686,
+      "learning_rate": 2.488566047888082e-05,
+      "loss": 0.1897,
+      "step": 1855
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.3024115562438965,
+      "learning_rate": 2.4952919020715632e-05,
+      "loss": 0.1886,
+      "step": 1860
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.0095221996307373,
+      "learning_rate": 2.5020177562550446e-05,
+      "loss": 0.2005,
+      "step": 1865
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9537557363510132,
+      "learning_rate": 2.508743610438526e-05,
+      "loss": 0.1996,
+      "step": 1870
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.2144302129745483,
+      "learning_rate": 2.5154694646220072e-05,
+      "loss": 0.2224,
+      "step": 1875
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.864310622215271,
+      "learning_rate": 2.5221953188054885e-05,
+      "loss": 0.2359,
+      "step": 1880
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.8763705492019653,
+      "learning_rate": 2.5289211729889695e-05,
+      "loss": 0.1772,
+      "step": 1885
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.2501022815704346,
+      "learning_rate": 2.535647027172451e-05,
+      "loss": 0.1973,
+      "step": 1890
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 22.93429946899414,
+      "learning_rate": 2.5423728813559322e-05,
+      "loss": 0.2025,
+      "step": 1895
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.8246068954467773,
+      "learning_rate": 2.5490987355394135e-05,
+      "loss": 0.2578,
+      "step": 1900
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.5132076740264893,
+      "learning_rate": 2.5558245897228948e-05,
+      "loss": 0.191,
+      "step": 1905
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.2082728147506714,
+      "learning_rate": 2.5625504439063765e-05,
+      "loss": 0.205,
+      "step": 1910
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.7007734775543213,
+      "learning_rate": 2.5692762980898578e-05,
+      "loss": 0.2016,
+      "step": 1915
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.175850868225098,
+      "learning_rate": 2.576002152273339e-05,
+      "loss": 0.2172,
+      "step": 1920
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.0627433061599731,
+      "learning_rate": 2.5827280064568198e-05,
+      "loss": 0.1878,
+      "step": 1925
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.005064845085144,
+      "learning_rate": 2.589453860640301e-05,
+      "loss": 0.2042,
+      "step": 1930
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.762886643409729,
+      "learning_rate": 2.5961797148237828e-05,
+      "loss": 0.2334,
+      "step": 1935
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.3690932989120483,
+      "learning_rate": 2.602905569007264e-05,
+      "loss": 0.1999,
+      "step": 1940
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.8110887408256531,
+      "learning_rate": 2.6096314231907454e-05,
+      "loss": 0.198,
+      "step": 1945
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.4105325937271118,
+      "learning_rate": 2.6163572773742268e-05,
+      "loss": 0.1984,
+      "step": 1950
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.1216415166854858,
+      "learning_rate": 2.623083131557708e-05,
+      "loss": 0.2031,
+      "step": 1955
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.2657769918441772,
+      "learning_rate": 2.6298089857411894e-05,
+      "loss": 0.1897,
+      "step": 1960
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.246349573135376,
+      "learning_rate": 2.6365348399246704e-05,
+      "loss": 0.1596,
+      "step": 1965
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.0972946882247925,
+      "learning_rate": 2.6432606941081517e-05,
+      "loss": 0.2206,
+      "step": 1970
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.0011210441589355,
+      "learning_rate": 2.649986548291633e-05,
+      "loss": 0.2094,
+      "step": 1975
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.6088778972625732,
+      "learning_rate": 2.6567124024751144e-05,
+      "loss": 0.1945,
+      "step": 1980
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.121163010597229,
+      "learning_rate": 2.6634382566585957e-05,
+      "loss": 0.1911,
+      "step": 1985
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.4630978107452393,
+      "learning_rate": 2.670164110842077e-05,
+      "loss": 0.2013,
+      "step": 1990
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 15.178384780883789,
+      "learning_rate": 2.6768899650255587e-05,
+      "loss": 0.1768,
+      "step": 1995
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.4255266189575195,
+      "learning_rate": 2.68361581920904e-05,
+      "loss": 0.1948,
+      "step": 2000
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.9294538497924805,
+      "learning_rate": 2.6903416733925207e-05,
+      "loss": 0.2167,
+      "step": 2005
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.1127851009368896,
+      "learning_rate": 2.697067527576002e-05,
+      "loss": 0.251,
+      "step": 2010
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7663888931274414,
+      "learning_rate": 2.7037933817594833e-05,
+      "loss": 0.1848,
+      "step": 2015
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.2002959251403809,
+      "learning_rate": 2.710519235942965e-05,
+      "loss": 0.203,
+      "step": 2020
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.9071945548057556,
+      "learning_rate": 2.7172450901264463e-05,
+      "loss": 0.1841,
+      "step": 2025
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.0894787311553955,
+      "learning_rate": 2.7239709443099276e-05,
+      "loss": 0.203,
+      "step": 2030
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.9670848846435547,
+      "learning_rate": 2.730696798493409e-05,
+      "loss": 0.2326,
+      "step": 2035
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.8053321838378906,
+      "learning_rate": 2.7374226526768903e-05,
+      "loss": 0.1884,
+      "step": 2040
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.9916960597038269,
+      "learning_rate": 2.7441485068603713e-05,
+      "loss": 0.1843,
+      "step": 2045
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.6158232688903809,
+      "learning_rate": 2.7508743610438526e-05,
+      "loss": 0.2108,
+      "step": 2050
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.4162769317626953,
+      "learning_rate": 2.757600215227334e-05,
+      "loss": 0.1793,
+      "step": 2055
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.0673177242279053,
+      "learning_rate": 2.7643260694108153e-05,
+      "loss": 0.1714,
+      "step": 2060
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.2585564851760864,
+      "learning_rate": 2.7710519235942966e-05,
+      "loss": 0.1737,
+      "step": 2065
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.9342767596244812,
+      "learning_rate": 2.777777777777778e-05,
+      "loss": 0.1738,
+      "step": 2070
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.80026912689209,
+      "learning_rate": 2.7845036319612596e-05,
+      "loss": 0.1997,
+      "step": 2075
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.875598669052124,
+      "learning_rate": 2.791229486144741e-05,
+      "loss": 0.1795,
+      "step": 2080
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.5565993785858154,
+      "learning_rate": 2.7979553403282216e-05,
+      "loss": 0.172,
+      "step": 2085
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.0654990673065186,
+      "learning_rate": 2.804681194511703e-05,
+      "loss": 0.1924,
+      "step": 2090
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.200706720352173,
+      "learning_rate": 2.8114070486951842e-05,
+      "loss": 0.213,
+      "step": 2095
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.121198058128357,
+      "learning_rate": 2.818132902878666e-05,
+      "loss": 0.1983,
+      "step": 2100
+    },
+    {
+      "epoch": 0.79,
+      "eval_loss": 0.17107431590557098,
+      "eval_runtime": 357.5488,
+      "eval_samples_per_second": 63.742,
+      "eval_steps_per_second": 7.968,
+      "eval_wer": 0.14772764873271127,
+      "step": 2100
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.698148488998413,
+      "learning_rate": 2.8248587570621472e-05,
+      "loss": 0.2128,
+      "step": 2105
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6483871936798096,
+      "learning_rate": 2.8315846112456285e-05,
+      "loss": 0.2004,
+      "step": 2110
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.3125154972076416,
+      "learning_rate": 2.83831046542911e-05,
+      "loss": 0.1958,
+      "step": 2115
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.06911301612854,
+      "learning_rate": 2.8450363196125912e-05,
+      "loss": 0.1793,
+      "step": 2120
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.083073616027832,
+      "learning_rate": 2.851762173796072e-05,
+      "loss": 0.196,
+      "step": 2125
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.4168145656585693,
+      "learning_rate": 2.8584880279795535e-05,
+      "loss": 0.1929,
+      "step": 2130
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.41092050075531,
+      "learning_rate": 2.8652138821630348e-05,
+      "loss": 0.1795,
+      "step": 2135
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.1256417036056519,
+      "learning_rate": 2.871939736346516e-05,
+      "loss": 0.2109,
+      "step": 2140
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8200010657310486,
+      "learning_rate": 2.8786655905299975e-05,
+      "loss": 0.1891,
+      "step": 2145
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.943932294845581,
+      "learning_rate": 2.8853914447134788e-05,
+      "loss": 0.1844,
+      "step": 2150
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.838498592376709,
+      "learning_rate": 2.89211729889696e-05,
+      "loss": 0.2097,
+      "step": 2155
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.3750524520874023,
+      "learning_rate": 2.8988431530804418e-05,
+      "loss": 0.1819,
+      "step": 2160
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.2923299074172974,
+      "learning_rate": 2.9055690072639224e-05,
+      "loss": 0.2085,
+      "step": 2165
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.9227744340896606,
+      "learning_rate": 2.9122948614474038e-05,
+      "loss": 0.1946,
+      "step": 2170
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.8773796558380127,
+      "learning_rate": 2.919020715630885e-05,
+      "loss": 0.2224,
+      "step": 2175
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.422987937927246,
+      "learning_rate": 2.9257465698143664e-05,
+      "loss": 0.1934,
+      "step": 2180
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7351481318473816,
+      "learning_rate": 2.932472423997848e-05,
+      "loss": 0.1893,
+      "step": 2185
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.9060851335525513,
+      "learning_rate": 2.9391982781813294e-05,
+      "loss": 0.2107,
+      "step": 2190
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.9987771511077881,
+      "learning_rate": 2.9459241323648107e-05,
+      "loss": 0.1912,
+      "step": 2195
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8838289976119995,
+      "learning_rate": 2.9526499865482914e-05,
+      "loss": 0.1716,
+      "step": 2200
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.9973472952842712,
+      "learning_rate": 2.959375840731773e-05,
+      "loss": 0.2313,
+      "step": 2205
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.138644218444824,
+      "learning_rate": 2.9661016949152544e-05,
+      "loss": 0.1959,
+      "step": 2210
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8018471598625183,
+      "learning_rate": 2.9728275490987357e-05,
+      "loss": 0.1759,
+      "step": 2215
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.042300820350647,
+      "learning_rate": 2.979553403282217e-05,
+      "loss": 0.2003,
+      "step": 2220
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8881530165672302,
+      "learning_rate": 2.9862792574656983e-05,
+      "loss": 0.1857,
+      "step": 2225
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.0730048418045044,
+      "learning_rate": 2.9930051116491797e-05,
+      "loss": 0.1803,
+      "step": 2230
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.26365327835083,
+      "learning_rate": 2.999730965832661e-05,
+      "loss": 0.2044,
+      "step": 2235
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.454659342765808,
+      "learning_rate": 3.006456820016142e-05,
+      "loss": 0.1901,
+      "step": 2240
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.56609308719635,
+      "learning_rate": 3.0131826741996233e-05,
+      "loss": 0.2159,
+      "step": 2245
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.6059943437576294,
+      "learning_rate": 3.0199085283831046e-05,
+      "loss": 0.2086,
+      "step": 2250
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.6269097328186035,
+      "learning_rate": 3.026634382566586e-05,
+      "loss": 0.1579,
+      "step": 2255
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.079280376434326,
+      "learning_rate": 3.0333602367500673e-05,
+      "loss": 0.214,
+      "step": 2260
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.8409821391105652,
+      "learning_rate": 3.040086090933549e-05,
+      "loss": 0.1765,
+      "step": 2265
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.9875451326370239,
+      "learning_rate": 3.0468119451170303e-05,
+      "loss": 0.1918,
+      "step": 2270
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.1287016868591309,
+      "learning_rate": 3.053537799300511e-05,
+      "loss": 0.1838,
+      "step": 2275
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.101073980331421,
+      "learning_rate": 3.060263653483992e-05,
+      "loss": 0.2051,
+      "step": 2280
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.8420178890228271,
+      "learning_rate": 3.066989507667474e-05,
+      "loss": 0.1932,
+      "step": 2285
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.9675130844116211,
+      "learning_rate": 3.073715361850955e-05,
+      "loss": 0.1829,
+      "step": 2290
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.8261280655860901,
+      "learning_rate": 3.0804412160344366e-05,
+      "loss": 0.1823,
+      "step": 2295
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.1558027267456055,
+      "learning_rate": 3.0871670702179176e-05,
+      "loss": 0.2087,
+      "step": 2300
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.086066722869873,
+      "learning_rate": 3.093892924401399e-05,
+      "loss": 0.1922,
+      "step": 2305
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.0070490837097168,
+      "learning_rate": 3.100618778584881e-05,
+      "loss": 0.1895,
+      "step": 2310
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.0676946640014648,
+      "learning_rate": 3.107344632768362e-05,
+      "loss": 0.1771,
+      "step": 2315
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.0908974409103394,
+      "learning_rate": 3.114070486951843e-05,
+      "loss": 0.1812,
+      "step": 2320
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.0786138772964478,
+      "learning_rate": 3.120796341135324e-05,
+      "loss": 0.2001,
+      "step": 2325
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.9205860495567322,
+      "learning_rate": 3.1275221953188055e-05,
+      "loss": 0.1985,
+      "step": 2330
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.5036931037902832,
+      "learning_rate": 3.134248049502287e-05,
+      "loss": 0.2025,
+      "step": 2335
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.8630493879318237,
+      "learning_rate": 3.140973903685768e-05,
+      "loss": 0.1759,
+      "step": 2340
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.3396941423416138,
+      "learning_rate": 3.14769975786925e-05,
+      "loss": 0.1882,
+      "step": 2345
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.6374388933181763,
+      "learning_rate": 3.154425612052731e-05,
+      "loss": 0.1706,
+      "step": 2350
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.9011550545692444,
+      "learning_rate": 3.1611514662362125e-05,
+      "loss": 0.1713,
+      "step": 2355
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.356629729270935,
+      "learning_rate": 3.1678773204196935e-05,
+      "loss": 0.1889,
+      "step": 2360
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.3063596487045288,
+      "learning_rate": 3.1746031746031745e-05,
+      "loss": 0.1949,
+      "step": 2365
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.447098970413208,
+      "learning_rate": 3.181329028786656e-05,
+      "loss": 0.206,
+      "step": 2370
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.135978102684021,
+      "learning_rate": 3.188054882970137e-05,
+      "loss": 0.2009,
+      "step": 2375
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.4577934741973877,
+      "learning_rate": 3.194780737153619e-05,
+      "loss": 0.2074,
+      "step": 2380
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.6658282279968262,
+      "learning_rate": 3.2015065913371e-05,
+      "loss": 0.1861,
+      "step": 2385
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.3513625860214233,
+      "learning_rate": 3.2082324455205814e-05,
+      "loss": 0.2266,
+      "step": 2390
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.197036623954773,
+      "learning_rate": 3.214958299704063e-05,
+      "loss": 0.2198,
+      "step": 2395
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.0936663150787354,
+      "learning_rate": 3.2216841538875434e-05,
+      "loss": 0.1796,
+      "step": 2400
+    },
+    {
+      "epoch": 0.9,
+      "eval_loss": 0.1664888709783554,
+      "eval_runtime": 343.3008,
+      "eval_samples_per_second": 66.388,
+      "eval_steps_per_second": 8.299,
+      "eval_wer": 0.14263615908940508,
+      "step": 2400
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.5868946313858032,
+      "learning_rate": 3.228410008071025e-05,
+      "loss": 0.1753,
+      "step": 2405
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.086012601852417,
+      "learning_rate": 3.235135862254506e-05,
+      "loss": 0.195,
+      "step": 2410
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.1508128643035889,
+      "learning_rate": 3.241861716437988e-05,
+      "loss": 0.2205,
+      "step": 2415
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8650568723678589,
+      "learning_rate": 3.2485875706214694e-05,
+      "loss": 0.1753,
+      "step": 2420
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.1367069482803345,
+      "learning_rate": 3.2553134248049504e-05,
+      "loss": 0.2035,
+      "step": 2425
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.267732620239258,
+      "learning_rate": 3.262039278988432e-05,
+      "loss": 0.1951,
+      "step": 2430
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.8955305814743042,
+      "learning_rate": 3.268765133171913e-05,
+      "loss": 0.1751,
+      "step": 2435
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.737474799156189,
+      "learning_rate": 3.275490987355394e-05,
+      "loss": 0.174,
+      "step": 2440
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.9678319692611694,
+      "learning_rate": 3.282216841538876e-05,
+      "loss": 0.1909,
+      "step": 2445
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.0070288181304932,
+      "learning_rate": 3.288942695722357e-05,
+      "loss": 0.1868,
+      "step": 2450
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.5490509271621704,
+      "learning_rate": 3.295668549905838e-05,
+      "loss": 0.1892,
+      "step": 2455
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.0689626932144165,
+      "learning_rate": 3.302394404089319e-05,
+      "loss": 0.1764,
+      "step": 2460
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.3539146184921265,
+      "learning_rate": 3.309120258272801e-05,
+      "loss": 0.1829,
+      "step": 2465
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.9297118186950684,
+      "learning_rate": 3.315846112456282e-05,
+      "loss": 0.2051,
+      "step": 2470
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.45232093334198,
+      "learning_rate": 3.3225719666397636e-05,
+      "loss": 0.2083,
+      "step": 2475
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.2929126024246216,
+      "learning_rate": 3.3292978208232446e-05,
+      "loss": 0.1801,
+      "step": 2480
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.8578481674194336,
+      "learning_rate": 3.3360236750067256e-05,
+      "loss": 0.193,
+      "step": 2485
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.8273435831069946,
+      "learning_rate": 3.342749529190207e-05,
+      "loss": 0.2083,
+      "step": 2490
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.3022762537002563,
+      "learning_rate": 3.349475383373688e-05,
+      "loss": 0.2044,
+      "step": 2495
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.0528521537780762,
+      "learning_rate": 3.35620123755717e-05,
+      "loss": 0.1811,
+      "step": 2500
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.8947622776031494,
+      "learning_rate": 3.3629270917406516e-05,
+      "loss": 0.1649,
+      "step": 2505
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.297065258026123,
+      "learning_rate": 3.3696529459241326e-05,
+      "loss": 0.2006,
+      "step": 2510
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.9753395915031433,
+      "learning_rate": 3.376378800107614e-05,
+      "loss": 0.1864,
+      "step": 2515
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.9703760147094727,
+      "learning_rate": 3.383104654291095e-05,
+      "loss": 0.195,
+      "step": 2520
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.5742170810699463,
+      "learning_rate": 3.389830508474576e-05,
+      "loss": 0.2144,
+      "step": 2525
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.9958549737930298,
+      "learning_rate": 3.396556362658058e-05,
+      "loss": 0.1655,
+      "step": 2530
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7199857831001282,
+      "learning_rate": 3.403282216841539e-05,
+      "loss": 0.2055,
+      "step": 2535
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.229958176612854,
+      "learning_rate": 3.4100080710250205e-05,
+      "loss": 0.1786,
+      "step": 2540
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.6161264181137085,
+      "learning_rate": 3.4167339252085015e-05,
+      "loss": 0.1846,
+      "step": 2545
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5578131675720215,
+      "learning_rate": 3.423459779391983e-05,
+      "loss": 0.1921,
+      "step": 2550
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8049275279045105,
+      "learning_rate": 3.430185633575464e-05,
+      "loss": 0.1742,
+      "step": 2555
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.0607887506484985,
+      "learning_rate": 3.436911487758945e-05,
+      "loss": 0.1853,
+      "step": 2560
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.642946720123291,
+      "learning_rate": 3.443637341942427e-05,
+      "loss": 0.1646,
+      "step": 2565
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.9879059195518494,
+      "learning_rate": 3.450363196125908e-05,
+      "loss": 0.2247,
+      "step": 2570
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.1038026809692383,
+      "learning_rate": 3.4570890503093895e-05,
+      "loss": 0.1896,
+      "step": 2575
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.1367892026901245,
+      "learning_rate": 3.463814904492871e-05,
+      "loss": 0.1791,
+      "step": 2580
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.2722300291061401,
+      "learning_rate": 3.470540758676352e-05,
+      "loss": 0.1883,
+      "step": 2585
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.9158148765563965,
+      "learning_rate": 3.477266612859834e-05,
+      "loss": 0.1606,
+      "step": 2590
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.8352373242378235,
+      "learning_rate": 3.483992467043315e-05,
+      "loss": 0.1554,
+      "step": 2595
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.0104358196258545,
+      "learning_rate": 3.490718321226796e-05,
+      "loss": 0.1809,
+      "step": 2600
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.5048420429229736,
+      "learning_rate": 3.4974441754102774e-05,
+      "loss": 0.1689,
+      "step": 2605
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 33.753849029541016,
+      "learning_rate": 3.5041700295937584e-05,
+      "loss": 0.1777,
+      "step": 2610
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.0280221700668335,
+      "learning_rate": 3.51089588377724e-05,
+      "loss": 0.1586,
+      "step": 2615
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.1569583415985107,
+      "learning_rate": 3.517621737960721e-05,
+      "loss": 0.1728,
+      "step": 2620
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.9094194173812866,
+      "learning_rate": 3.524347592144203e-05,
+      "loss": 0.1834,
+      "step": 2625
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.4927579164505005,
+      "learning_rate": 3.531073446327684e-05,
+      "loss": 0.174,
+      "step": 2630
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.1765927076339722,
+      "learning_rate": 3.537799300511165e-05,
+      "loss": 0.1837,
+      "step": 2635
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.1112360954284668,
+      "learning_rate": 3.5445251546946464e-05,
+      "loss": 0.2069,
+      "step": 2640
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.883026123046875,
+      "learning_rate": 3.5512510088781274e-05,
+      "loss": 0.1965,
+      "step": 2645
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.2431875467300415,
+      "learning_rate": 3.557976863061609e-05,
+      "loss": 0.1689,
+      "step": 2650
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.2642476558685303,
+      "learning_rate": 3.56470271724509e-05,
+      "loss": 0.1772,
+      "step": 2655
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7983075976371765,
+      "learning_rate": 3.571428571428572e-05,
+      "loss": 0.1656,
+      "step": 2660
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.027163028717041,
+      "learning_rate": 3.5781544256120534e-05,
+      "loss": 0.1876,
+      "step": 2665
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.603554606437683,
+      "learning_rate": 3.5848802797955343e-05,
+      "loss": 0.162,
+      "step": 2670
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.8891953229904175,
+      "learning_rate": 3.591606133979015e-05,
+      "loss": 0.178,
+      "step": 2675
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.9294222593307495,
+      "learning_rate": 3.598331988162496e-05,
+      "loss": 0.1607,
+      "step": 2680
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.1016396284103394,
+      "learning_rate": 3.605057842345978e-05,
+      "loss": 0.1906,
+      "step": 2685
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.8925282955169678,
+      "learning_rate": 3.6117836965294597e-05,
+      "loss": 0.1887,
+      "step": 2690
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.0900254249572754,
+      "learning_rate": 3.6185095507129406e-05,
+      "loss": 0.1656,
+      "step": 2695
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.0671287775039673,
+      "learning_rate": 3.625235404896422e-05,
+      "loss": 0.1569,
+      "step": 2700
+    },
+    {
+      "epoch": 1.02,
+      "eval_loss": 0.1641453206539154,
+      "eval_runtime": 342.5844,
+      "eval_samples_per_second": 66.527,
+      "eval_steps_per_second": 8.316,
+      "eval_wer": 0.1324759281318245,
+      "step": 2700
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.9765105247497559,
+      "learning_rate": 3.631961259079903e-05,
+      "loss": 0.1535,
+      "step": 2705
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.6181786060333252,
+      "learning_rate": 3.638687113263385e-05,
+      "loss": 0.179,
+      "step": 2710
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.7979306578636169,
+      "learning_rate": 3.645412967446866e-05,
+      "loss": 0.1515,
+      "step": 2715
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.9192932844161987,
+      "learning_rate": 3.652138821630347e-05,
+      "loss": 0.179,
+      "step": 2720
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.1763132810592651,
+      "learning_rate": 3.6588646758138286e-05,
+      "loss": 0.1665,
+      "step": 2725
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 2.1913676261901855,
+      "learning_rate": 3.6655905299973096e-05,
+      "loss": 0.194,
+      "step": 2730
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.9093385338783264,
+      "learning_rate": 3.672316384180791e-05,
+      "loss": 0.1419,
+      "step": 2735
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.7085216641426086,
+      "learning_rate": 3.679042238364272e-05,
+      "loss": 0.1745,
+      "step": 2740
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.9004052877426147,
+      "learning_rate": 3.685768092547754e-05,
+      "loss": 0.168,
+      "step": 2745
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.1891593933105469,
+      "learning_rate": 3.691148775894539e-05,
+      "loss": 0.2122,
+      "step": 2750
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.9683598279953003,
+      "learning_rate": 3.69787463007802e-05,
+      "loss": 0.1763,
+      "step": 2755
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.9511354565620422,
+      "learning_rate": 3.7046004842615015e-05,
+      "loss": 0.1612,
+      "step": 2760
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.988720178604126,
+      "learning_rate": 3.7113263384449825e-05,
+      "loss": 0.1932,
+      "step": 2765
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.9230719208717346,
+      "learning_rate": 3.718052192628464e-05,
+      "loss": 0.1661,
+      "step": 2770
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.6773403286933899,
+      "learning_rate": 3.724778046811945e-05,
+      "loss": 0.1525,
+      "step": 2775
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.9711774587631226,
+      "learning_rate": 3.731503900995427e-05,
+      "loss": 0.1839,
+      "step": 2780
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.9966487288475037,
+      "learning_rate": 3.738229755178908e-05,
+      "loss": 0.1659,
+      "step": 2785
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 12.543924331665039,
+      "learning_rate": 3.7449556093623894e-05,
+      "loss": 0.1952,
+      "step": 2790
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.082942008972168,
+      "learning_rate": 3.7516814635458704e-05,
+      "loss": 0.181,
+      "step": 2795
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.083632469177246,
+      "learning_rate": 3.7584073177293514e-05,
+      "loss": 0.1506,
+      "step": 2800
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.0529345273971558,
+      "learning_rate": 3.765133171912833e-05,
+      "loss": 0.1702,
+      "step": 2805
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.9386705756187439,
+      "learning_rate": 3.771859026096314e-05,
+      "loss": 0.1639,
+      "step": 2810
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.7492910623550415,
+      "learning_rate": 3.778584880279796e-05,
+      "loss": 0.1641,
+      "step": 2815
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.3244742155075073,
+      "learning_rate": 3.7853107344632774e-05,
+      "loss": 0.1585,
+      "step": 2820
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.6878337860107422,
+      "learning_rate": 3.7920365886467584e-05,
+      "loss": 0.2087,
+      "step": 2825
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 2.5463967323303223,
+      "learning_rate": 3.79876244283024e-05,
+      "loss": 0.2073,
+      "step": 2830
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 5.008793354034424,
+      "learning_rate": 3.8054882970137204e-05,
+      "loss": 0.2131,
+      "step": 2835
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.333203911781311,
+      "learning_rate": 3.812214151197202e-05,
+      "loss": 0.162,
+      "step": 2840
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.6690267324447632,
+      "learning_rate": 3.818940005380684e-05,
+      "loss": 0.2104,
+      "step": 2845
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.0743372440338135,
+      "learning_rate": 3.825665859564165e-05,
+      "loss": 0.1584,
+      "step": 2850
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.215438723564148,
+      "learning_rate": 3.8323917137476463e-05,
+      "loss": 0.1547,
+      "step": 2855
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.972049355506897,
+      "learning_rate": 3.839117567931127e-05,
+      "loss": 0.1642,
+      "step": 2860
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.4244087934494019,
+      "learning_rate": 3.844498251277913e-05,
+      "loss": 0.2157,
+      "step": 2865
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.9279956817626953,
+      "learning_rate": 3.851224105461394e-05,
+      "loss": 0.1593,
+      "step": 2870
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.1965911388397217,
+      "learning_rate": 3.857949959644875e-05,
+      "loss": 0.1913,
+      "step": 2875
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.0124505758285522,
+      "learning_rate": 3.864675813828356e-05,
+      "loss": 0.1703,
+      "step": 2880
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 2.8864803314208984,
+      "learning_rate": 3.8714016680118376e-05,
+      "loss": 0.2079,
+      "step": 2885
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.0078617334365845,
+      "learning_rate": 3.878127522195319e-05,
+      "loss": 0.159,
+      "step": 2890
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.9530019760131836,
+      "learning_rate": 3.8848533763788e-05,
+      "loss": 0.168,
+      "step": 2895
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.0602315664291382,
+      "learning_rate": 3.891579230562282e-05,
+      "loss": 0.1639,
+      "step": 2900
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.8479179739952087,
+      "learning_rate": 3.898305084745763e-05,
+      "loss": 0.1598,
+      "step": 2905
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.3728115558624268,
+      "learning_rate": 3.9050309389292445e-05,
+      "loss": 0.1594,
+      "step": 2910
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.9466823935508728,
+      "learning_rate": 3.9117567931127255e-05,
+      "loss": 0.1606,
+      "step": 2915
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.225555658340454,
+      "learning_rate": 3.9184826472962065e-05,
+      "loss": 0.1489,
+      "step": 2920
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.3602200746536255,
+      "learning_rate": 3.925208501479688e-05,
+      "loss": 0.1845,
+      "step": 2925
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.9880752563476562,
+      "learning_rate": 3.931934355663169e-05,
+      "loss": 0.1805,
+      "step": 2930
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.7916092276573181,
+      "learning_rate": 3.938660209846651e-05,
+      "loss": 0.1583,
+      "step": 2935
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.7080309391021729,
+      "learning_rate": 3.945386064030132e-05,
+      "loss": 0.1658,
+      "step": 2940
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.2263010740280151,
+      "learning_rate": 3.9521119182136135e-05,
+      "loss": 0.1681,
+      "step": 2945
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 2.5591046810150146,
+      "learning_rate": 3.958837772397095e-05,
+      "loss": 0.1845,
+      "step": 2950
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.0637853145599365,
+      "learning_rate": 3.9655636265805754e-05,
+      "loss": 0.1771,
+      "step": 2955
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.2432622909545898,
+      "learning_rate": 3.972289480764057e-05,
+      "loss": 0.1902,
+      "step": 2960
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.3372734785079956,
+      "learning_rate": 3.979015334947538e-05,
+      "loss": 0.1743,
+      "step": 2965
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.0920542478561401,
+      "learning_rate": 3.98574118913102e-05,
+      "loss": 0.1939,
+      "step": 2970
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.9923748970031738,
+      "learning_rate": 3.9924670433145014e-05,
+      "loss": 0.1596,
+      "step": 2975
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 3.2066454887390137,
+      "learning_rate": 3.9991928974979824e-05,
+      "loss": 0.191,
+      "step": 2980
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 3.276280403137207,
+      "learning_rate": 4.005918751681464e-05,
+      "loss": 0.1767,
+      "step": 2985
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.5679292678833008,
+      "learning_rate": 4.012644605864945e-05,
+      "loss": 0.1814,
+      "step": 2990
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.3838121891021729,
+      "learning_rate": 4.019370460048426e-05,
+      "loss": 0.1894,
+      "step": 2995
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.2008754014968872,
+      "learning_rate": 4.026096314231908e-05,
+      "loss": 0.1776,
+      "step": 3000
+    },
+    {
+      "epoch": 1.13,
+      "eval_loss": 0.15891925990581512,
+      "eval_runtime": 350.9783,
+      "eval_samples_per_second": 64.936,
+      "eval_steps_per_second": 8.117,
+      "eval_wer": 0.1407211633909073,
+      "step": 3000
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.1781939268112183,
+      "learning_rate": 4.032822168415389e-05,
+      "loss": 0.1655,
+      "step": 3005
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.136367678642273,
+      "learning_rate": 4.0395480225988704e-05,
+      "loss": 0.1557,
+      "step": 3010
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 3.3085150718688965,
+      "learning_rate": 4.0462738767823514e-05,
+      "loss": 0.1699,
+      "step": 3015
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.0148067474365234,
+      "learning_rate": 4.052999730965833e-05,
+      "loss": 0.1619,
+      "step": 3020
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 7.891255855560303,
+      "learning_rate": 4.059725585149314e-05,
+      "loss": 0.223,
+      "step": 3025
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.9084330797195435,
+      "learning_rate": 4.066451439332796e-05,
+      "loss": 0.1896,
+      "step": 3030
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.5072202682495117,
+      "learning_rate": 4.073177293516277e-05,
+      "loss": 0.1835,
+      "step": 3035
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.1229376792907715,
+      "learning_rate": 4.0799031476997577e-05,
+      "loss": 0.1674,
+      "step": 3040
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.7050045728683472,
+      "learning_rate": 4.086629001883239e-05,
+      "loss": 0.1836,
+      "step": 3045
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.593251347541809,
+      "learning_rate": 4.09335485606672e-05,
+      "loss": 0.1966,
+      "step": 3050
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.1783024072647095,
+      "learning_rate": 4.100080710250202e-05,
+      "loss": 0.1776,
+      "step": 3055
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.8477684259414673,
+      "learning_rate": 4.1068065644336836e-05,
+      "loss": 0.1817,
+      "step": 3060
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.439737319946289,
+      "learning_rate": 4.1135324186171646e-05,
+      "loss": 0.1605,
+      "step": 3065
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.5581740140914917,
+      "learning_rate": 4.120258272800646e-05,
+      "loss": 0.192,
+      "step": 3070
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.8211709856987,
+      "learning_rate": 4.126984126984127e-05,
+      "loss": 0.1467,
+      "step": 3075
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 3.331674098968506,
+      "learning_rate": 4.133709981167608e-05,
+      "loss": 0.2015,
+      "step": 3080
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.7677055597305298,
+      "learning_rate": 4.14043583535109e-05,
+      "loss": 0.1809,
+      "step": 3085
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.2478415966033936,
+      "learning_rate": 4.147161689534571e-05,
+      "loss": 0.1714,
+      "step": 3090
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.2321388721466064,
+      "learning_rate": 4.1538875437180526e-05,
+      "loss": 0.1506,
+      "step": 3095
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.3224626779556274,
+      "learning_rate": 4.1606133979015336e-05,
+      "loss": 0.1733,
+      "step": 3100
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 2.080465316772461,
+      "learning_rate": 4.167339252085015e-05,
+      "loss": 0.1614,
+      "step": 3105
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.4414575099945068,
+      "learning_rate": 4.174065106268496e-05,
+      "loss": 0.1934,
+      "step": 3110
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.0855170488357544,
+      "learning_rate": 4.180790960451977e-05,
+      "loss": 0.1712,
+      "step": 3115
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.4533028602600098,
+      "learning_rate": 4.187516814635459e-05,
+      "loss": 0.1838,
+      "step": 3120
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 3.967257261276245,
+      "learning_rate": 4.19424266881894e-05,
+      "loss": 0.1838,
+      "step": 3125
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.1230443716049194,
+      "learning_rate": 4.2009685230024215e-05,
+      "loss": 0.1709,
+      "step": 3130
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.5892051458358765,
+      "learning_rate": 4.207694377185903e-05,
+      "loss": 0.1787,
+      "step": 3135
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.088104009628296,
+      "learning_rate": 4.214420231369384e-05,
+      "loss": 0.1623,
+      "step": 3140
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.6897289752960205,
+      "learning_rate": 4.221146085552866e-05,
+      "loss": 0.1962,
+      "step": 3145
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.2557629346847534,
+      "learning_rate": 4.227871939736347e-05,
+      "loss": 0.1816,
+      "step": 3150
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.5183990001678467,
+      "learning_rate": 4.234597793919828e-05,
+      "loss": 0.1726,
+      "step": 3155
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.9610652327537537,
+      "learning_rate": 4.2413236481033095e-05,
+      "loss": 0.1707,
+      "step": 3160
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.1233092546463013,
+      "learning_rate": 4.2480495022867905e-05,
+      "loss": 0.1839,
+      "step": 3165
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.1543560028076172,
+      "learning_rate": 4.254775356470272e-05,
+      "loss": 0.1772,
+      "step": 3170
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.8285064697265625,
+      "learning_rate": 4.261501210653753e-05,
+      "loss": 0.1795,
+      "step": 3175
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.0327891111373901,
+      "learning_rate": 4.268227064837235e-05,
+      "loss": 0.1899,
+      "step": 3180
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.5567407608032227,
+      "learning_rate": 4.274952919020716e-05,
+      "loss": 0.2047,
+      "step": 3185
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.9652482867240906,
+      "learning_rate": 4.2816787732041974e-05,
+      "loss": 0.1521,
+      "step": 3190
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.0588191747665405,
+      "learning_rate": 4.2884046273876784e-05,
+      "loss": 0.1515,
+      "step": 3195
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.8769431114196777,
+      "learning_rate": 4.2951304815711594e-05,
+      "loss": 0.1735,
+      "step": 3200
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.0823622941970825,
+      "learning_rate": 4.301856335754641e-05,
+      "loss": 0.1808,
+      "step": 3205
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 2.161386489868164,
+      "learning_rate": 4.308582189938122e-05,
+      "loss": 0.1869,
+      "step": 3210
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.6155903339385986,
+      "learning_rate": 4.315308044121604e-05,
+      "loss": 0.1821,
+      "step": 3215
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.9346185326576233,
+      "learning_rate": 4.3220338983050854e-05,
+      "loss": 0.1631,
+      "step": 3220
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.9511622190475464,
+      "learning_rate": 4.3287597524885664e-05,
+      "loss": 0.165,
+      "step": 3225
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.2343895435333252,
+      "learning_rate": 4.3354856066720474e-05,
+      "loss": 0.1876,
+      "step": 3230
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.6236696243286133,
+      "learning_rate": 4.3422114608555284e-05,
+      "loss": 0.1772,
+      "step": 3235
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.2224787473678589,
+      "learning_rate": 4.34893731503901e-05,
+      "loss": 0.1708,
+      "step": 3240
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.95892333984375,
+      "learning_rate": 4.355663169222492e-05,
+      "loss": 0.1762,
+      "step": 3245
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 2.5917956829071045,
+      "learning_rate": 4.362389023405973e-05,
+      "loss": 0.1892,
+      "step": 3250
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.2812261581420898,
+      "learning_rate": 4.3691148775894543e-05,
+      "loss": 0.1723,
+      "step": 3255
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.1995842456817627,
+      "learning_rate": 4.375840731772935e-05,
+      "loss": 0.1905,
+      "step": 3260
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.336096167564392,
+      "learning_rate": 4.382566585956417e-05,
+      "loss": 0.1788,
+      "step": 3265
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.2569266557693481,
+      "learning_rate": 4.389292440139898e-05,
+      "loss": 0.1893,
+      "step": 3270
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.218036413192749,
+      "learning_rate": 4.396018294323379e-05,
+      "loss": 0.1974,
+      "step": 3275
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.8960794806480408,
+      "learning_rate": 4.4027441485068606e-05,
+      "loss": 0.2063,
+      "step": 3280
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.9423807859420776,
+      "learning_rate": 4.4094700026903416e-05,
+      "loss": 0.1796,
+      "step": 3285
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.858022153377533,
+      "learning_rate": 4.416195856873823e-05,
+      "loss": 0.1706,
+      "step": 3290
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.3597792387008667,
+      "learning_rate": 4.422921711057304e-05,
+      "loss": 0.1514,
+      "step": 3295
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.4573347568511963,
+      "learning_rate": 4.429647565240786e-05,
+      "loss": 0.1521,
+      "step": 3300
+    },
+    {
+      "epoch": 1.24,
+      "eval_loss": 0.1759900152683258,
+      "eval_runtime": 343.8958,
+      "eval_samples_per_second": 66.273,
+      "eval_steps_per_second": 8.284,
+      "eval_wer": 0.15696553834954668,
+      "step": 3300
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.8293449282646179,
+      "learning_rate": 4.4363734194242676e-05,
+      "loss": 0.1743,
+      "step": 3305
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.9640608429908752,
+      "learning_rate": 4.443099273607748e-05,
+      "loss": 0.1758,
+      "step": 3310
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.431788444519043,
+      "learning_rate": 4.4498251277912296e-05,
+      "loss": 0.1703,
+      "step": 3315
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.161786675453186,
+      "learning_rate": 4.4565509819747106e-05,
+      "loss": 0.1786,
+      "step": 3320
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.6935396194458008,
+      "learning_rate": 4.463276836158192e-05,
+      "loss": 0.1794,
+      "step": 3325
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.1057621240615845,
+      "learning_rate": 4.470002690341674e-05,
+      "loss": 0.1782,
+      "step": 3330
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.1191602945327759,
+      "learning_rate": 4.476728544525155e-05,
+      "loss": 0.1757,
+      "step": 3335
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.2520352602005005,
+      "learning_rate": 4.4834543987086366e-05,
+      "loss": 0.182,
+      "step": 3340
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.242197871208191,
+      "learning_rate": 4.4901802528921175e-05,
+      "loss": 0.1734,
+      "step": 3345
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.9863011240959167,
+      "learning_rate": 4.4969061070755985e-05,
+      "loss": 0.156,
+      "step": 3350
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 3.0695974826812744,
+      "learning_rate": 4.50363196125908e-05,
+      "loss": 0.1903,
+      "step": 3355
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.0145655870437622,
+      "learning_rate": 4.510357815442561e-05,
+      "loss": 0.1794,
+      "step": 3360
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.0827836990356445,
+      "learning_rate": 4.517083669626043e-05,
+      "loss": 0.1765,
+      "step": 3365
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.4207769632339478,
+      "learning_rate": 4.523809523809524e-05,
+      "loss": 0.1658,
+      "step": 3370
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.0543575286865234,
+      "learning_rate": 4.5305353779930055e-05,
+      "loss": 0.1857,
+      "step": 3375
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.8837730884552002,
+      "learning_rate": 4.5372612321764865e-05,
+      "loss": 0.1893,
+      "step": 3380
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 5.119034767150879,
+      "learning_rate": 4.543987086359968e-05,
+      "loss": 0.186,
+      "step": 3385
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.3015881776809692,
+      "learning_rate": 4.550712940543449e-05,
+      "loss": 0.2072,
+      "step": 3390
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.446882963180542,
+      "learning_rate": 4.55743879472693e-05,
+      "loss": 0.2063,
+      "step": 3395
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.5108122825622559,
+      "learning_rate": 4.564164648910412e-05,
+      "loss": 0.1648,
+      "step": 3400
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.4953631162643433,
+      "learning_rate": 4.570890503093893e-05,
+      "loss": 0.1772,
+      "step": 3405
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.3225328922271729,
+      "learning_rate": 4.5776163572773744e-05,
+      "loss": 0.1792,
+      "step": 3410
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.035876750946045,
+      "learning_rate": 4.584342211460856e-05,
+      "loss": 0.1849,
+      "step": 3415
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.8946778178215027,
+      "learning_rate": 4.591068065644337e-05,
+      "loss": 0.1788,
+      "step": 3420
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.8172470331192017,
+      "learning_rate": 4.597793919827819e-05,
+      "loss": 0.159,
+      "step": 3425
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.1445375680923462,
+      "learning_rate": 4.6045197740113e-05,
+      "loss": 0.1887,
+      "step": 3430
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.3172004222869873,
+      "learning_rate": 4.611245628194781e-05,
+      "loss": 0.1749,
+      "step": 3435
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.119693636894226,
+      "learning_rate": 4.6179714823782624e-05,
+      "loss": 0.1543,
+      "step": 3440
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.424845576286316,
+      "learning_rate": 4.6246973365617434e-05,
+      "loss": 0.1729,
+      "step": 3445
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.8938009142875671,
+      "learning_rate": 4.631423190745225e-05,
+      "loss": 0.1558,
+      "step": 3450
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.1971361637115479,
+      "learning_rate": 4.638149044928706e-05,
+      "loss": 0.1635,
+      "step": 3455
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.8629947304725647,
+      "learning_rate": 4.644874899112188e-05,
+      "loss": 0.1644,
+      "step": 3460
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.3908476829528809,
+      "learning_rate": 4.651600753295669e-05,
+      "loss": 0.1599,
+      "step": 3465
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.0362669229507446,
+      "learning_rate": 4.65832660747915e-05,
+      "loss": 0.1822,
+      "step": 3470
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 13.378885269165039,
+      "learning_rate": 4.6650524616626313e-05,
+      "loss": 0.1764,
+      "step": 3475
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 2.5136680603027344,
+      "learning_rate": 4.671778315846112e-05,
+      "loss": 0.2122,
+      "step": 3480
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.920466959476471,
+      "learning_rate": 4.678504170029594e-05,
+      "loss": 0.1901,
+      "step": 3485
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.0931224822998047,
+      "learning_rate": 4.685230024213076e-05,
+      "loss": 0.1862,
+      "step": 3490
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 2.3703625202178955,
+      "learning_rate": 4.6919558783965566e-05,
+      "loss": 0.2167,
+      "step": 3495
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.3064347505569458,
+      "learning_rate": 4.698681732580038e-05,
+      "loss": 0.2133,
+      "step": 3500
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 2.3508777618408203,
+      "learning_rate": 4.705407586763519e-05,
+      "loss": 0.2176,
+      "step": 3505
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 3.372924566268921,
+      "learning_rate": 4.712133440947e-05,
+      "loss": 0.1914,
+      "step": 3510
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.356210708618164,
+      "learning_rate": 4.718859295130482e-05,
+      "loss": 0.1867,
+      "step": 3515
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.9219199419021606,
+      "learning_rate": 4.725585149313963e-05,
+      "loss": 0.1846,
+      "step": 3520
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.9569295644760132,
+      "learning_rate": 4.7323110034974446e-05,
+      "loss": 0.1657,
+      "step": 3525
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.2573210000991821,
+      "learning_rate": 4.7390368576809256e-05,
+      "loss": 0.1978,
+      "step": 3530
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.3519845008850098,
+      "learning_rate": 4.745762711864407e-05,
+      "loss": 0.1778,
+      "step": 3535
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.2347899675369263,
+      "learning_rate": 4.752488566047888e-05,
+      "loss": 0.1748,
+      "step": 3540
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.9839012622833252,
+      "learning_rate": 4.75921442023137e-05,
+      "loss": 0.1582,
+      "step": 3545
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.4501760005950928,
+      "learning_rate": 4.765940274414851e-05,
+      "loss": 0.1761,
+      "step": 3550
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 4.814388751983643,
+      "learning_rate": 4.772666128598332e-05,
+      "loss": 0.1525,
+      "step": 3555
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.3463459014892578,
+      "learning_rate": 4.7793919827818136e-05,
+      "loss": 0.1654,
+      "step": 3560
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.2179330587387085,
+      "learning_rate": 4.7861178369652945e-05,
+      "loss": 0.1666,
+      "step": 3565
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.8596850037574768,
+      "learning_rate": 4.792843691148776e-05,
+      "loss": 0.1557,
+      "step": 3570
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.174802303314209,
+      "learning_rate": 4.799569545332258e-05,
+      "loss": 0.1631,
+      "step": 3575
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.4201500415802002,
+      "learning_rate": 4.806295399515739e-05,
+      "loss": 0.1516,
+      "step": 3580
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.8228991031646729,
+      "learning_rate": 4.8130212536992205e-05,
+      "loss": 0.1754,
+      "step": 3585
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.8405484557151794,
+      "learning_rate": 4.819747107882701e-05,
+      "loss": 0.1682,
+      "step": 3590
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.7070761919021606,
+      "learning_rate": 4.8264729620661825e-05,
+      "loss": 0.1748,
+      "step": 3595
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 5.835540294647217,
+      "learning_rate": 4.833198816249664e-05,
+      "loss": 0.2282,
+      "step": 3600
+    },
+    {
+      "epoch": 1.36,
+      "eval_loss": 0.16513101756572723,
+      "eval_runtime": 360.828,
+      "eval_samples_per_second": 63.163,
+      "eval_steps_per_second": 7.896,
+      "eval_wer": 0.13878341936337768,
+      "step": 3600
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.9683034420013428,
+      "learning_rate": 4.839924670433145e-05,
+      "loss": 0.1843,
+      "step": 3605
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.6088361740112305,
+      "learning_rate": 4.846650524616627e-05,
+      "loss": 0.174,
+      "step": 3610
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.9753746390342712,
+      "learning_rate": 4.853376378800108e-05,
+      "loss": 0.1846,
+      "step": 3615
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.110234022140503,
+      "learning_rate": 4.8601022329835895e-05,
+      "loss": 0.1738,
+      "step": 3620
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.8439876437187195,
+      "learning_rate": 4.8668280871670705e-05,
+      "loss": 0.1814,
+      "step": 3625
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.9755458235740662,
+      "learning_rate": 4.8722087705138554e-05,
+      "loss": 0.1995,
+      "step": 3630
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.7656239867210388,
+      "learning_rate": 4.8789346246973364e-05,
+      "loss": 0.1659,
+      "step": 3635
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.1936314105987549,
+      "learning_rate": 4.885660478880818e-05,
+      "loss": 0.1626,
+      "step": 3640
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.2369545698165894,
+      "learning_rate": 4.8923863330643e-05,
+      "loss": 0.1904,
+      "step": 3645
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 3.2011802196502686,
+      "learning_rate": 4.899112187247781e-05,
+      "loss": 0.2002,
+      "step": 3650
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.1917798519134521,
+      "learning_rate": 4.9058380414312623e-05,
+      "loss": 0.2081,
+      "step": 3655
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.7763285040855408,
+      "learning_rate": 4.912563895614743e-05,
+      "loss": 0.1859,
+      "step": 3660
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.001961588859558,
+      "learning_rate": 4.919289749798225e-05,
+      "loss": 0.1665,
+      "step": 3665
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.377252459526062,
+      "learning_rate": 4.926015603981706e-05,
+      "loss": 0.1841,
+      "step": 3670
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.082650065422058,
+      "learning_rate": 4.932741458165187e-05,
+      "loss": 0.1992,
+      "step": 3675
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.9457229971885681,
+      "learning_rate": 4.9394673123486686e-05,
+      "loss": 0.1789,
+      "step": 3680
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 2.665595531463623,
+      "learning_rate": 4.9461931665321496e-05,
+      "loss": 0.1956,
+      "step": 3685
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.1224653720855713,
+      "learning_rate": 4.952919020715631e-05,
+      "loss": 0.1892,
+      "step": 3690
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 18.871780395507812,
+      "learning_rate": 4.959644874899112e-05,
+      "loss": 0.2265,
+      "step": 3695
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.1702632904052734,
+      "learning_rate": 4.966370729082594e-05,
+      "loss": 0.1712,
+      "step": 3700
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.8938783407211304,
+      "learning_rate": 4.9730965832660756e-05,
+      "loss": 0.1537,
+      "step": 3705
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.3430206775665283,
+      "learning_rate": 4.979822437449556e-05,
+      "loss": 0.2079,
+      "step": 3710
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.3486568927764893,
+      "learning_rate": 4.9865482916330376e-05,
+      "loss": 0.1617,
+      "step": 3715
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.851841151714325,
+      "learning_rate": 4.9932741458165186e-05,
+      "loss": 0.1922,
+      "step": 3720
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.1524040699005127,
+      "learning_rate": 5e-05,
+      "loss": 0.1883,
+      "step": 3725
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.8665148019790649,
+      "learning_rate": 4.999999724399743e-05,
+      "loss": 0.1605,
+      "step": 3730
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.8434327840805054,
+      "learning_rate": 4.9999988975990335e-05,
+      "loss": 0.1506,
+      "step": 3735
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.9524747133255005,
+      "learning_rate": 4.999997519598053e-05,
+      "loss": 0.168,
+      "step": 3740
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.2099045515060425,
+      "learning_rate": 4.999995590397105e-05,
+      "loss": 0.1703,
+      "step": 3745
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.1699497699737549,
+      "learning_rate": 4.999993109996617e-05,
+      "loss": 0.1819,
+      "step": 3750
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.9737915396690369,
+      "learning_rate": 4.999990078397133e-05,
+      "loss": 0.1532,
+      "step": 3755
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.8586001396179199,
+      "learning_rate": 4.9999864955993236e-05,
+      "loss": 0.1859,
+      "step": 3760
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.0415858030319214,
+      "learning_rate": 4.9999823616039776e-05,
+      "loss": 0.151,
+      "step": 3765
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.8435890674591064,
+      "learning_rate": 4.9999776764120066e-05,
+      "loss": 0.1462,
+      "step": 3770
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.612954020500183,
+      "learning_rate": 4.9999724400244434e-05,
+      "loss": 0.1456,
+      "step": 3775
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.8953659534454346,
+      "learning_rate": 4.999966652442444e-05,
+      "loss": 0.1606,
+      "step": 3780
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.1730568408966064,
+      "learning_rate": 4.999960313667282e-05,
+      "loss": 0.1927,
+      "step": 3785
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.012624979019165,
+      "learning_rate": 4.9999534237003574e-05,
+      "loss": 0.1713,
+      "step": 3790
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.361119031906128,
+      "learning_rate": 4.9999459825431875e-05,
+      "loss": 0.1805,
+      "step": 3795
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.299597978591919,
+      "learning_rate": 4.9999379901974144e-05,
+      "loss": 0.1823,
+      "step": 3800
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.0337384939193726,
+      "learning_rate": 4.9999294466647985e-05,
+      "loss": 0.1667,
+      "step": 3805
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.5598517656326294,
+      "learning_rate": 4.999920351947225e-05,
+      "loss": 0.1793,
+      "step": 3810
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.3236010074615479,
+      "learning_rate": 4.9999107060466985e-05,
+      "loss": 0.1897,
+      "step": 3815
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 2.208864212036133,
+      "learning_rate": 4.9999005089653464e-05,
+      "loss": 0.1867,
+      "step": 3820
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.133978247642517,
+      "learning_rate": 4.999889760705415e-05,
+      "loss": 0.1608,
+      "step": 3825
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.532672643661499,
+      "learning_rate": 4.999878461269277e-05,
+      "loss": 0.1734,
+      "step": 3830
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.289676308631897,
+      "learning_rate": 4.999866610659421e-05,
+      "loss": 0.1651,
+      "step": 3835
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.9877331852912903,
+      "learning_rate": 4.999854208878462e-05,
+      "loss": 0.1597,
+      "step": 3840
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.7448598742485046,
+      "learning_rate": 4.999841255929132e-05,
+      "loss": 0.1669,
+      "step": 3845
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.849123477935791,
+      "learning_rate": 4.999827751814289e-05,
+      "loss": 0.1884,
+      "step": 3850
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.065721869468689,
+      "learning_rate": 4.999813696536909e-05,
+      "loss": 0.1729,
+      "step": 3855
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.0890185832977295,
+      "learning_rate": 4.9997990901000924e-05,
+      "loss": 0.1746,
+      "step": 3860
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.9440321922302246,
+      "learning_rate": 4.999783932507058e-05,
+      "loss": 0.1671,
+      "step": 3865
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.8460837006568909,
+      "learning_rate": 4.9997682237611486e-05,
+      "loss": 0.1487,
+      "step": 3870
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.569832682609558,
+      "learning_rate": 4.999751963865828e-05,
+      "loss": 0.1767,
+      "step": 3875
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.4627352952957153,
+      "learning_rate": 4.99973515282468e-05,
+      "loss": 0.1849,
+      "step": 3880
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.9669816493988037,
+      "learning_rate": 4.9997177906414124e-05,
+      "loss": 0.1836,
+      "step": 3885
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.569061040878296,
+      "learning_rate": 4.9996998773198526e-05,
+      "loss": 0.1713,
+      "step": 3890
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.032231092453003,
+      "learning_rate": 4.999681412863949e-05,
+      "loss": 0.1836,
+      "step": 3895
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.8653714060783386,
+      "learning_rate": 4.999662397277776e-05,
+      "loss": 0.1384,
+      "step": 3900
+    },
+    {
+      "epoch": 1.47,
+      "eval_loss": 0.15624000132083893,
+      "eval_runtime": 341.6862,
+      "eval_samples_per_second": 66.702,
+      "eval_steps_per_second": 8.338,
+      "eval_wer": 0.12114932830388458,
+      "step": 3900
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.629183292388916,
+      "learning_rate": 4.999642830565522e-05,
+      "loss": 0.1707,
+      "step": 3905
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.3875324726104736,
+      "learning_rate": 4.999622712731504e-05,
+      "loss": 0.1609,
+      "step": 3910
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.9344199895858765,
+      "learning_rate": 4.999602043780156e-05,
+      "loss": 0.1578,
+      "step": 3915
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.135634422302246,
+      "learning_rate": 4.9995808237160364e-05,
+      "loss": 0.1603,
+      "step": 3920
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.257941484451294,
+      "learning_rate": 4.999559052543823e-05,
+      "loss": 0.2123,
+      "step": 3925
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.1599656343460083,
+      "learning_rate": 4.999536730268315e-05,
+      "loss": 0.1789,
+      "step": 3930
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.8688915967941284,
+      "learning_rate": 4.999513856894437e-05,
+      "loss": 0.1503,
+      "step": 3935
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.443287968635559,
+      "learning_rate": 4.999490432427228e-05,
+      "loss": 0.1582,
+      "step": 3940
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.6412478685379028,
+      "learning_rate": 4.999466456871856e-05,
+      "loss": 0.1534,
+      "step": 3945
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.7799890637397766,
+      "learning_rate": 4.9994419302336066e-05,
+      "loss": 0.154,
+      "step": 3950
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.3910584449768066,
+      "learning_rate": 4.999416852517886e-05,
+      "loss": 0.153,
+      "step": 3955
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.9851388335227966,
+      "learning_rate": 4.9993912237302245e-05,
+      "loss": 0.1726,
+      "step": 3960
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.172812819480896,
+      "learning_rate": 4.999365043876272e-05,
+      "loss": 0.1972,
+      "step": 3965
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.9145379066467285,
+      "learning_rate": 4.999338312961801e-05,
+      "loss": 0.1565,
+      "step": 3970
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.9822158813476562,
+      "learning_rate": 4.9993110309927046e-05,
+      "loss": 0.2251,
+      "step": 3975
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.2184572219848633,
+      "learning_rate": 4.999283197975e-05,
+      "loss": 0.1668,
+      "step": 3980
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.9308606386184692,
+      "learning_rate": 4.999254813914821e-05,
+      "loss": 0.1543,
+      "step": 3985
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.9054309129714966,
+      "learning_rate": 4.9992258788184274e-05,
+      "loss": 0.1867,
+      "step": 3990
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 2.366084098815918,
+      "learning_rate": 4.999196392692199e-05,
+      "loss": 0.1485,
+      "step": 3995
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.9666925668716431,
+      "learning_rate": 4.9991663555426354e-05,
+      "loss": 0.1731,
+      "step": 4000
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.8880798816680908,
+      "learning_rate": 4.99913576737636e-05,
+      "loss": 0.1631,
+      "step": 4005
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 2.6021339893341064,
+      "learning_rate": 4.9991046282001184e-05,
+      "loss": 0.1668,
+      "step": 4010
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.814507007598877,
+      "learning_rate": 4.999072938020775e-05,
+      "loss": 0.1651,
+      "step": 4015
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.8402716517448425,
+      "learning_rate": 4.999040696845315e-05,
+      "loss": 0.1791,
+      "step": 4020
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.9569932818412781,
+      "learning_rate": 4.9990079046808494e-05,
+      "loss": 0.1717,
+      "step": 4025
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.0011552572250366,
+      "learning_rate": 4.9989745615346075e-05,
+      "loss": 0.1435,
+      "step": 4030
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.815840721130371,
+      "learning_rate": 4.9989406674139404e-05,
+      "loss": 0.1684,
+      "step": 4035
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 2.8425769805908203,
+      "learning_rate": 4.9989062223263216e-05,
+      "loss": 0.17,
+      "step": 4040
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.9893917441368103,
+      "learning_rate": 4.998871226279346e-05,
+      "loss": 0.1911,
+      "step": 4045
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 2.195355176925659,
+      "learning_rate": 4.998835679280728e-05,
+      "loss": 0.159,
+      "step": 4050
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.1777340173721313,
+      "learning_rate": 4.998799581338306e-05,
+      "loss": 0.1768,
+      "step": 4055
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 9.148725509643555,
+      "learning_rate": 4.9987629324600395e-05,
+      "loss": 0.2249,
+      "step": 4060
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.6743226647377014,
+      "learning_rate": 4.998725732654008e-05,
+      "loss": 0.2068,
+      "step": 4065
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 3.9217522144317627,
+      "learning_rate": 4.998687981928413e-05,
+      "loss": 0.1991,
+      "step": 4070
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 6.364414215087891,
+      "learning_rate": 4.998649680291578e-05,
+      "loss": 0.1717,
+      "step": 4075
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 6.8271613121032715,
+      "learning_rate": 4.99861082775195e-05,
+      "loss": 0.1741,
+      "step": 4080
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.5911943912506104,
+      "learning_rate": 4.9985714243180906e-05,
+      "loss": 0.1743,
+      "step": 4085
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.1743755340576172,
+      "learning_rate": 4.998531469998692e-05,
+      "loss": 0.1859,
+      "step": 4090
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.0271023511886597,
+      "learning_rate": 4.99849096480256e-05,
+      "loss": 0.1919,
+      "step": 4095
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.2307546138763428,
+      "learning_rate": 4.9984499087386274e-05,
+      "loss": 0.1745,
+      "step": 4100
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 4.071069717407227,
+      "learning_rate": 4.998408301815945e-05,
+      "loss": 0.1551,
+      "step": 4105
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.796509861946106,
+      "learning_rate": 4.998366144043688e-05,
+      "loss": 0.1619,
+      "step": 4110
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 3.109466314315796,
+      "learning_rate": 4.9983234354311496e-05,
+      "loss": 0.1749,
+      "step": 4115
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 3.3903563022613525,
+      "learning_rate": 4.9982801759877466e-05,
+      "loss": 0.2019,
+      "step": 4120
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.8734415769577026,
+      "learning_rate": 4.998236365723017e-05,
+      "loss": 0.1472,
+      "step": 4125
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.9183098077774048,
+      "learning_rate": 4.99819200464662e-05,
+      "loss": 0.1662,
+      "step": 4130
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.9466953873634338,
+      "learning_rate": 4.998147092768337e-05,
+      "loss": 0.1726,
+      "step": 4135
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.9105796813964844,
+      "learning_rate": 4.998101630098069e-05,
+      "loss": 0.1574,
+      "step": 4140
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.1506848335266113,
+      "learning_rate": 4.99805561664584e-05,
+      "loss": 0.1412,
+      "step": 4145
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.2937140464782715,
+      "learning_rate": 4.998009052421796e-05,
+      "loss": 0.1977,
+      "step": 4150
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.0511424541473389,
+      "learning_rate": 4.9979619374362024e-05,
+      "loss": 0.1834,
+      "step": 4155
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 2.416187047958374,
+      "learning_rate": 4.9979142716994476e-05,
+      "loss": 0.1742,
+      "step": 4160
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.9834833145141602,
+      "learning_rate": 4.997866055222041e-05,
+      "loss": 0.1607,
+      "step": 4165
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.7763827443122864,
+      "learning_rate": 4.9978172880146134e-05,
+      "loss": 0.1607,
+      "step": 4170
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.1321146488189697,
+      "learning_rate": 4.9977679700879164e-05,
+      "loss": 0.1618,
+      "step": 4175
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 7.22818660736084,
+      "learning_rate": 4.997718101452824e-05,
+      "loss": 0.1549,
+      "step": 4180
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.299710988998413,
+      "learning_rate": 4.9976676821203314e-05,
+      "loss": 0.1776,
+      "step": 4185
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 6.282206058502197,
+      "learning_rate": 4.997616712101556e-05,
+      "loss": 0.1852,
+      "step": 4190
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.2167057991027832,
+      "learning_rate": 4.997565191407733e-05,
+      "loss": 0.1651,
+      "step": 4195
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.018463134765625,
+      "learning_rate": 4.9975131200502246e-05,
+      "loss": 0.1701,
+      "step": 4200
+    },
+    {
+      "epoch": 1.58,
+      "eval_loss": 0.15161743760108948,
+      "eval_runtime": 343.6851,
+      "eval_samples_per_second": 66.314,
+      "eval_steps_per_second": 8.29,
+      "eval_wer": 0.1189799649262127,
+      "step": 4200
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 2.062075138092041,
+      "learning_rate": 4.9974604980405093e-05,
+      "loss": 0.1329,
+      "step": 4205
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.125655174255371,
+      "learning_rate": 4.9974073253901906e-05,
+      "loss": 0.1528,
+      "step": 4210
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.1863205432891846,
+      "learning_rate": 4.9973536021109915e-05,
+      "loss": 0.1542,
+      "step": 4215
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.8338540196418762,
+      "learning_rate": 4.9972993282147574e-05,
+      "loss": 0.1764,
+      "step": 4220
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.3029398918151855,
+      "learning_rate": 4.997244503713454e-05,
+      "loss": 0.1416,
+      "step": 4225
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.7819846868515015,
+      "learning_rate": 4.9971891286191695e-05,
+      "loss": 0.1528,
+      "step": 4230
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.1202632188796997,
+      "learning_rate": 4.997133202944112e-05,
+      "loss": 0.1286,
+      "step": 4235
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.8742148280143738,
+      "learning_rate": 4.997076726700613e-05,
+      "loss": 0.1597,
+      "step": 4240
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.954116702079773,
+      "learning_rate": 4.997019699901125e-05,
+      "loss": 0.1709,
+      "step": 4245
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.4541547298431396,
+      "learning_rate": 4.9969621225582195e-05,
+      "loss": 0.1618,
+      "step": 4250
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.4215017557144165,
+      "learning_rate": 4.996903994684592e-05,
+      "loss": 0.159,
+      "step": 4255
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 5.174014568328857,
+      "learning_rate": 4.9968453162930595e-05,
+      "loss": 0.1869,
+      "step": 4260
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.9055848717689514,
+      "learning_rate": 4.996786087396558e-05,
+      "loss": 0.1627,
+      "step": 4265
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.7991906404495239,
+      "learning_rate": 4.9967263080081474e-05,
+      "loss": 0.2156,
+      "step": 4270
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.5315312147140503,
+      "learning_rate": 4.9966659781410064e-05,
+      "loss": 0.1634,
+      "step": 4275
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.026758074760437,
+      "learning_rate": 4.9966050978084386e-05,
+      "loss": 0.1596,
+      "step": 4280
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.9494081139564514,
+      "learning_rate": 4.996543667023865e-05,
+      "loss": 0.1526,
+      "step": 4285
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.8328499794006348,
+      "learning_rate": 4.996481685800831e-05,
+      "loss": 0.1664,
+      "step": 4290
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.7483776807785034,
+      "learning_rate": 4.996419154153002e-05,
+      "loss": 0.1829,
+      "step": 4295
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.9861835837364197,
+      "learning_rate": 4.996356072094164e-05,
+      "loss": 0.1464,
+      "step": 4300
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.7990335822105408,
+      "learning_rate": 4.996292439638227e-05,
+      "loss": 0.1626,
+      "step": 4305
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.0307897329330444,
+      "learning_rate": 4.99622825679922e-05,
+      "loss": 0.1652,
+      "step": 4310
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.0875548124313354,
+      "learning_rate": 4.9961635235912935e-05,
+      "loss": 0.1671,
+      "step": 4315
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.8530448079109192,
+      "learning_rate": 4.996098240028721e-05,
+      "loss": 0.1717,
+      "step": 4320
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.7367186546325684,
+      "learning_rate": 4.996032406125894e-05,
+      "loss": 0.1508,
+      "step": 4325
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.7808917760849,
+      "learning_rate": 4.995966021897332e-05,
+      "loss": 0.1726,
+      "step": 4330
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.8300977945327759,
+      "learning_rate": 4.995899087357666e-05,
+      "loss": 0.1733,
+      "step": 4335
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.8856766819953918,
+      "learning_rate": 4.995831602521658e-05,
+      "loss": 0.1976,
+      "step": 4340
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.4146891832351685,
+      "learning_rate": 4.995763567404184e-05,
+      "loss": 0.1439,
+      "step": 4345
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.7983535528182983,
+      "learning_rate": 4.9956949820202464e-05,
+      "loss": 0.1656,
+      "step": 4350
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.9048216938972473,
+      "learning_rate": 4.995625846384966e-05,
+      "loss": 0.1524,
+      "step": 4355
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.9650282263755798,
+      "learning_rate": 4.995556160513587e-05,
+      "loss": 0.1926,
+      "step": 4360
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.0725921392440796,
+      "learning_rate": 4.995485924421472e-05,
+      "loss": 0.1612,
+      "step": 4365
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.2737367153167725,
+      "learning_rate": 4.995415138124108e-05,
+      "loss": 0.1524,
+      "step": 4370
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.0467036962509155,
+      "learning_rate": 4.995343801637102e-05,
+      "loss": 0.1688,
+      "step": 4375
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.7611225843429565,
+      "learning_rate": 4.9952719149761814e-05,
+      "loss": 0.1698,
+      "step": 4380
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.9313662052154541,
+      "learning_rate": 4.995199478157196e-05,
+      "loss": 0.1417,
+      "step": 4385
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.9136959314346313,
+      "learning_rate": 4.995126491196117e-05,
+      "loss": 0.1529,
+      "step": 4390
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.636729896068573,
+      "learning_rate": 4.995052954109036e-05,
+      "loss": 0.1669,
+      "step": 4395
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.8130836486816406,
+      "learning_rate": 4.994978866912168e-05,
+      "loss": 0.1496,
+      "step": 4400
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.975873589515686,
+      "learning_rate": 4.994904229621846e-05,
+      "loss": 0.1533,
+      "step": 4405
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.413733720779419,
+      "learning_rate": 4.994829042254527e-05,
+      "loss": 0.1443,
+      "step": 4410
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 3.059051990509033,
+      "learning_rate": 4.994753304826788e-05,
+      "loss": 0.153,
+      "step": 4415
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.5588415861129761,
+      "learning_rate": 4.9946770173553275e-05,
+      "loss": 0.1408,
+      "step": 4420
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.641609787940979,
+      "learning_rate": 4.994600179856966e-05,
+      "loss": 0.1681,
+      "step": 4425
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.7481341361999512,
+      "learning_rate": 4.994522792348644e-05,
+      "loss": 0.163,
+      "step": 4430
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.13042414188385,
+      "learning_rate": 4.9944448548474244e-05,
+      "loss": 0.1497,
+      "step": 4435
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.7852217555046082,
+      "learning_rate": 4.99436636737049e-05,
+      "loss": 0.1445,
+      "step": 4440
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.2825710773468018,
+      "learning_rate": 4.9942873299351464e-05,
+      "loss": 0.152,
+      "step": 4445
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.3754384517669678,
+      "learning_rate": 4.9942077425588195e-05,
+      "loss": 0.1406,
+      "step": 4450
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.4862149953842163,
+      "learning_rate": 4.994127605259058e-05,
+      "loss": 0.1479,
+      "step": 4455
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 3.0437076091766357,
+      "learning_rate": 4.994046918053529e-05,
+      "loss": 0.1644,
+      "step": 4460
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.5922364592552185,
+      "learning_rate": 4.993965680960023e-05,
+      "loss": 0.1597,
+      "step": 4465
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.208236813545227,
+      "learning_rate": 4.993883893996451e-05,
+      "loss": 0.1594,
+      "step": 4470
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.2552003860473633,
+      "learning_rate": 4.993801557180845e-05,
+      "loss": 0.188,
+      "step": 4475
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.9966633319854736,
+      "learning_rate": 4.9937186705313604e-05,
+      "loss": 0.1629,
+      "step": 4480
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.7919297814369202,
+      "learning_rate": 4.9936352340662706e-05,
+      "loss": 0.1939,
+      "step": 4485
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.5858938694000244,
+      "learning_rate": 4.9935512478039716e-05,
+      "loss": 0.1614,
+      "step": 4490
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.9647216796875,
+      "learning_rate": 4.993466711762981e-05,
+      "loss": 0.152,
+      "step": 4495
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.536423683166504,
+      "learning_rate": 4.9933816259619376e-05,
+      "loss": 0.1756,
+      "step": 4500
+    },
+    {
+      "epoch": 1.69,
+      "eval_loss": 0.14690713584423065,
+      "eval_runtime": 333.1601,
+      "eval_samples_per_second": 68.409,
+      "eval_steps_per_second": 8.551,
+      "eval_wer": 0.12032418436900272,
+      "step": 4500
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.6433957815170288,
+      "learning_rate": 4.993295990419601e-05,
+      "loss": 0.1649,
+      "step": 4505
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.7061463594436646,
+      "learning_rate": 4.993209805154852e-05,
+      "loss": 0.1853,
+      "step": 4510
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 2.2344307899475098,
+      "learning_rate": 4.9931230701866926e-05,
+      "loss": 0.1686,
+      "step": 4515
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.1541404724121094,
+      "learning_rate": 4.993035785534247e-05,
+      "loss": 0.1889,
+      "step": 4520
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.9450957179069519,
+      "learning_rate": 4.9929479512167585e-05,
+      "loss": 0.1551,
+      "step": 4525
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.8680835962295532,
+      "learning_rate": 4.992859567253593e-05,
+      "loss": 0.1479,
+      "step": 4530
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.6638008952140808,
+      "learning_rate": 4.9927706336642385e-05,
+      "loss": 0.2194,
+      "step": 4535
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.8829554319381714,
+      "learning_rate": 4.992681150468303e-05,
+      "loss": 0.1583,
+      "step": 4540
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.5697282552719116,
+      "learning_rate": 4.992591117685513e-05,
+      "loss": 0.1719,
+      "step": 4545
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.2131922245025635,
+      "learning_rate": 4.992500535335724e-05,
+      "loss": 0.1652,
+      "step": 4550
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.1503419876098633,
+      "learning_rate": 4.9924094034389025e-05,
+      "loss": 0.1614,
+      "step": 4555
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.4217567443847656,
+      "learning_rate": 4.9923177220151446e-05,
+      "loss": 0.1408,
+      "step": 4560
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.0131118297576904,
+      "learning_rate": 4.9922254910846636e-05,
+      "loss": 0.157,
+      "step": 4565
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.7946292161941528,
+      "learning_rate": 4.992132710667793e-05,
+      "loss": 0.1659,
+      "step": 4570
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.9991573095321655,
+      "learning_rate": 4.9920393807849906e-05,
+      "loss": 0.1579,
+      "step": 4575
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 2.415388345718384,
+      "learning_rate": 4.991945501456834e-05,
+      "loss": 0.1561,
+      "step": 4580
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 5.059329986572266,
+      "learning_rate": 4.991851072704021e-05,
+      "loss": 0.182,
+      "step": 4585
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.5428355932235718,
+      "learning_rate": 4.991756094547371e-05,
+      "loss": 0.2072,
+      "step": 4590
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.9356845617294312,
+      "learning_rate": 4.9916605670078254e-05,
+      "loss": 0.1402,
+      "step": 4595
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.056833028793335,
+      "learning_rate": 4.991564490106446e-05,
+      "loss": 0.2042,
+      "step": 4600
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.5836195945739746,
+      "learning_rate": 4.991467863864416e-05,
+      "loss": 0.1422,
+      "step": 4605
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 3.130746364593506,
+      "learning_rate": 4.991370688303039e-05,
+      "loss": 0.1637,
+      "step": 4610
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.9702777862548828,
+      "learning_rate": 4.991272963443741e-05,
+      "loss": 0.1618,
+      "step": 4615
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.6662006378173828,
+      "learning_rate": 4.991174689308068e-05,
+      "loss": 0.1894,
+      "step": 4620
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.6733968257904053,
+      "learning_rate": 4.991075865917687e-05,
+      "loss": 0.1443,
+      "step": 4625
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.8633238673210144,
+      "learning_rate": 4.990976493294388e-05,
+      "loss": 0.1623,
+      "step": 4630
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.9704465270042419,
+      "learning_rate": 4.990876571460079e-05,
+      "loss": 0.2001,
+      "step": 4635
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 6.2354350090026855,
+      "learning_rate": 4.990776100436793e-05,
+      "loss": 0.1741,
+      "step": 4640
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.0551279783248901,
+      "learning_rate": 4.9906750802466794e-05,
+      "loss": 0.1688,
+      "step": 4645
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.851539134979248,
+      "learning_rate": 4.9905735109120125e-05,
+      "loss": 0.1507,
+      "step": 4650
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.5565102100372314,
+      "learning_rate": 4.990471392455186e-05,
+      "loss": 0.1297,
+      "step": 4655
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.162804126739502,
+      "learning_rate": 4.990368724898715e-05,
+      "loss": 0.2039,
+      "step": 4660
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.0225718021392822,
+      "learning_rate": 4.9902655082652366e-05,
+      "loss": 0.1739,
+      "step": 4665
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.9165812730789185,
+      "learning_rate": 4.9901617425775067e-05,
+      "loss": 0.1671,
+      "step": 4670
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 3.626378297805786,
+      "learning_rate": 4.990057427858404e-05,
+      "loss": 0.1626,
+      "step": 4675
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.1790755987167358,
+      "learning_rate": 4.989952564130928e-05,
+      "loss": 0.1803,
+      "step": 4680
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.0086643695831299,
+      "learning_rate": 4.989847151418199e-05,
+      "loss": 0.1855,
+      "step": 4685
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.5969728827476501,
+      "learning_rate": 4.9897411897434583e-05,
+      "loss": 0.1346,
+      "step": 4690
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.90949547290802,
+      "learning_rate": 4.9896346791300684e-05,
+      "loss": 0.1884,
+      "step": 4695
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 42.233055114746094,
+      "learning_rate": 4.989527619601513e-05,
+      "loss": 0.1853,
+      "step": 4700
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.9414621591567993,
+      "learning_rate": 4.989420011181396e-05,
+      "loss": 0.1742,
+      "step": 4705
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.1032065153121948,
+      "learning_rate": 4.989311853893444e-05,
+      "loss": 0.1429,
+      "step": 4710
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.2506531476974487,
+      "learning_rate": 4.9892031477615025e-05,
+      "loss": 0.1837,
+      "step": 4715
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.8768670558929443,
+      "learning_rate": 4.98909389280954e-05,
+      "loss": 0.1479,
+      "step": 4720
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.7473987340927124,
+      "learning_rate": 4.988984089061643e-05,
+      "loss": 0.1653,
+      "step": 4725
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.7386647462844849,
+      "learning_rate": 4.988873736542025e-05,
+      "loss": 0.1854,
+      "step": 4730
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.7257795929908752,
+      "learning_rate": 4.988762835275011e-05,
+      "loss": 0.1455,
+      "step": 4735
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.3313937187194824,
+      "learning_rate": 4.988651385285058e-05,
+      "loss": 0.1665,
+      "step": 4740
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.7676001787185669,
+      "learning_rate": 4.9885393865967355e-05,
+      "loss": 0.1693,
+      "step": 4745
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.5955161452293396,
+      "learning_rate": 4.9884268392347374e-05,
+      "loss": 0.1677,
+      "step": 4750
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.6203986406326294,
+      "learning_rate": 4.988313743223879e-05,
+      "loss": 0.156,
+      "step": 4755
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.4381760358810425,
+      "learning_rate": 4.9882000985890956e-05,
+      "loss": 0.1707,
+      "step": 4760
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.7708421945571899,
+      "learning_rate": 4.988085905355442e-05,
+      "loss": 0.165,
+      "step": 4765
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.710453450679779,
+      "learning_rate": 4.987971163548097e-05,
+      "loss": 0.1532,
+      "step": 4770
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.0742347240447998,
+      "learning_rate": 4.987855873192359e-05,
+      "loss": 0.1937,
+      "step": 4775
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.0155588388442993,
+      "learning_rate": 4.987740034313646e-05,
+      "loss": 0.1554,
+      "step": 4780
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.0998890399932861,
+      "learning_rate": 4.9876236469375e-05,
+      "loss": 0.1782,
+      "step": 4785
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.8582411408424377,
+      "learning_rate": 4.98750671108958e-05,
+      "loss": 0.1598,
+      "step": 4790
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.8275126218795776,
+      "learning_rate": 4.98738922679567e-05,
+      "loss": 0.1572,
+      "step": 4795
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.7812132239341736,
+      "learning_rate": 4.987271194081672e-05,
+      "loss": 0.167,
+      "step": 4800
+    },
+    {
+      "epoch": 1.81,
+      "eval_loss": 0.1384541243314743,
+      "eval_runtime": 328.4965,
+      "eval_samples_per_second": 69.38,
+      "eval_steps_per_second": 8.673,
+      "eval_wer": 0.11514997352921713,
+      "step": 4800
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.00557279586792,
+      "learning_rate": 4.98715261297361e-05,
+      "loss": 0.1534,
+      "step": 4805
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.829555094242096,
+      "learning_rate": 4.987033483497629e-05,
+      "loss": 0.1494,
+      "step": 4810
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.0176321268081665,
+      "learning_rate": 4.986913805679994e-05,
+      "loss": 0.1307,
+      "step": 4815
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.6296998858451843,
+      "learning_rate": 4.986793579547092e-05,
+      "loss": 0.1706,
+      "step": 4820
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.0653918981552124,
+      "learning_rate": 4.986672805125431e-05,
+      "loss": 0.1611,
+      "step": 4825
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.0894889831542969,
+      "learning_rate": 4.9865514824416385e-05,
+      "loss": 0.1825,
+      "step": 4830
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.5467034578323364,
+      "learning_rate": 4.986429611522465e-05,
+      "loss": 0.1606,
+      "step": 4835
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.9150000810623169,
+      "learning_rate": 4.986307192394779e-05,
+      "loss": 0.161,
+      "step": 4840
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.9007552266120911,
+      "learning_rate": 4.9861842250855725e-05,
+      "loss": 0.1382,
+      "step": 4845
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.9415110349655151,
+      "learning_rate": 4.986060709621957e-05,
+      "loss": 0.1631,
+      "step": 4850
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.3723894357681274,
+      "learning_rate": 4.985936646031165e-05,
+      "loss": 0.1654,
+      "step": 4855
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.9437314867973328,
+      "learning_rate": 4.985812034340551e-05,
+      "loss": 0.1653,
+      "step": 4860
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.9050099849700928,
+      "learning_rate": 4.985686874577589e-05,
+      "loss": 0.1579,
+      "step": 4865
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.8051374554634094,
+      "learning_rate": 4.985561166769874e-05,
+      "loss": 0.1559,
+      "step": 4870
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 4.0360918045043945,
+      "learning_rate": 4.985434910945122e-05,
+      "loss": 0.172,
+      "step": 4875
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.7933779358863831,
+      "learning_rate": 4.9853081071311704e-05,
+      "loss": 0.1602,
+      "step": 4880
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.7482435703277588,
+      "learning_rate": 4.9851807553559757e-05,
+      "loss": 0.1452,
+      "step": 4885
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.6117326617240906,
+      "learning_rate": 4.985052855647618e-05,
+      "loss": 0.1418,
+      "step": 4890
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.7311630845069885,
+      "learning_rate": 4.984924408034296e-05,
+      "loss": 0.1603,
+      "step": 4895
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.6815735697746277,
+      "learning_rate": 4.9847954125443305e-05,
+      "loss": 0.1738,
+      "step": 4900
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.6389229893684387,
+      "learning_rate": 4.984665869206161e-05,
+      "loss": 0.1499,
+      "step": 4905
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.9959733486175537,
+      "learning_rate": 4.9845357780483504e-05,
+      "loss": 0.145,
+      "step": 4910
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.8103878498077393,
+      "learning_rate": 4.984405139099581e-05,
+      "loss": 0.1574,
+      "step": 4915
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.641474187374115,
+      "learning_rate": 4.984273952388655e-05,
+      "loss": 0.1424,
+      "step": 4920
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 2.4348208904266357,
+      "learning_rate": 4.984142217944499e-05,
+      "loss": 0.1457,
+      "step": 4925
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.5225803852081299,
+      "learning_rate": 4.984009935796155e-05,
+      "loss": 0.1293,
+      "step": 4930
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.1669846773147583,
+      "learning_rate": 4.983877105972791e-05,
+      "loss": 0.1522,
+      "step": 4935
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.239125370979309,
+      "learning_rate": 4.9837437285036906e-05,
+      "loss": 0.1708,
+      "step": 4940
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.560187816619873,
+      "learning_rate": 4.983609803418264e-05,
+      "loss": 0.159,
+      "step": 4945
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.7422255873680115,
+      "learning_rate": 4.983475330746037e-05,
+      "loss": 0.1396,
+      "step": 4950
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.9650576710700989,
+      "learning_rate": 4.9833403105166585e-05,
+      "loss": 0.1591,
+      "step": 4955
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.2351810932159424,
+      "learning_rate": 4.9832047427598985e-05,
+      "loss": 0.1593,
+      "step": 4960
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.9720680117607117,
+      "learning_rate": 4.983068627505647e-05,
+      "loss": 0.162,
+      "step": 4965
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.202864170074463,
+      "learning_rate": 4.9829319647839136e-05,
+      "loss": 0.1425,
+      "step": 4970
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.73358154296875,
+      "learning_rate": 4.98279475462483e-05,
+      "loss": 0.1306,
+      "step": 4975
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.9871314167976379,
+      "learning_rate": 4.982656997058649e-05,
+      "loss": 0.1576,
+      "step": 4980
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.7367146611213684,
+      "learning_rate": 4.982518692115744e-05,
+      "loss": 0.1484,
+      "step": 4985
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.0497846603393555,
+      "learning_rate": 4.9823798398266066e-05,
+      "loss": 0.176,
+      "step": 4990
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.112668514251709,
+      "learning_rate": 4.982240440221852e-05,
+      "loss": 0.1444,
+      "step": 4995
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.1213332414627075,
+      "learning_rate": 4.982100493332215e-05,
+      "loss": 0.1676,
+      "step": 5000
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.7841285467147827,
+      "learning_rate": 4.981959999188551e-05,
+      "loss": 0.1451,
+      "step": 5005
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.6666841506958008,
+      "learning_rate": 4.9818189578218375e-05,
+      "loss": 0.1494,
+      "step": 5010
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.975662887096405,
+      "learning_rate": 4.981677369263169e-05,
+      "loss": 0.1495,
+      "step": 5015
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.8885636925697327,
+      "learning_rate": 4.981535233543765e-05,
+      "loss": 0.1461,
+      "step": 5020
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 5.57780647277832,
+      "learning_rate": 4.981392550694962e-05,
+      "loss": 0.1574,
+      "step": 5025
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.040246605873108,
+      "learning_rate": 4.98124932074822e-05,
+      "loss": 0.1476,
+      "step": 5030
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.5935861468315125,
+      "learning_rate": 4.981105543735117e-05,
+      "loss": 0.1481,
+      "step": 5035
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.7847604751586914,
+      "learning_rate": 4.9809612196873555e-05,
+      "loss": 0.1513,
+      "step": 5040
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.0112693309783936,
+      "learning_rate": 4.980816348636753e-05,
+      "loss": 0.1365,
+      "step": 5045
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.8237004280090332,
+      "learning_rate": 4.9806709306152534e-05,
+      "loss": 0.1857,
+      "step": 5050
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.5945549011230469,
+      "learning_rate": 4.980524965654916e-05,
+      "loss": 0.1505,
+      "step": 5055
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.803919792175293,
+      "learning_rate": 4.980378453787925e-05,
+      "loss": 0.1709,
+      "step": 5060
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.7791391611099243,
+      "learning_rate": 4.9802313950465826e-05,
+      "loss": 0.1325,
+      "step": 5065
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.0427775382995605,
+      "learning_rate": 4.980083789463312e-05,
+      "loss": 0.1415,
+      "step": 5070
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.932044267654419,
+      "learning_rate": 4.9799356370706594e-05,
+      "loss": 0.1643,
+      "step": 5075
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.4509750604629517,
+      "learning_rate": 4.979786937901287e-05,
+      "loss": 0.1505,
+      "step": 5080
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.3959046602249146,
+      "learning_rate": 4.9796376919879804e-05,
+      "loss": 0.1642,
+      "step": 5085
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.4368891716003418,
+      "learning_rate": 4.979487899363647e-05,
+      "loss": 0.1575,
+      "step": 5090
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.7688291668891907,
+      "learning_rate": 4.9793375600613114e-05,
+      "loss": 0.1457,
+      "step": 5095
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.9328327178955078,
+      "learning_rate": 4.979186674114121e-05,
+      "loss": 0.1516,
+      "step": 5100
+    },
+    {
+      "epoch": 1.92,
+      "eval_loss": 0.14146040380001068,
+      "eval_runtime": 328.7804,
+      "eval_samples_per_second": 69.32,
+      "eval_steps_per_second": 8.665,
+      "eval_wer": 0.11254425584011647,
+      "step": 5100
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.311266303062439,
+      "learning_rate": 4.9790352415553434e-05,
+      "loss": 0.1453,
+      "step": 5105
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.6027989387512207,
+      "learning_rate": 4.978883262418367e-05,
+      "loss": 0.1455,
+      "step": 5110
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.7381126284599304,
+      "learning_rate": 4.978730736736698e-05,
+      "loss": 0.1793,
+      "step": 5115
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.5193963050842285,
+      "learning_rate": 4.9785776645439676e-05,
+      "loss": 0.1869,
+      "step": 5120
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.9653743505477905,
+      "learning_rate": 4.978424045873925e-05,
+      "loss": 0.1463,
+      "step": 5125
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.5014756917953491,
+      "learning_rate": 4.9782698807604385e-05,
+      "loss": 0.1767,
+      "step": 5130
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.3769853115081787,
+      "learning_rate": 4.978115169237499e-05,
+      "loss": 0.161,
+      "step": 5135
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.9955167174339294,
+      "learning_rate": 4.977959911339219e-05,
+      "loss": 0.1594,
+      "step": 5140
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.7471234202384949,
+      "learning_rate": 4.9778041070998284e-05,
+      "loss": 0.1568,
+      "step": 5145
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.9103015065193176,
+      "learning_rate": 4.9776477565536784e-05,
+      "loss": 0.1338,
+      "step": 5150
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.3470547199249268,
+      "learning_rate": 4.977490859735242e-05,
+      "loss": 0.1653,
+      "step": 5155
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.6749018430709839,
+      "learning_rate": 4.977333416679112e-05,
+      "loss": 0.151,
+      "step": 5160
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.9364789724349976,
+      "learning_rate": 4.977175427420001e-05,
+      "loss": 0.1593,
+      "step": 5165
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.5562646985054016,
+      "learning_rate": 4.977016891992743e-05,
+      "loss": 0.149,
+      "step": 5170
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.9740204215049744,
+      "learning_rate": 4.976857810432291e-05,
+      "loss": 0.1511,
+      "step": 5175
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.0942884683609009,
+      "learning_rate": 4.9766981827737206e-05,
+      "loss": 0.1692,
+      "step": 5180
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.3451478481292725,
+      "learning_rate": 4.976538009052226e-05,
+      "loss": 0.1572,
+      "step": 5185
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.1559349298477173,
+      "learning_rate": 4.976377289303121e-05,
+      "loss": 0.1395,
+      "step": 5190
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.7259495258331299,
+      "learning_rate": 4.9762160235618437e-05,
+      "loss": 0.1587,
+      "step": 5195
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.4958502054214478,
+      "learning_rate": 4.976054211863949e-05,
+      "loss": 0.1561,
+      "step": 5200
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.3462657928466797,
+      "learning_rate": 4.975891854245112e-05,
+      "loss": 0.1565,
+      "step": 5205
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.0927257537841797,
+      "learning_rate": 4.9757289507411294e-05,
+      "loss": 0.1495,
+      "step": 5210
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.9764543771743774,
+      "learning_rate": 4.97556550138792e-05,
+      "loss": 0.171,
+      "step": 5215
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.0617835521697998,
+      "learning_rate": 4.9754015062215206e-05,
+      "loss": 0.1568,
+      "step": 5220
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.8799893856048584,
+      "learning_rate": 4.975236965278087e-05,
+      "loss": 0.1452,
+      "step": 5225
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.3555277585983276,
+      "learning_rate": 4.9750718785938996e-05,
+      "loss": 0.1654,
+      "step": 5230
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 5.218792915344238,
+      "learning_rate": 4.974906246205356e-05,
+      "loss": 0.155,
+      "step": 5235
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.9155853390693665,
+      "learning_rate": 4.974740068148974e-05,
+      "loss": 0.15,
+      "step": 5240
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.7979586124420166,
+      "learning_rate": 4.974573344461394e-05,
+      "loss": 0.1698,
+      "step": 5245
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.0437504053115845,
+      "learning_rate": 4.974406075179374e-05,
+      "loss": 0.1585,
+      "step": 5250
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.8467562198638916,
+      "learning_rate": 4.9742382603397944e-05,
+      "loss": 0.1581,
+      "step": 5255
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.5232844948768616,
+      "learning_rate": 4.974069899979654e-05,
+      "loss": 0.1617,
+      "step": 5260
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.8070797920227051,
+      "learning_rate": 4.9739009941360736e-05,
+      "loss": 0.1526,
+      "step": 5265
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.5830070972442627,
+      "learning_rate": 4.9737315428462946e-05,
+      "loss": 0.1503,
+      "step": 5270
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.7734190821647644,
+      "learning_rate": 4.9735615461476756e-05,
+      "loss": 0.1697,
+      "step": 5275
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.8063217401504517,
+      "learning_rate": 4.9733910040777e-05,
+      "loss": 0.1524,
+      "step": 5280
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.6108196973800659,
+      "learning_rate": 4.9732199166739664e-05,
+      "loss": 0.1672,
+      "step": 5285
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.6462995409965515,
+      "learning_rate": 4.973048283974198e-05,
+      "loss": 0.1535,
+      "step": 5290
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.7417263984680176,
+      "learning_rate": 4.972876106016234e-05,
+      "loss": 0.1748,
+      "step": 5295
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.7168163061141968,
+      "learning_rate": 4.97270338283804e-05,
+      "loss": 0.1642,
+      "step": 5300
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.6040955781936646,
+      "learning_rate": 4.972530114477695e-05,
+      "loss": 0.1262,
+      "step": 5305
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.082634449005127,
+      "learning_rate": 4.972356300973403e-05,
+      "loss": 0.1506,
+      "step": 5310
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.6132025718688965,
+      "learning_rate": 4.972181942363485e-05,
+      "loss": 0.1416,
+      "step": 5315
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.6767434477806091,
+      "learning_rate": 4.972007038686385e-05,
+      "loss": 0.1285,
+      "step": 5320
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 1.3485394716262817,
+      "learning_rate": 4.9718315899806646e-05,
+      "loss": 0.1352,
+      "step": 5325
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 1.297582983970642,
+      "learning_rate": 4.9716555962850074e-05,
+      "loss": 0.1496,
+      "step": 5330
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.7809275388717651,
+      "learning_rate": 4.971479057638216e-05,
+      "loss": 0.141,
+      "step": 5335
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.7213215827941895,
+      "learning_rate": 4.971301974079215e-05,
+      "loss": 0.14,
+      "step": 5340
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.7752698659896851,
+      "learning_rate": 4.9711243456470464e-05,
+      "loss": 0.1373,
+      "step": 5345
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 1.2434266805648804,
+      "learning_rate": 4.9709461723808745e-05,
+      "loss": 0.1643,
+      "step": 5350
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.8750342726707458,
+      "learning_rate": 4.9707674543199824e-05,
+      "loss": 0.1448,
+      "step": 5355
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.6641231775283813,
+      "learning_rate": 4.970588191503774e-05,
+      "loss": 0.1337,
+      "step": 5360
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 1.229873538017273,
+      "learning_rate": 4.970408383971774e-05,
+      "loss": 0.1568,
+      "step": 5365
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.7113270163536072,
+      "learning_rate": 4.9702280317636256e-05,
+      "loss": 0.1335,
+      "step": 5370
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.5180246233940125,
+      "learning_rate": 4.9700471349190935e-05,
+      "loss": 0.14,
+      "step": 5375
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.895698070526123,
+      "learning_rate": 4.9698656934780606e-05,
+      "loss": 0.1254,
+      "step": 5380
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.6878983378410339,
+      "learning_rate": 4.9696837074805325e-05,
+      "loss": 0.1195,
+      "step": 5385
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.8192896842956543,
+      "learning_rate": 4.9695011769666335e-05,
+      "loss": 0.1253,
+      "step": 5390
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.8286742568016052,
+      "learning_rate": 4.969318101976607e-05,
+      "loss": 0.1271,
+      "step": 5395
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 1.4134165048599243,
+      "learning_rate": 4.969134482550818e-05,
+      "loss": 0.1309,
+      "step": 5400
+    },
+    {
+      "epoch": 2.03,
+      "eval_loss": 0.14486746490001678,
+      "eval_runtime": 328.5631,
+      "eval_samples_per_second": 69.366,
+      "eval_steps_per_second": 8.671,
+      "eval_wer": 0.11267040566474754,
+      "step": 5400
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 1.4804235696792603,
+      "learning_rate": 4.968950318729751e-05,
+      "loss": 0.1697,
+      "step": 5405
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.9230585694313049,
+      "learning_rate": 4.9687656105540104e-05,
+      "loss": 0.1472,
+      "step": 5410
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.7488617897033691,
+      "learning_rate": 4.9685803580643206e-05,
+      "loss": 0.1376,
+      "step": 5415
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.6868006587028503,
+      "learning_rate": 4.9683945613015264e-05,
+      "loss": 0.1228,
+      "step": 5420
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 1.6682804822921753,
+      "learning_rate": 4.9682082203065924e-05,
+      "loss": 0.1257,
+      "step": 5425
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.777127742767334,
+      "learning_rate": 4.968021335120602e-05,
+      "loss": 0.1402,
+      "step": 5430
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.8282036185264587,
+      "learning_rate": 4.967833905784761e-05,
+      "loss": 0.147,
+      "step": 5435
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.1067842245101929,
+      "learning_rate": 4.967645932340392e-05,
+      "loss": 0.1423,
+      "step": 5440
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.9508112668991089,
+      "learning_rate": 4.9674574148289424e-05,
+      "loss": 0.1196,
+      "step": 5445
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.7283747792243958,
+      "learning_rate": 4.967268353291974e-05,
+      "loss": 0.1608,
+      "step": 5450
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.9536224007606506,
+      "learning_rate": 4.9670787477711725e-05,
+      "loss": 0.1495,
+      "step": 5455
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.9457405805587769,
+      "learning_rate": 4.9668885983083414e-05,
+      "loss": 0.1392,
+      "step": 5460
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.9726963043212891,
+      "learning_rate": 4.966697904945405e-05,
+      "loss": 0.1641,
+      "step": 5465
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.6165057420730591,
+      "learning_rate": 4.966506667724409e-05,
+      "loss": 0.1634,
+      "step": 5470
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 1.10140860080719,
+      "learning_rate": 4.966314886687514e-05,
+      "loss": 0.1452,
+      "step": 5475
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 1.1103501319885254,
+      "learning_rate": 4.966122561877008e-05,
+      "loss": 0.1258,
+      "step": 5480
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.7427185773849487,
+      "learning_rate": 4.965929693335292e-05,
+      "loss": 0.139,
+      "step": 5485
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.940016508102417,
+      "learning_rate": 4.965736281104889e-05,
+      "loss": 0.1377,
+      "step": 5490
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.6074488162994385,
+      "learning_rate": 4.965542325228446e-05,
+      "loss": 0.1454,
+      "step": 5495
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.6096179485321045,
+      "learning_rate": 4.9653478257487236e-05,
+      "loss": 0.1335,
+      "step": 5500
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.677111029624939,
+      "learning_rate": 4.9651527827086064e-05,
+      "loss": 0.1356,
+      "step": 5505
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 1.0961294174194336,
+      "learning_rate": 4.964957196151097e-05,
+      "loss": 0.1652,
+      "step": 5510
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.993168294429779,
+      "learning_rate": 4.9647610661193176e-05,
+      "loss": 0.1308,
+      "step": 5515
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.5909833908081055,
+      "learning_rate": 4.964564392656512e-05,
+      "loss": 0.1306,
+      "step": 5520
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 1.548743724822998,
+      "learning_rate": 4.964367175806043e-05,
+      "loss": 0.1339,
+      "step": 5525
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.7050294876098633,
+      "learning_rate": 4.964169415611393e-05,
+      "loss": 0.1506,
+      "step": 5530
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 1.0210274457931519,
+      "learning_rate": 4.9639711121161634e-05,
+      "loss": 0.1474,
+      "step": 5535
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.8528026342391968,
+      "learning_rate": 4.963772265364077e-05,
+      "loss": 0.1469,
+      "step": 5540
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.7811522483825684,
+      "learning_rate": 4.963572875398975e-05,
+      "loss": 0.1714,
+      "step": 5545
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.6667211651802063,
+      "learning_rate": 4.96337294226482e-05,
+      "loss": 0.1428,
+      "step": 5550
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 3.62852144241333,
+      "learning_rate": 4.963172466005691e-05,
+      "loss": 0.1718,
+      "step": 5555
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.8301425576210022,
+      "learning_rate": 4.962971446665792e-05,
+      "loss": 0.1492,
+      "step": 5560
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.9716572761535645,
+      "learning_rate": 4.962769884289441e-05,
+      "loss": 0.13,
+      "step": 5565
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.7408800721168518,
+      "learning_rate": 4.9625677789210804e-05,
+      "loss": 0.136,
+      "step": 5570
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.7961171865463257,
+      "learning_rate": 4.96236513060527e-05,
+      "loss": 0.1404,
+      "step": 5575
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.0075358152389526,
+      "learning_rate": 4.962161939386689e-05,
+      "loss": 0.1518,
+      "step": 5580
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 3.7874200344085693,
+      "learning_rate": 4.961958205310139e-05,
+      "loss": 0.13,
+      "step": 5585
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 1.0250333547592163,
+      "learning_rate": 4.961753928420537e-05,
+      "loss": 0.1612,
+      "step": 5590
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 1.419983148574829,
+      "learning_rate": 4.9615491087629226e-05,
+      "loss": 0.1582,
+      "step": 5595
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.9660770297050476,
+      "learning_rate": 4.961343746382455e-05,
+      "loss": 0.1447,
+      "step": 5600
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.9092869162559509,
+      "learning_rate": 4.961137841324413e-05,
+      "loss": 0.138,
+      "step": 5605
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 1.4980789422988892,
+      "learning_rate": 4.960931393634194e-05,
+      "loss": 0.1232,
+      "step": 5610
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.8407796025276184,
+      "learning_rate": 4.9607244033573156e-05,
+      "loss": 0.1914,
+      "step": 5615
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.9257882833480835,
+      "learning_rate": 4.960516870539414e-05,
+      "loss": 0.149,
+      "step": 5620
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.8622299432754517,
+      "learning_rate": 4.960308795226248e-05,
+      "loss": 0.1736,
+      "step": 5625
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.5688603520393372,
+      "learning_rate": 4.960100177463693e-05,
+      "loss": 0.1305,
+      "step": 5630
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.9023870229721069,
+      "learning_rate": 4.9598910172977454e-05,
+      "loss": 0.1341,
+      "step": 5635
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.6877835392951965,
+      "learning_rate": 4.959681314774521e-05,
+      "loss": 0.147,
+      "step": 5640
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.848576009273529,
+      "learning_rate": 4.959471069940255e-05,
+      "loss": 0.1626,
+      "step": 5645
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.6969914436340332,
+      "learning_rate": 4.959260282841302e-05,
+      "loss": 0.1184,
+      "step": 5650
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.6146367788314819,
+      "learning_rate": 4.9590489535241355e-05,
+      "loss": 0.1608,
+      "step": 5655
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.6508395671844482,
+      "learning_rate": 4.9588370820353516e-05,
+      "loss": 0.135,
+      "step": 5660
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.8212695121765137,
+      "learning_rate": 4.958624668421662e-05,
+      "loss": 0.1307,
+      "step": 5665
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.6195183396339417,
+      "learning_rate": 4.9584117127299e-05,
+      "loss": 0.1504,
+      "step": 5670
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.0033502578735352,
+      "learning_rate": 4.9581982150070194e-05,
+      "loss": 0.1425,
+      "step": 5675
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.2747350931167603,
+      "learning_rate": 4.95798417530009e-05,
+      "loss": 0.1367,
+      "step": 5680
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.3030251264572144,
+      "learning_rate": 4.9577695936563053e-05,
+      "loss": 0.1336,
+      "step": 5685
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.6721092462539673,
+      "learning_rate": 4.957554470122975e-05,
+      "loss": 0.1318,
+      "step": 5690
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.8976878523826599,
+      "learning_rate": 4.957338804747531e-05,
+      "loss": 0.1447,
+      "step": 5695
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.5446593165397644,
+      "learning_rate": 4.957122597577521e-05,
+      "loss": 0.1352,
+      "step": 5700
+    },
+    {
+      "epoch": 2.15,
+      "eval_loss": 0.13123564422130585,
+      "eval_runtime": 328.6758,
+      "eval_samples_per_second": 69.342,
+      "eval_steps_per_second": 8.668,
+      "eval_wer": 0.10461956521739131,
+      "step": 5700
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.7231523394584656,
+      "learning_rate": 4.956905848660617e-05,
+      "loss": 0.134,
+      "step": 5705
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.7739360332489014,
+      "learning_rate": 4.9566885580446065e-05,
+      "loss": 0.1312,
+      "step": 5710
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.5566073060035706,
+      "learning_rate": 4.956470725777397e-05,
+      "loss": 0.1288,
+      "step": 5715
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.5600452423095703,
+      "learning_rate": 4.956296070007038e-05,
+      "loss": 0.13,
+      "step": 5720
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.6121380925178528,
+      "learning_rate": 4.956077262888782e-05,
+      "loss": 0.122,
+      "step": 5725
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.6983882784843445,
+      "learning_rate": 4.9558579142541075e-05,
+      "loss": 0.1529,
+      "step": 5730
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.6412332653999329,
+      "learning_rate": 4.955638024151374e-05,
+      "loss": 0.1238,
+      "step": 5735
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.6804934740066528,
+      "learning_rate": 4.955417592629065e-05,
+      "loss": 0.1097,
+      "step": 5740
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.8537352681159973,
+      "learning_rate": 4.955196619735781e-05,
+      "loss": 0.1415,
+      "step": 5745
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 1.7510582208633423,
+      "learning_rate": 4.954975105520241e-05,
+      "loss": 0.1393,
+      "step": 5750
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.7123726606369019,
+      "learning_rate": 4.954753050031285e-05,
+      "loss": 0.1427,
+      "step": 5755
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.985660970211029,
+      "learning_rate": 4.954530453317874e-05,
+      "loss": 0.1444,
+      "step": 5760
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.5345569252967834,
+      "learning_rate": 4.954307315429082e-05,
+      "loss": 0.1443,
+      "step": 5765
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.6254405379295349,
+      "learning_rate": 4.9540836364141094e-05,
+      "loss": 0.1309,
+      "step": 5770
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.9977956414222717,
+      "learning_rate": 4.953859416322273e-05,
+      "loss": 0.143,
+      "step": 5775
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.7780291438102722,
+      "learning_rate": 4.9536346552030075e-05,
+      "loss": 0.1363,
+      "step": 5780
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 1.019315481185913,
+      "learning_rate": 4.95340935310587e-05,
+      "loss": 0.1391,
+      "step": 5785
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.678236186504364,
+      "learning_rate": 4.953183510080534e-05,
+      "loss": 0.1383,
+      "step": 5790
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.6958431005477905,
+      "learning_rate": 4.9529571261767924e-05,
+      "loss": 0.1381,
+      "step": 5795
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 1.6696650981903076,
+      "learning_rate": 4.95273020144456e-05,
+      "loss": 0.133,
+      "step": 5800
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.5488206148147583,
+      "learning_rate": 4.9525027359338696e-05,
+      "loss": 0.1312,
+      "step": 5805
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.795659065246582,
+      "learning_rate": 4.9522747296948706e-05,
+      "loss": 0.1362,
+      "step": 5810
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.6193950176239014,
+      "learning_rate": 4.9520461827778365e-05,
+      "loss": 0.1152,
+      "step": 5815
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.5706254839897156,
+      "learning_rate": 4.951817095233156e-05,
+      "loss": 0.1217,
+      "step": 5820
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 1.1922762393951416,
+      "learning_rate": 4.951587467111337e-05,
+      "loss": 0.1551,
+      "step": 5825
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 2.589902639389038,
+      "learning_rate": 4.9513572984630105e-05,
+      "loss": 0.1288,
+      "step": 5830
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.8596647381782532,
+      "learning_rate": 4.951126589338924e-05,
+      "loss": 0.1255,
+      "step": 5835
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.6189730167388916,
+      "learning_rate": 4.950895339789942e-05,
+      "loss": 0.1372,
+      "step": 5840
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.8075405359268188,
+      "learning_rate": 4.950663549867053e-05,
+      "loss": 0.1333,
+      "step": 5845
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.334312915802002,
+      "learning_rate": 4.9504312196213596e-05,
+      "loss": 0.1307,
+      "step": 5850
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.5502147674560547,
+      "learning_rate": 4.9501983491040885e-05,
+      "loss": 0.1304,
+      "step": 5855
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.9113101959228516,
+      "learning_rate": 4.949964938366581e-05,
+      "loss": 0.1431,
+      "step": 5860
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 2.66496205329895,
+      "learning_rate": 4.949730987460301e-05,
+      "loss": 0.1348,
+      "step": 5865
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.879149854183197,
+      "learning_rate": 4.9494964964368296e-05,
+      "loss": 0.1247,
+      "step": 5870
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.491457462310791,
+      "learning_rate": 4.949261465347866e-05,
+      "loss": 0.1368,
+      "step": 5875
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 2.164935350418091,
+      "learning_rate": 4.949025894245233e-05,
+      "loss": 0.1444,
+      "step": 5880
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 1.0364428758621216,
+      "learning_rate": 4.9487897831808664e-05,
+      "loss": 0.1437,
+      "step": 5885
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.9372974634170532,
+      "learning_rate": 4.948600505591905e-05,
+      "loss": 0.1528,
+      "step": 5890
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.8672482371330261,
+      "learning_rate": 4.9483634227276864e-05,
+      "loss": 0.1591,
+      "step": 5895
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.777405858039856,
+      "learning_rate": 4.948125800047797e-05,
+      "loss": 0.1567,
+      "step": 5900
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.931729793548584,
+      "learning_rate": 4.947887637604628e-05,
+      "loss": 0.1405,
+      "step": 5905
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.7462069988250732,
+      "learning_rate": 4.947648935450689e-05,
+      "loss": 0.1487,
+      "step": 5910
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.7155471444129944,
+      "learning_rate": 4.9474096936386104e-05,
+      "loss": 0.1534,
+      "step": 5915
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.8451410531997681,
+      "learning_rate": 4.9471699122211404e-05,
+      "loss": 0.1316,
+      "step": 5920
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 1.0760977268218994,
+      "learning_rate": 4.946929591251144e-05,
+      "loss": 0.1572,
+      "step": 5925
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.7779541015625,
+      "learning_rate": 4.946688730781609e-05,
+      "loss": 0.1527,
+      "step": 5930
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 1.0388118028640747,
+      "learning_rate": 4.94644733086564e-05,
+      "loss": 0.1323,
+      "step": 5935
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.7712625861167908,
+      "learning_rate": 4.946205391556461e-05,
+      "loss": 0.1322,
+      "step": 5940
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 3.1851577758789062,
+      "learning_rate": 4.9459629129074136e-05,
+      "loss": 0.1628,
+      "step": 5945
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.7296850085258484,
+      "learning_rate": 4.9457198949719605e-05,
+      "loss": 0.1545,
+      "step": 5950
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.7276253700256348,
+      "learning_rate": 4.9454763378036834e-05,
+      "loss": 0.14,
+      "step": 5955
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 1.5968475341796875,
+      "learning_rate": 4.94523224145628e-05,
+      "loss": 0.1638,
+      "step": 5960
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 2.794768810272217,
+      "learning_rate": 4.94498760598357e-05,
+      "loss": 0.1352,
+      "step": 5965
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.967302680015564,
+      "learning_rate": 4.9447424314394895e-05,
+      "loss": 0.1432,
+      "step": 5970
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.7929945588111877,
+      "learning_rate": 4.944496717878096e-05,
+      "loss": 0.1356,
+      "step": 5975
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.1957675218582153,
+      "learning_rate": 4.944250465353564e-05,
+      "loss": 0.1881,
+      "step": 5980
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.1174230575561523,
+      "learning_rate": 4.9440036739201865e-05,
+      "loss": 0.1168,
+      "step": 5985
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 2.840014934539795,
+      "learning_rate": 4.9437563436323776e-05,
+      "loss": 0.1292,
+      "step": 5990
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 3.749016284942627,
+      "learning_rate": 4.9435084745446666e-05,
+      "loss": 0.1722,
+      "step": 5995
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.802821159362793,
+      "learning_rate": 4.9432600667117056e-05,
+      "loss": 0.1508,
+      "step": 6000
+    },
+    {
+      "epoch": 2.26,
+      "eval_loss": 0.13933046162128448,
+      "eval_runtime": 328.7167,
+      "eval_samples_per_second": 69.333,
+      "eval_steps_per_second": 8.667,
+      "eval_wer": 0.10712601747071669,
+      "step": 6000
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 1.366395354270935,
+      "learning_rate": 4.9430111201882635e-05,
+      "loss": 0.163,
+      "step": 6005
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.9769381880760193,
+      "learning_rate": 4.942761635029227e-05,
+      "loss": 0.143,
+      "step": 6010
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 1.250303030014038,
+      "learning_rate": 4.9425116112896033e-05,
+      "loss": 0.1598,
+      "step": 6015
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.9101468324661255,
+      "learning_rate": 4.942261049024518e-05,
+      "loss": 0.1476,
+      "step": 6020
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.7861135005950928,
+      "learning_rate": 4.9420099482892144e-05,
+      "loss": 0.1778,
+      "step": 6025
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.938271164894104,
+      "learning_rate": 4.941758309139055e-05,
+      "loss": 0.1445,
+      "step": 6030
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.7871306538581848,
+      "learning_rate": 4.941506131629523e-05,
+      "loss": 0.1294,
+      "step": 6035
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.8223226070404053,
+      "learning_rate": 4.941253415816217e-05,
+      "loss": 0.119,
+      "step": 6040
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.193043828010559,
+      "learning_rate": 4.941000161754855e-05,
+      "loss": 0.1612,
+      "step": 6045
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.8297756910324097,
+      "learning_rate": 4.9407463695012776e-05,
+      "loss": 0.1518,
+      "step": 6050
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.9389541745185852,
+      "learning_rate": 4.940492039111438e-05,
+      "loss": 0.1318,
+      "step": 6055
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.9064432978630066,
+      "learning_rate": 4.940237170641412e-05,
+      "loss": 0.1398,
+      "step": 6060
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.2143980264663696,
+      "learning_rate": 4.9399817641473935e-05,
+      "loss": 0.1596,
+      "step": 6065
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.6395060420036316,
+      "learning_rate": 4.939725819685694e-05,
+      "loss": 0.1597,
+      "step": 6070
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 2.3046085834503174,
+      "learning_rate": 4.939469337312745e-05,
+      "loss": 0.1331,
+      "step": 6075
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.6680417656898499,
+      "learning_rate": 4.9392123170850944e-05,
+      "loss": 0.1398,
+      "step": 6080
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 1.0138882398605347,
+      "learning_rate": 4.9389547590594116e-05,
+      "loss": 0.1505,
+      "step": 6085
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 1.4139388799667358,
+      "learning_rate": 4.938696663292482e-05,
+      "loss": 0.1488,
+      "step": 6090
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 1.381995439529419,
+      "learning_rate": 4.9384380298412106e-05,
+      "loss": 0.1428,
+      "step": 6095
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 1.3032971620559692,
+      "learning_rate": 4.938178858762622e-05,
+      "loss": 0.1555,
+      "step": 6100
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 1.0044642686843872,
+      "learning_rate": 4.9379191501138565e-05,
+      "loss": 0.1366,
+      "step": 6105
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 1.2266291379928589,
+      "learning_rate": 4.937658903952177e-05,
+      "loss": 0.1638,
+      "step": 6110
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.9061530828475952,
+      "learning_rate": 4.9373981203349605e-05,
+      "loss": 0.1646,
+      "step": 6115
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.7566392421722412,
+      "learning_rate": 4.937136799319706e-05,
+      "loss": 0.1299,
+      "step": 6120
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 1.0662040710449219,
+      "learning_rate": 4.936874940964029e-05,
+      "loss": 0.1718,
+      "step": 6125
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.9356708526611328,
+      "learning_rate": 4.936612545325665e-05,
+      "loss": 0.139,
+      "step": 6130
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.9278246760368347,
+      "learning_rate": 4.9363496124624654e-05,
+      "loss": 0.1599,
+      "step": 6135
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.9249376654624939,
+      "learning_rate": 4.936086142432403e-05,
+      "loss": 0.1255,
+      "step": 6140
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.6730533838272095,
+      "learning_rate": 4.9358221352935666e-05,
+      "loss": 0.1451,
+      "step": 6145
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 1.7042384147644043,
+      "learning_rate": 4.9355575911041666e-05,
+      "loss": 0.1442,
+      "step": 6150
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.843647301197052,
+      "learning_rate": 4.935292509922527e-05,
+      "loss": 0.1421,
+      "step": 6155
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.685804009437561,
+      "learning_rate": 4.935026891807095e-05,
+      "loss": 0.1259,
+      "step": 6160
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.951064944267273,
+      "learning_rate": 4.934760736816434e-05,
+      "loss": 0.135,
+      "step": 6165
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.830935537815094,
+      "learning_rate": 4.9344940450092246e-05,
+      "loss": 0.1448,
+      "step": 6170
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.9615690112113953,
+      "learning_rate": 4.934226816444268e-05,
+      "loss": 0.128,
+      "step": 6175
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.2887710332870483,
+      "learning_rate": 4.933959051180483e-05,
+      "loss": 0.1399,
+      "step": 6180
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.1694103479385376,
+      "learning_rate": 4.933690749276906e-05,
+      "loss": 0.1232,
+      "step": 6185
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.5503979921340942,
+      "learning_rate": 4.9334219107926924e-05,
+      "loss": 0.1485,
+      "step": 6190
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.8017725944519043,
+      "learning_rate": 4.933152535787116e-05,
+      "loss": 0.1314,
+      "step": 6195
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.3640562295913696,
+      "learning_rate": 4.932882624319568e-05,
+      "loss": 0.1812,
+      "step": 6200
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.7592419981956482,
+      "learning_rate": 4.9326121764495596e-05,
+      "loss": 0.1646,
+      "step": 6205
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.9565640687942505,
+      "learning_rate": 4.932341192236719e-05,
+      "loss": 0.1194,
+      "step": 6210
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.8074353337287903,
+      "learning_rate": 4.9320696717407924e-05,
+      "loss": 0.1351,
+      "step": 6215
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 2.460293769836426,
+      "learning_rate": 4.931797615021644e-05,
+      "loss": 0.1449,
+      "step": 6220
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.4247808456420898,
+      "learning_rate": 4.931525022139258e-05,
+      "loss": 0.1478,
+      "step": 6225
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.9104633927345276,
+      "learning_rate": 4.9312518931537354e-05,
+      "loss": 0.135,
+      "step": 6230
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.7951221466064453,
+      "learning_rate": 4.930978228125296e-05,
+      "loss": 0.1385,
+      "step": 6235
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 1.1165971755981445,
+      "learning_rate": 4.930704027114277e-05,
+      "loss": 0.1438,
+      "step": 6240
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.6317349672317505,
+      "learning_rate": 4.9304292901811354e-05,
+      "loss": 0.1393,
+      "step": 6245
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.8548485040664673,
+      "learning_rate": 4.9301540173864434e-05,
+      "loss": 0.1151,
+      "step": 6250
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.8319147825241089,
+      "learning_rate": 4.929878208790895e-05,
+      "loss": 0.1463,
+      "step": 6255
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.7275090217590332,
+      "learning_rate": 4.9296018644553e-05,
+      "loss": 0.1488,
+      "step": 6260
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.7814785838127136,
+      "learning_rate": 4.9293249844405856e-05,
+      "loss": 0.153,
+      "step": 6265
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.6442232728004456,
+      "learning_rate": 4.9290475688078e-05,
+      "loss": 0.1269,
+      "step": 6270
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 1.9818882942199707,
+      "learning_rate": 4.928769617618107e-05,
+      "loss": 0.1331,
+      "step": 6275
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.6372818946838379,
+      "learning_rate": 4.928491130932791e-05,
+      "loss": 0.1374,
+      "step": 6280
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 1.0020110607147217,
+      "learning_rate": 4.92821210881325e-05,
+      "loss": 0.1387,
+      "step": 6285
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.733284592628479,
+      "learning_rate": 4.927932551321004e-05,
+      "loss": 0.1257,
+      "step": 6290
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.8808489441871643,
+      "learning_rate": 4.927652458517691e-05,
+      "loss": 0.131,
+      "step": 6295
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.6504075527191162,
+      "learning_rate": 4.927371830465065e-05,
+      "loss": 0.1438,
+      "step": 6300
+    },
+    {
+      "epoch": 2.37,
+      "eval_loss": 0.1316407471895218,
+      "eval_runtime": 358.3694,
+      "eval_samples_per_second": 63.596,
+      "eval_steps_per_second": 7.95,
+      "eval_wer": 0.10251224273707894,
+      "step": 6300
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.7966932654380798,
+      "learning_rate": 4.927090667224998e-05,
+      "loss": 0.1408,
+      "step": 6305
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 1.3176876306533813,
+      "learning_rate": 4.926808968859483e-05,
+      "loss": 0.1509,
+      "step": 6310
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 1.2390415668487549,
+      "learning_rate": 4.926526735430628e-05,
+      "loss": 0.1492,
+      "step": 6315
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.6468360424041748,
+      "learning_rate": 4.926243967000659e-05,
+      "loss": 0.1181,
+      "step": 6320
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.5288561582565308,
+      "learning_rate": 4.925960663631922e-05,
+      "loss": 0.1342,
+      "step": 6325
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.8845801949501038,
+      "learning_rate": 4.925676825386879e-05,
+      "loss": 0.1393,
+      "step": 6330
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.7463661432266235,
+      "learning_rate": 4.925392452328112e-05,
+      "loss": 0.1695,
+      "step": 6335
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 1.140329360961914,
+      "learning_rate": 4.925107544518317e-05,
+      "loss": 0.1342,
+      "step": 6340
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.6256007552146912,
+      "learning_rate": 4.9248221020203136e-05,
+      "loss": 0.1476,
+      "step": 6345
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.8086591958999634,
+      "learning_rate": 4.924536124897035e-05,
+      "loss": 0.1304,
+      "step": 6350
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 1.1309864521026611,
+      "learning_rate": 4.9242496132115324e-05,
+      "loss": 0.1744,
+      "step": 6355
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.7822628617286682,
+      "learning_rate": 4.9239625670269765e-05,
+      "loss": 0.1283,
+      "step": 6360
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.5437459349632263,
+      "learning_rate": 4.92373254528254e-05,
+      "loss": 0.1216,
+      "step": 6365
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.7769615650177002,
+      "learning_rate": 4.923444537159255e-05,
+      "loss": 0.1395,
+      "step": 6370
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.2147082090377808,
+      "learning_rate": 4.9231559947144204e-05,
+      "loss": 0.1288,
+      "step": 6375
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.9443001747131348,
+      "learning_rate": 4.9228669180116535e-05,
+      "loss": 0.1322,
+      "step": 6380
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.0771030187606812,
+      "learning_rate": 4.922577307114691e-05,
+      "loss": 0.1134,
+      "step": 6385
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 1.8807294368743896,
+      "learning_rate": 4.9222871620873856e-05,
+      "loss": 0.1259,
+      "step": 6390
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.9258590936660767,
+      "learning_rate": 4.921996482993709e-05,
+      "loss": 0.1292,
+      "step": 6395
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.7376443147659302,
+      "learning_rate": 4.9217052698977506e-05,
+      "loss": 0.1338,
+      "step": 6400
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.929614245891571,
+      "learning_rate": 4.921413522863716e-05,
+      "loss": 0.1232,
+      "step": 6405
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.9406668543815613,
+      "learning_rate": 4.9211212419559305e-05,
+      "loss": 0.1522,
+      "step": 6410
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.6864051818847656,
+      "learning_rate": 4.920828427238836e-05,
+      "loss": 0.1652,
+      "step": 6415
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.8289921283721924,
+      "learning_rate": 4.920535078776993e-05,
+      "loss": 0.1275,
+      "step": 6420
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.0505247116088867,
+      "learning_rate": 4.9202411966350784e-05,
+      "loss": 0.1195,
+      "step": 6425
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.8016154170036316,
+      "learning_rate": 4.9199467808778864e-05,
+      "loss": 0.1486,
+      "step": 6430
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.7477118968963623,
+      "learning_rate": 4.919651831570331e-05,
+      "loss": 0.1907,
+      "step": 6435
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.8171904683113098,
+      "learning_rate": 4.9193563487774434e-05,
+      "loss": 0.1618,
+      "step": 6440
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 26.902006149291992,
+      "learning_rate": 4.9190603325643705e-05,
+      "loss": 0.1716,
+      "step": 6445
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.6617954969406128,
+      "learning_rate": 4.918763782996378e-05,
+      "loss": 0.1382,
+      "step": 6450
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 8.283758163452148,
+      "learning_rate": 4.9184667001388504e-05,
+      "loss": 0.1409,
+      "step": 6455
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.8804893493652344,
+      "learning_rate": 4.918169084057287e-05,
+      "loss": 0.149,
+      "step": 6460
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 1.7051013708114624,
+      "learning_rate": 4.917870934817307e-05,
+      "loss": 0.1517,
+      "step": 6465
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.9064131379127502,
+      "learning_rate": 4.917572252484647e-05,
+      "loss": 0.1209,
+      "step": 6470
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.0296118259429932,
+      "learning_rate": 4.91727303712516e-05,
+      "loss": 0.1386,
+      "step": 6475
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.1122374534606934,
+      "learning_rate": 4.9169732888048165e-05,
+      "loss": 0.1335,
+      "step": 6480
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.1262705326080322,
+      "learning_rate": 4.916673007589705e-05,
+      "loss": 0.136,
+      "step": 6485
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.9449793696403503,
+      "learning_rate": 4.9163721935460337e-05,
+      "loss": 0.1371,
+      "step": 6490
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.732397198677063,
+      "learning_rate": 4.916070846740124e-05,
+      "loss": 0.1296,
+      "step": 6495
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.6037105917930603,
+      "learning_rate": 4.9157689672384174e-05,
+      "loss": 0.1361,
+      "step": 6500
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 1.0030683279037476,
+      "learning_rate": 4.915466555107472e-05,
+      "loss": 0.1392,
+      "step": 6505
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.7280795574188232,
+      "learning_rate": 4.915163610413965e-05,
+      "loss": 0.142,
+      "step": 6510
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.8218936324119568,
+      "learning_rate": 4.914860133224688e-05,
+      "loss": 0.1409,
+      "step": 6515
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.8449922204017639,
+      "learning_rate": 4.914556123606553e-05,
+      "loss": 0.1401,
+      "step": 6520
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.9599225521087646,
+      "learning_rate": 4.914251581626588e-05,
+      "loss": 0.1388,
+      "step": 6525
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 1.8811895847320557,
+      "learning_rate": 4.9139465073519374e-05,
+      "loss": 0.1437,
+      "step": 6530
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.7941332459449768,
+      "learning_rate": 4.913640900849866e-05,
+      "loss": 0.1615,
+      "step": 6535
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 1.022691249847412,
+      "learning_rate": 4.913334762187752e-05,
+      "loss": 0.1291,
+      "step": 6540
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.9060131311416626,
+      "learning_rate": 4.913028091433093e-05,
+      "loss": 0.1404,
+      "step": 6545
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.8678662776947021,
+      "learning_rate": 4.912720888653506e-05,
+      "loss": 0.1487,
+      "step": 6550
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 2.1097733974456787,
+      "learning_rate": 4.9124131539167206e-05,
+      "loss": 0.1347,
+      "step": 6555
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 1.1271727085113525,
+      "learning_rate": 4.912104887290587e-05,
+      "loss": 0.1301,
+      "step": 6560
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.9581430554389954,
+      "learning_rate": 4.911796088843074e-05,
+      "loss": 0.1411,
+      "step": 6565
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.9186280965805054,
+      "learning_rate": 4.9114867586422617e-05,
+      "loss": 0.1361,
+      "step": 6570
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 2.2340848445892334,
+      "learning_rate": 4.9111768967563546e-05,
+      "loss": 0.1535,
+      "step": 6575
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.071394681930542,
+      "learning_rate": 4.9108665032536694e-05,
+      "loss": 0.1587,
+      "step": 6580
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.1373658180236816,
+      "learning_rate": 4.910555578202641e-05,
+      "loss": 0.1287,
+      "step": 6585
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.9361284375190735,
+      "learning_rate": 4.9102441216718257e-05,
+      "loss": 0.1328,
+      "step": 6590
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 2.0420515537261963,
+      "learning_rate": 4.909932133729889e-05,
+      "loss": 0.1652,
+      "step": 6595
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.7470790147781372,
+      "learning_rate": 4.909619614445621e-05,
+      "loss": 0.1413,
+      "step": 6600
+    },
+    {
+      "epoch": 2.49,
+      "eval_loss": 0.13696463406085968,
+      "eval_runtime": 346.6686,
+      "eval_samples_per_second": 65.743,
+      "eval_steps_per_second": 8.218,
+      "eval_wer": 0.10719012639798822,
+      "step": 6600
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.6145943999290466,
+      "learning_rate": 4.909306563887925e-05,
+      "loss": 0.1393,
+      "step": 6605
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.6376383304595947,
+      "learning_rate": 4.9089929821258234e-05,
+      "loss": 0.1449,
+      "step": 6610
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.7879248261451721,
+      "learning_rate": 4.9086788692284534e-05,
+      "loss": 0.1429,
+      "step": 6615
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.5231431126594543,
+      "learning_rate": 4.908364225265072e-05,
+      "loss": 0.1314,
+      "step": 6620
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.7156097888946533,
+      "learning_rate": 4.908049050305051e-05,
+      "loss": 0.1388,
+      "step": 6625
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.7103233337402344,
+      "learning_rate": 4.90773334441788e-05,
+      "loss": 0.1306,
+      "step": 6630
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.8446791172027588,
+      "learning_rate": 4.907417107673167e-05,
+      "loss": 0.1552,
+      "step": 6635
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.9607967734336853,
+      "learning_rate": 4.907100340140636e-05,
+      "loss": 0.1404,
+      "step": 6640
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.760793149471283,
+      "learning_rate": 4.906783041890126e-05,
+      "loss": 0.1246,
+      "step": 6645
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.246391773223877,
+      "learning_rate": 4.906465212991597e-05,
+      "loss": 0.1537,
+      "step": 6650
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 2.0128471851348877,
+      "learning_rate": 4.906146853515123e-05,
+      "loss": 0.113,
+      "step": 6655
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 1.1551294326782227,
+      "learning_rate": 4.905827963530897e-05,
+      "loss": 0.1453,
+      "step": 6660
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 1.5330767631530762,
+      "learning_rate": 4.9055085431092265e-05,
+      "loss": 0.1276,
+      "step": 6665
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.6763330101966858,
+      "learning_rate": 4.905188592320538e-05,
+      "loss": 0.1254,
+      "step": 6670
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.9245719313621521,
+      "learning_rate": 4.904868111235374e-05,
+      "loss": 0.1272,
+      "step": 6675
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 1.0257134437561035,
+      "learning_rate": 4.9045470999243956e-05,
+      "loss": 0.1464,
+      "step": 6680
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 1.0971448421478271,
+      "learning_rate": 4.9042255584583786e-05,
+      "loss": 0.1328,
+      "step": 6685
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 1.0139062404632568,
+      "learning_rate": 4.903903486908216e-05,
+      "loss": 0.1395,
+      "step": 6690
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.7547524571418762,
+      "learning_rate": 4.9035808853449185e-05,
+      "loss": 0.1339,
+      "step": 6695
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.7293165326118469,
+      "learning_rate": 4.903257753839614e-05,
+      "loss": 0.1174,
+      "step": 6700
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.8482643961906433,
+      "learning_rate": 4.902934092463546e-05,
+      "loss": 0.1342,
+      "step": 6705
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 11.732702255249023,
+      "learning_rate": 4.902609901288075e-05,
+      "loss": 0.1564,
+      "step": 6710
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.9433574080467224,
+      "learning_rate": 4.902285180384679e-05,
+      "loss": 0.189,
+      "step": 6715
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.6246347427368164,
+      "learning_rate": 4.9019599298249535e-05,
+      "loss": 0.1317,
+      "step": 6720
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.7281513810157776,
+      "learning_rate": 4.9016341496806087e-05,
+      "loss": 0.1449,
+      "step": 6725
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.7813048958778381,
+      "learning_rate": 4.9013078400234736e-05,
+      "loss": 0.1293,
+      "step": 6730
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.921598494052887,
+      "learning_rate": 4.900981000925492e-05,
+      "loss": 0.1333,
+      "step": 6735
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.7274336814880371,
+      "learning_rate": 4.900653632458725e-05,
+      "loss": 0.1632,
+      "step": 6740
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.8828667402267456,
+      "learning_rate": 4.900325734695353e-05,
+      "loss": 0.134,
+      "step": 6745
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.6555948853492737,
+      "learning_rate": 4.8999973077076696e-05,
+      "loss": 0.1491,
+      "step": 6750
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.7132114768028259,
+      "learning_rate": 4.8996683515680866e-05,
+      "loss": 0.1269,
+      "step": 6755
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.5767226815223694,
+      "learning_rate": 4.899338866349132e-05,
+      "loss": 0.1394,
+      "step": 6760
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.6916750073432922,
+      "learning_rate": 4.899008852123451e-05,
+      "loss": 0.1304,
+      "step": 6765
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.732676088809967,
+      "learning_rate": 4.898678308963806e-05,
+      "loss": 0.1238,
+      "step": 6770
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.8567782044410706,
+      "learning_rate": 4.8983472369430736e-05,
+      "loss": 0.1279,
+      "step": 6775
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 1.0105549097061157,
+      "learning_rate": 4.89801563613425e-05,
+      "loss": 0.1289,
+      "step": 6780
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.6153063774108887,
+      "learning_rate": 4.897683506610447e-05,
+      "loss": 0.154,
+      "step": 6785
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 1.3825997114181519,
+      "learning_rate": 4.89735084844489e-05,
+      "loss": 0.151,
+      "step": 6790
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 2.300419330596924,
+      "learning_rate": 4.8970176617109264e-05,
+      "loss": 0.1499,
+      "step": 6795
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.7483996152877808,
+      "learning_rate": 4.896683946482016e-05,
+      "loss": 0.142,
+      "step": 6800
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.5335126519203186,
+      "learning_rate": 4.896349702831736e-05,
+      "loss": 0.1198,
+      "step": 6805
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 1.705177664756775,
+      "learning_rate": 4.896014930833782e-05,
+      "loss": 0.1381,
+      "step": 6810
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 3.3025259971618652,
+      "learning_rate": 4.895679630561963e-05,
+      "loss": 0.1553,
+      "step": 6815
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.6509557366371155,
+      "learning_rate": 4.895343802090207e-05,
+      "loss": 0.1321,
+      "step": 6820
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.9139301776885986,
+      "learning_rate": 4.895007445492557e-05,
+      "loss": 0.1471,
+      "step": 6825
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.8868838548660278,
+      "learning_rate": 4.894670560843174e-05,
+      "loss": 0.1353,
+      "step": 6830
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 4.048460483551025,
+      "learning_rate": 4.8943331482163335e-05,
+      "loss": 0.148,
+      "step": 6835
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 1.6404019594192505,
+      "learning_rate": 4.8939952076864274e-05,
+      "loss": 0.16,
+      "step": 6840
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 1.2283729314804077,
+      "learning_rate": 4.893656739327967e-05,
+      "loss": 0.1967,
+      "step": 6845
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.7045356035232544,
+      "learning_rate": 4.893317743215576e-05,
+      "loss": 0.1424,
+      "step": 6850
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 1.1238758563995361,
+      "learning_rate": 4.892978219423998e-05,
+      "loss": 0.1433,
+      "step": 6855
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 1.80108642578125,
+      "learning_rate": 4.8926381680280906e-05,
+      "loss": 0.163,
+      "step": 6860
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.6069390177726746,
+      "learning_rate": 4.892297589102827e-05,
+      "loss": 0.1373,
+      "step": 6865
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.9916033148765564,
+      "learning_rate": 4.8919564827233e-05,
+      "loss": 0.1488,
+      "step": 6870
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 1.271288275718689,
+      "learning_rate": 4.8916148489647164e-05,
+      "loss": 0.1401,
+      "step": 6875
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 1.2095807790756226,
+      "learning_rate": 4.8912726879023994e-05,
+      "loss": 0.1535,
+      "step": 6880
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 7.420047283172607,
+      "learning_rate": 4.890929999611789e-05,
+      "loss": 0.1375,
+      "step": 6885
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.7246400713920593,
+      "learning_rate": 4.89058678416844e-05,
+      "loss": 0.1324,
+      "step": 6890
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 1.4963173866271973,
+      "learning_rate": 4.890243041648026e-05,
+      "loss": 0.1451,
+      "step": 6895
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 2.027754545211792,
+      "learning_rate": 4.889898772126336e-05,
+      "loss": 0.1456,
+      "step": 6900
+    },
+    {
+      "epoch": 2.6,
+      "eval_loss": 0.1360301375389099,
+      "eval_runtime": 357.2427,
+      "eval_samples_per_second": 63.797,
+      "eval_steps_per_second": 7.975,
+      "eval_wer": 0.1129309774336576,
+      "step": 6900
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 1.2794420719146729,
+      "learning_rate": 4.889553975679272e-05,
+      "loss": 0.1702,
+      "step": 6905
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 2.196462392807007,
+      "learning_rate": 4.8892086523828574e-05,
+      "loss": 0.1277,
+      "step": 6910
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.5619906783103943,
+      "learning_rate": 4.8888628023132275e-05,
+      "loss": 0.1425,
+      "step": 6915
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.6266874670982361,
+      "learning_rate": 4.888516425546636e-05,
+      "loss": 0.127,
+      "step": 6920
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 1.1283832788467407,
+      "learning_rate": 4.888169522159453e-05,
+      "loss": 0.1389,
+      "step": 6925
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.9725835919380188,
+      "learning_rate": 4.887822092228162e-05,
+      "loss": 0.134,
+      "step": 6930
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.6389516592025757,
+      "learning_rate": 4.887474135829366e-05,
+      "loss": 0.1373,
+      "step": 6935
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.6490597128868103,
+      "learning_rate": 4.887125653039782e-05,
+      "loss": 0.1234,
+      "step": 6940
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.6022095084190369,
+      "learning_rate": 4.886776643936244e-05,
+      "loss": 0.1248,
+      "step": 6945
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 1.0632058382034302,
+      "learning_rate": 4.8864271085957e-05,
+      "loss": 0.1355,
+      "step": 6950
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.9414571523666382,
+      "learning_rate": 4.8860770470952165e-05,
+      "loss": 0.1322,
+      "step": 6955
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 2.2274961471557617,
+      "learning_rate": 4.8857264595119764e-05,
+      "loss": 0.1324,
+      "step": 6960
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 1.054241418838501,
+      "learning_rate": 4.885375345923275e-05,
+      "loss": 0.1308,
+      "step": 6965
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 1.3285417556762695,
+      "learning_rate": 4.8850237064065286e-05,
+      "loss": 0.1256,
+      "step": 6970
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.6504650115966797,
+      "learning_rate": 4.884671541039265e-05,
+      "loss": 0.1385,
+      "step": 6975
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.9060842394828796,
+      "learning_rate": 4.884318849899129e-05,
+      "loss": 0.1432,
+      "step": 6980
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.9944954514503479,
+      "learning_rate": 4.8839656330638836e-05,
+      "loss": 0.1338,
+      "step": 6985
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 1.9839235544204712,
+      "learning_rate": 4.883611890611406e-05,
+      "loss": 0.1416,
+      "step": 6990
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.586404025554657,
+      "learning_rate": 4.883257622619688e-05,
+      "loss": 0.126,
+      "step": 6995
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.9899946451187134,
+      "learning_rate": 4.88290282916684e-05,
+      "loss": 0.1372,
+      "step": 7000
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 1.381420373916626,
+      "learning_rate": 4.882547510331087e-05,
+      "loss": 0.1482,
+      "step": 7005
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.9300584197044373,
+      "learning_rate": 4.882191666190769e-05,
+      "loss": 0.1421,
+      "step": 7010
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 1.1738147735595703,
+      "learning_rate": 4.881835296824342e-05,
+      "loss": 0.1637,
+      "step": 7015
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.5128684043884277,
+      "learning_rate": 4.88147840231038e-05,
+      "loss": 0.1324,
+      "step": 7020
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.8674966096878052,
+      "learning_rate": 4.88112098272757e-05,
+      "loss": 0.1557,
+      "step": 7025
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 1.105479121208191,
+      "learning_rate": 4.8807630381547176e-05,
+      "loss": 0.1494,
+      "step": 7030
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.6220070719718933,
+      "learning_rate": 4.880404568670741e-05,
+      "loss": 0.1292,
+      "step": 7035
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 1.244698405265808,
+      "learning_rate": 4.8800455743546756e-05,
+      "loss": 0.1376,
+      "step": 7040
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 7.456350326538086,
+      "learning_rate": 4.879686055285673e-05,
+      "loss": 0.125,
+      "step": 7045
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.644436776638031,
+      "learning_rate": 4.8793260115429994e-05,
+      "loss": 0.1679,
+      "step": 7050
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.6966572999954224,
+      "learning_rate": 4.8789654432060385e-05,
+      "loss": 0.1236,
+      "step": 7055
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.6554312109947205,
+      "learning_rate": 4.878604350354288e-05,
+      "loss": 0.1329,
+      "step": 7060
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 2.1683268547058105,
+      "learning_rate": 4.878242733067361e-05,
+      "loss": 0.1437,
+      "step": 7065
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.5948340892791748,
+      "learning_rate": 4.877880591424987e-05,
+      "loss": 0.1436,
+      "step": 7070
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.6776853203773499,
+      "learning_rate": 4.8775179255070124e-05,
+      "loss": 0.1165,
+      "step": 7075
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.8031582236289978,
+      "learning_rate": 4.8771547353933965e-05,
+      "loss": 0.125,
+      "step": 7080
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.8894427418708801,
+      "learning_rate": 4.8767910211642165e-05,
+      "loss": 0.1337,
+      "step": 7085
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.5981336236000061,
+      "learning_rate": 4.8764267828996634e-05,
+      "loss": 0.1183,
+      "step": 7090
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 2.503417491912842,
+      "learning_rate": 4.876062020680045e-05,
+      "loss": 0.1528,
+      "step": 7095
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.6375592947006226,
+      "learning_rate": 4.8756967345857834e-05,
+      "loss": 0.1195,
+      "step": 7100
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.9869526624679565,
+      "learning_rate": 4.8753309246974185e-05,
+      "loss": 0.1295,
+      "step": 7105
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 1.4211264848709106,
+      "learning_rate": 4.8749645910956025e-05,
+      "loss": 0.1237,
+      "step": 7110
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.9116244316101074,
+      "learning_rate": 4.874597733861106e-05,
+      "loss": 0.1424,
+      "step": 7115
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 1.5253331661224365,
+      "learning_rate": 4.8742303530748125e-05,
+      "loss": 0.1492,
+      "step": 7120
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.5958783030509949,
+      "learning_rate": 4.8738624488177234e-05,
+      "loss": 0.1277,
+      "step": 7125
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.849671483039856,
+      "learning_rate": 4.873494021170953e-05,
+      "loss": 0.1322,
+      "step": 7130
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 1.0361336469650269,
+      "learning_rate": 4.8731250702157335e-05,
+      "loss": 0.1251,
+      "step": 7135
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.8894281983375549,
+      "learning_rate": 4.8727555960334115e-05,
+      "loss": 0.1383,
+      "step": 7140
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 6.309716701507568,
+      "learning_rate": 4.872385598705448e-05,
+      "loss": 0.1354,
+      "step": 7145
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 5.492419242858887,
+      "learning_rate": 4.8720150783134196e-05,
+      "loss": 0.1405,
+      "step": 7150
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.7048467397689819,
+      "learning_rate": 4.871644034939019e-05,
+      "loss": 0.1481,
+      "step": 7155
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.991845965385437,
+      "learning_rate": 4.871272468664055e-05,
+      "loss": 0.1327,
+      "step": 7160
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.9893251657485962,
+      "learning_rate": 4.870900379570449e-05,
+      "loss": 0.1381,
+      "step": 7165
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 2.8745369911193848,
+      "learning_rate": 4.870527767740241e-05,
+      "loss": 0.1506,
+      "step": 7170
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.640715479850769,
+      "learning_rate": 4.870154633255583e-05,
+      "loss": 0.1427,
+      "step": 7175
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.7568126320838928,
+      "learning_rate": 4.869780976198744e-05,
+      "loss": 0.1289,
+      "step": 7180
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 1.0072859525680542,
+      "learning_rate": 4.8694067966521095e-05,
+      "loss": 0.1313,
+      "step": 7185
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 1.1212821006774902,
+      "learning_rate": 4.869032094698177e-05,
+      "loss": 0.1554,
+      "step": 7190
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.5119507312774658,
+      "learning_rate": 4.8686568704195614e-05,
+      "loss": 0.1311,
+      "step": 7195
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 4.779508590698242,
+      "learning_rate": 4.8682811238989923e-05,
+      "loss": 0.151,
+      "step": 7200
+    },
+    {
+      "epoch": 2.71,
+      "eval_loss": 0.1381547600030899,
+      "eval_runtime": 340.2253,
+      "eval_samples_per_second": 66.988,
+      "eval_steps_per_second": 8.374,
+      "eval_wer": 0.10548813778042486,
+      "step": 7200
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 2.002885580062866,
+      "learning_rate": 4.867904855219315e-05,
+      "loss": 0.1412,
+      "step": 7205
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 1.3475871086120605,
+      "learning_rate": 4.867528064463488e-05,
+      "loss": 0.1302,
+      "step": 7210
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.8996849060058594,
+      "learning_rate": 4.8671507517145865e-05,
+      "loss": 0.1472,
+      "step": 7215
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.8186680674552917,
+      "learning_rate": 4.8667729170558006e-05,
+      "loss": 0.147,
+      "step": 7220
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 1.4446626901626587,
+      "learning_rate": 4.8663945605704367e-05,
+      "loss": 0.1394,
+      "step": 7225
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 1.8337488174438477,
+      "learning_rate": 4.866015682341913e-05,
+      "loss": 0.1487,
+      "step": 7230
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 1.0524581670761108,
+      "learning_rate": 4.865636282453766e-05,
+      "loss": 0.125,
+      "step": 7235
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 1.0678493976593018,
+      "learning_rate": 4.8652563609896446e-05,
+      "loss": 0.1363,
+      "step": 7240
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.8506509065628052,
+      "learning_rate": 4.864875918033315e-05,
+      "loss": 0.154,
+      "step": 7245
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.8889586329460144,
+      "learning_rate": 4.864494953668657e-05,
+      "loss": 0.1354,
+      "step": 7250
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 5.761272430419922,
+      "learning_rate": 4.864113467979666e-05,
+      "loss": 0.1354,
+      "step": 7255
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.7074782848358154,
+      "learning_rate": 4.8637314610504514e-05,
+      "loss": 0.1322,
+      "step": 7260
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.6419563293457031,
+      "learning_rate": 4.863348932965238e-05,
+      "loss": 0.1219,
+      "step": 7265
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 1.207193374633789,
+      "learning_rate": 4.862965883808367e-05,
+      "loss": 0.1368,
+      "step": 7270
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 4.059045791625977,
+      "learning_rate": 4.862582313664292e-05,
+      "loss": 0.1273,
+      "step": 7275
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 1.8989766836166382,
+      "learning_rate": 4.8621982226175836e-05,
+      "loss": 0.1457,
+      "step": 7280
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 21.187536239624023,
+      "learning_rate": 4.8618136107529254e-05,
+      "loss": 0.1434,
+      "step": 7285
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.931460976600647,
+      "learning_rate": 4.8614284781551164e-05,
+      "loss": 0.1215,
+      "step": 7290
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.9682580828666687,
+      "learning_rate": 4.861042824909072e-05,
+      "loss": 0.1348,
+      "step": 7295
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.9349156022071838,
+      "learning_rate": 4.8606566510998196e-05,
+      "loss": 0.1321,
+      "step": 7300
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 1.9747711420059204,
+      "learning_rate": 4.860269956812504e-05,
+      "loss": 0.1255,
+      "step": 7305
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.8634727597236633,
+      "learning_rate": 4.859882742132383e-05,
+      "loss": 0.1513,
+      "step": 7310
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.6778947114944458,
+      "learning_rate": 4.859495007144831e-05,
+      "loss": 0.1466,
+      "step": 7315
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 1.1910842657089233,
+      "learning_rate": 4.8591067519353345e-05,
+      "loss": 0.1291,
+      "step": 7320
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 9.965970993041992,
+      "learning_rate": 4.858717976589496e-05,
+      "loss": 0.1387,
+      "step": 7325
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 2.398820638656616,
+      "learning_rate": 4.858328681193034e-05,
+      "loss": 0.1962,
+      "step": 7330
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 1.043782114982605,
+      "learning_rate": 4.857938865831779e-05,
+      "loss": 0.1288,
+      "step": 7335
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.5726656317710876,
+      "learning_rate": 4.857548530591679e-05,
+      "loss": 0.1448,
+      "step": 7340
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.8872296214103699,
+      "learning_rate": 4.857157675558794e-05,
+      "loss": 0.1365,
+      "step": 7345
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.688933253288269,
+      "learning_rate": 4.856766300819301e-05,
+      "loss": 0.1512,
+      "step": 7350
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.7956893444061279,
+      "learning_rate": 4.856374406459489e-05,
+      "loss": 0.1448,
+      "step": 7355
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 1.1272779703140259,
+      "learning_rate": 4.855981992565764e-05,
+      "loss": 0.1499,
+      "step": 7360
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.7511281967163086,
+      "learning_rate": 4.855589059224645e-05,
+      "loss": 0.1277,
+      "step": 7365
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 1.4745033979415894,
+      "learning_rate": 4.855195606522767e-05,
+      "loss": 0.1454,
+      "step": 7370
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.5665118098258972,
+      "learning_rate": 4.854801634546876e-05,
+      "loss": 0.1366,
+      "step": 7375
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.9047695398330688,
+      "learning_rate": 4.854407143383838e-05,
+      "loss": 0.1515,
+      "step": 7380
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 8.252267837524414,
+      "learning_rate": 4.854012133120629e-05,
+      "loss": 0.1414,
+      "step": 7385
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 1.70575749874115,
+      "learning_rate": 4.853616603844341e-05,
+      "loss": 0.1415,
+      "step": 7390
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.8045769929885864,
+      "learning_rate": 4.85322055564218e-05,
+      "loss": 0.2336,
+      "step": 7395
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 1.5406668186187744,
+      "learning_rate": 4.852823988601468e-05,
+      "loss": 0.1554,
+      "step": 7400
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.825659453868866,
+      "learning_rate": 4.852426902809639e-05,
+      "loss": 0.1402,
+      "step": 7405
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.8376110792160034,
+      "learning_rate": 4.852029298354244e-05,
+      "loss": 0.1356,
+      "step": 7410
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.6655499935150146,
+      "learning_rate": 4.851631175322945e-05,
+      "loss": 0.1412,
+      "step": 7415
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.6092177629470825,
+      "learning_rate": 4.851232533803521e-05,
+      "loss": 0.1261,
+      "step": 7420
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.5395492315292358,
+      "learning_rate": 4.850833373883865e-05,
+      "loss": 0.1402,
+      "step": 7425
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.6742095351219177,
+      "learning_rate": 4.8504336956519844e-05,
+      "loss": 0.1304,
+      "step": 7430
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.5698046088218689,
+      "learning_rate": 4.8500334991959987e-05,
+      "loss": 0.1416,
+      "step": 7435
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.6993170380592346,
+      "learning_rate": 4.849632784604144e-05,
+      "loss": 0.1501,
+      "step": 7440
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.7156914472579956,
+      "learning_rate": 4.849231551964771e-05,
+      "loss": 0.1475,
+      "step": 7445
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.885330080986023,
+      "learning_rate": 4.848829801366342e-05,
+      "loss": 0.1507,
+      "step": 7450
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.9100447297096252,
+      "learning_rate": 4.848427532897437e-05,
+      "loss": 0.1827,
+      "step": 7455
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.7941574454307556,
+      "learning_rate": 4.8480247466467453e-05,
+      "loss": 0.1427,
+      "step": 7460
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.8409976363182068,
+      "learning_rate": 4.847621442703076e-05,
+      "loss": 0.1536,
+      "step": 7465
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 3.9116220474243164,
+      "learning_rate": 4.847217621155348e-05,
+      "loss": 0.1485,
+      "step": 7470
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.6189751625061035,
+      "learning_rate": 4.846813282092597e-05,
+      "loss": 0.1773,
+      "step": 7475
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.9027923941612244,
+      "learning_rate": 4.84640842560397e-05,
+      "loss": 0.1412,
+      "step": 7480
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 1.0196582078933716,
+      "learning_rate": 4.8460030517787326e-05,
+      "loss": 0.1375,
+      "step": 7485
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.9911620616912842,
+      "learning_rate": 4.84559716070626e-05,
+      "loss": 0.1291,
+      "step": 7490
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 1.1531399488449097,
+      "learning_rate": 4.845190752476043e-05,
+      "loss": 0.1423,
+      "step": 7495
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.5511499643325806,
+      "learning_rate": 4.844783827177687e-05,
+      "loss": 0.1257,
+      "step": 7500
+    },
+    {
+      "epoch": 2.82,
+      "eval_loss": 0.13953574001789093,
+      "eval_runtime": 339.8549,
+      "eval_samples_per_second": 67.061,
+      "eval_steps_per_second": 8.383,
+      "eval_wer": 0.1091320064853418,
+      "step": 7500
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.7815948724746704,
+      "learning_rate": 4.844376384900911e-05,
+      "loss": 0.1662,
+      "step": 7505
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.8163542747497559,
+      "learning_rate": 4.843968425735548e-05,
+      "loss": 0.1357,
+      "step": 7510
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.8427860140800476,
+      "learning_rate": 4.8435599497715453e-05,
+      "loss": 0.1267,
+      "step": 7515
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 1.1204466819763184,
+      "learning_rate": 4.843150957098962e-05,
+      "loss": 0.1371,
+      "step": 7520
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 1.0005865097045898,
+      "learning_rate": 4.8427414478079756e-05,
+      "loss": 0.1367,
+      "step": 7525
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.9354833960533142,
+      "learning_rate": 4.842331421988873e-05,
+      "loss": 0.1521,
+      "step": 7530
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 1.5336062908172607,
+      "learning_rate": 4.8419208797320564e-05,
+      "loss": 0.1396,
+      "step": 7535
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 1.329808235168457,
+      "learning_rate": 4.841509821128043e-05,
+      "loss": 0.124,
+      "step": 7540
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.7218471765518188,
+      "learning_rate": 4.8410982462674636e-05,
+      "loss": 0.1565,
+      "step": 7545
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 6.445298671722412,
+      "learning_rate": 4.840686155241062e-05,
+      "loss": 0.1757,
+      "step": 7550
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 1.429038405418396,
+      "learning_rate": 4.840273548139696e-05,
+      "loss": 0.1657,
+      "step": 7555
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.8010733723640442,
+      "learning_rate": 4.839860425054336e-05,
+      "loss": 0.1429,
+      "step": 7560
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.930989146232605,
+      "learning_rate": 4.83944678607607e-05,
+      "loss": 0.1307,
+      "step": 7565
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 2.590219497680664,
+      "learning_rate": 4.839032631296094e-05,
+      "loss": 0.1422,
+      "step": 7570
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.8220692873001099,
+      "learning_rate": 4.838617960805725e-05,
+      "loss": 0.1498,
+      "step": 7575
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 1.3354254961013794,
+      "learning_rate": 4.838202774696386e-05,
+      "loss": 0.1554,
+      "step": 7580
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.7347155809402466,
+      "learning_rate": 4.8377870730596194e-05,
+      "loss": 0.1218,
+      "step": 7585
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 1.7706823348999023,
+      "learning_rate": 4.8373708559870776e-05,
+      "loss": 0.1478,
+      "step": 7590
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.8192682266235352,
+      "learning_rate": 4.83695412357053e-05,
+      "loss": 0.1477,
+      "step": 7595
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.5958541631698608,
+      "learning_rate": 4.836536875901857e-05,
+      "loss": 0.1551,
+      "step": 7600
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.8686918020248413,
+      "learning_rate": 4.836119113073053e-05,
+      "loss": 0.1521,
+      "step": 7605
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.7984471917152405,
+      "learning_rate": 4.835700835176228e-05,
+      "loss": 0.1275,
+      "step": 7610
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.8589127659797668,
+      "learning_rate": 4.835282042303601e-05,
+      "loss": 0.1498,
+      "step": 7615
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 1.2806264162063599,
+      "learning_rate": 4.834862734547511e-05,
+      "loss": 0.1429,
+      "step": 7620
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 1.8588300943374634,
+      "learning_rate": 4.834442912000405e-05,
+      "loss": 0.144,
+      "step": 7625
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 1.1410589218139648,
+      "learning_rate": 4.834022574754846e-05,
+      "loss": 0.1722,
+      "step": 7630
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 1.192315697669983,
+      "learning_rate": 4.83360172290351e-05,
+      "loss": 0.1624,
+      "step": 7635
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.6814581751823425,
+      "learning_rate": 4.833180356539186e-05,
+      "loss": 0.1553,
+      "step": 7640
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.5996586680412292,
+      "learning_rate": 4.832758475754778e-05,
+      "loss": 0.145,
+      "step": 7645
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 1.142007827758789,
+      "learning_rate": 4.832336080643303e-05,
+      "loss": 0.1301,
+      "step": 7650
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.8181615471839905,
+      "learning_rate": 4.8319131712978885e-05,
+      "loss": 0.1474,
+      "step": 7655
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.5453782081604004,
+      "learning_rate": 4.831489747811779e-05,
+      "loss": 0.1432,
+      "step": 7660
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 1.1381460428237915,
+      "learning_rate": 4.8310658102783304e-05,
+      "loss": 0.1472,
+      "step": 7665
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.6505977511405945,
+      "learning_rate": 4.830641358791014e-05,
+      "loss": 0.1513,
+      "step": 7670
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.669451117515564,
+      "learning_rate": 4.830216393443411e-05,
+      "loss": 0.1356,
+      "step": 7675
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 1.0270744562149048,
+      "learning_rate": 4.8297909143292186e-05,
+      "loss": 0.1568,
+      "step": 7680
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.7490842342376709,
+      "learning_rate": 4.829364921542246e-05,
+      "loss": 0.1432,
+      "step": 7685
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.7833766937255859,
+      "learning_rate": 4.828938415176418e-05,
+      "loss": 0.1519,
+      "step": 7690
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 1.2223707437515259,
+      "learning_rate": 4.82851139532577e-05,
+      "loss": 0.1342,
+      "step": 7695
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.9714386463165283,
+      "learning_rate": 4.8280838620844506e-05,
+      "loss": 0.1509,
+      "step": 7700
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.7047361731529236,
+      "learning_rate": 4.827655815546722e-05,
+      "loss": 0.1548,
+      "step": 7705
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.6078805923461914,
+      "learning_rate": 4.827227255806962e-05,
+      "loss": 0.1426,
+      "step": 7710
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 1.0403228998184204,
+      "learning_rate": 4.826798182959658e-05,
+      "loss": 0.1269,
+      "step": 7715
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.6051403284072876,
+      "learning_rate": 4.826368597099412e-05,
+      "loss": 0.1429,
+      "step": 7720
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 1.608965277671814,
+      "learning_rate": 4.82593849832094e-05,
+      "loss": 0.1172,
+      "step": 7725
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.7561035752296448,
+      "learning_rate": 4.82550788671907e-05,
+      "loss": 0.1427,
+      "step": 7730
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.726631760597229,
+      "learning_rate": 4.8250767623887435e-05,
+      "loss": 0.1404,
+      "step": 7735
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.5835649371147156,
+      "learning_rate": 4.8246451254250145e-05,
+      "loss": 0.1324,
+      "step": 7740
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.4770151376724243,
+      "learning_rate": 4.8242129759230504e-05,
+      "loss": 0.1354,
+      "step": 7745
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.5990408658981323,
+      "learning_rate": 4.823780313978131e-05,
+      "loss": 0.1386,
+      "step": 7750
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 1.0062446594238281,
+      "learning_rate": 4.823347139685652e-05,
+      "loss": 0.1327,
+      "step": 7755
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 1.0581135749816895,
+      "learning_rate": 4.8229134531411166e-05,
+      "loss": 0.1462,
+      "step": 7760
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.6423236131668091,
+      "learning_rate": 4.822479254440147e-05,
+      "loss": 0.1549,
+      "step": 7765
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.5213544368743896,
+      "learning_rate": 4.8220445436784736e-05,
+      "loss": 0.1794,
+      "step": 7770
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.6475617289543152,
+      "learning_rate": 4.821609320951942e-05,
+      "loss": 0.1428,
+      "step": 7775
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.8852936029434204,
+      "learning_rate": 4.82117358635651e-05,
+      "loss": 0.1717,
+      "step": 7780
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 1.2626248598098755,
+      "learning_rate": 4.8207373399882494e-05,
+      "loss": 0.1448,
+      "step": 7785
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.6716387271881104,
+      "learning_rate": 4.820300581943342e-05,
+      "loss": 0.1369,
+      "step": 7790
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.8092980980873108,
+      "learning_rate": 4.819863312318087e-05,
+      "loss": 0.1366,
+      "step": 7795
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.8222695589065552,
+      "learning_rate": 4.819425531208891e-05,
+      "loss": 0.137,
+      "step": 7800
+    },
+    {
+      "epoch": 2.94,
+      "eval_loss": 0.13496482372283936,
+      "eval_runtime": 345.8806,
+      "eval_samples_per_second": 65.893,
+      "eval_steps_per_second": 8.237,
+      "eval_wer": 0.10381923764145325,
+      "step": 7800
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.7255913615226746,
+      "learning_rate": 4.818987238712278e-05,
+      "loss": 0.1493,
+      "step": 7805
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 1.5969575643539429,
+      "learning_rate": 4.818548434924881e-05,
+      "loss": 0.1505,
+      "step": 7810
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.6396997570991516,
+      "learning_rate": 4.818109119943449e-05,
+      "loss": 0.168,
+      "step": 7815
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.7796175479888916,
+      "learning_rate": 4.8176692938648424e-05,
+      "loss": 0.1379,
+      "step": 7820
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 1.3160333633422852,
+      "learning_rate": 4.8172289567860344e-05,
+      "loss": 0.1337,
+      "step": 7825
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 1.236918568611145,
+      "learning_rate": 4.816788108804108e-05,
+      "loss": 0.1342,
+      "step": 7830
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.9104110598564148,
+      "learning_rate": 4.816346750016264e-05,
+      "loss": 0.1316,
+      "step": 7835
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.4841606318950653,
+      "learning_rate": 4.815904880519813e-05,
+      "loss": 0.125,
+      "step": 7840
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 1.337124228477478,
+      "learning_rate": 4.8154625004121774e-05,
+      "loss": 0.1351,
+      "step": 7845
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.5650362968444824,
+      "learning_rate": 4.815019609790894e-05,
+      "loss": 0.1341,
+      "step": 7850
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.679772138595581,
+      "learning_rate": 4.814576208753612e-05,
+      "loss": 0.1497,
+      "step": 7855
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.5833603739738464,
+      "learning_rate": 4.8141322973980905e-05,
+      "loss": 0.1273,
+      "step": 7860
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.7591491937637329,
+      "learning_rate": 4.813687875822206e-05,
+      "loss": 0.1458,
+      "step": 7865
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.6721802353858948,
+      "learning_rate": 4.8132429441239425e-05,
+      "loss": 0.1346,
+      "step": 7870
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.8154709339141846,
+      "learning_rate": 4.812797502401399e-05,
+      "loss": 0.1575,
+      "step": 7875
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.6731926202774048,
+      "learning_rate": 4.812351550752787e-05,
+      "loss": 0.1427,
+      "step": 7880
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.9061952829360962,
+      "learning_rate": 4.81190508927643e-05,
+      "loss": 0.1485,
+      "step": 7885
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.4754640758037567,
+      "learning_rate": 4.811458118070764e-05,
+      "loss": 0.1386,
+      "step": 7890
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 1.6076382398605347,
+      "learning_rate": 4.8110106372343366e-05,
+      "loss": 0.1545,
+      "step": 7895
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 1.030163288116455,
+      "learning_rate": 4.81056264686581e-05,
+      "loss": 0.1553,
+      "step": 7900
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.649213433265686,
+      "learning_rate": 4.810114147063956e-05,
+      "loss": 0.1266,
+      "step": 7905
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.5791982412338257,
+      "learning_rate": 4.809665137927659e-05,
+      "loss": 0.1363,
+      "step": 7910
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.8975016474723816,
+      "learning_rate": 4.809215619555919e-05,
+      "loss": 0.1424,
+      "step": 7915
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.6749100685119629,
+      "learning_rate": 4.8087655920478436e-05,
+      "loss": 0.1353,
+      "step": 7920
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 1.0060081481933594,
+      "learning_rate": 4.808315055502657e-05,
+      "loss": 0.1507,
+      "step": 7925
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 1.5669864416122437,
+      "learning_rate": 4.8078640100196923e-05,
+      "loss": 0.1375,
+      "step": 7930
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 0.8764144778251648,
+      "learning_rate": 4.807412455698396e-05,
+      "loss": 0.1364,
+      "step": 7935
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 0.8155016899108887,
+      "learning_rate": 4.806960392638328e-05,
+      "loss": 0.1399,
+      "step": 7940
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 0.7208448052406311,
+      "learning_rate": 4.806507820939158e-05,
+      "loss": 0.1229,
+      "step": 7945
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 0.7936753034591675,
+      "learning_rate": 4.80605474070067e-05,
+      "loss": 0.152,
+      "step": 7950
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.814859926700592,
+      "learning_rate": 4.8056011520227584e-05,
+      "loss": 0.1489,
+      "step": 7955
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.8821942806243896,
+      "learning_rate": 4.805147055005431e-05,
+      "loss": 0.1173,
+      "step": 7960
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.6818812489509583,
+      "learning_rate": 4.804692449748808e-05,
+      "loss": 0.1383,
+      "step": 7965
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.7268828749656677,
+      "learning_rate": 4.80423733635312e-05,
+      "loss": 0.1281,
+      "step": 7970
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.6024926900863647,
+      "learning_rate": 4.8037817149187096e-05,
+      "loss": 0.1149,
+      "step": 7975
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.8752952814102173,
+      "learning_rate": 4.803325585546034e-05,
+      "loss": 0.1348,
+      "step": 7980
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 1.1683228015899658,
+      "learning_rate": 4.80286894833566e-05,
+      "loss": 0.1293,
+      "step": 7985
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.9751905798912048,
+      "learning_rate": 4.802411803388267e-05,
+      "loss": 0.1239,
+      "step": 7990
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.8060181140899658,
+      "learning_rate": 4.801954150804646e-05,
+      "loss": 0.1338,
+      "step": 7995
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 1.2890589237213135,
+      "learning_rate": 4.801495990685701e-05,
+      "loss": 0.1363,
+      "step": 8000
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.8100897669792175,
+      "learning_rate": 4.801037323132447e-05,
+      "loss": 0.1169,
+      "step": 8005
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 1.07380211353302,
+      "learning_rate": 4.800578148246011e-05,
+      "loss": 0.1303,
+      "step": 8010
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 1.572828769683838,
+      "learning_rate": 4.8001184661276325e-05,
+      "loss": 0.1151,
+      "step": 8015
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 0.6850970983505249,
+      "learning_rate": 4.799658276878662e-05,
+      "loss": 0.1305,
+      "step": 8020
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 0.9445905089378357,
+      "learning_rate": 4.799197580600562e-05,
+      "loss": 0.1225,
+      "step": 8025
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 0.5026190280914307,
+      "learning_rate": 4.798736377394907e-05,
+      "loss": 0.1153,
+      "step": 8030
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.5225924253463745,
+      "learning_rate": 4.798274667363383e-05,
+      "loss": 0.1212,
+      "step": 8035
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 1.3120522499084473,
+      "learning_rate": 4.797812450607787e-05,
+      "loss": 0.1317,
+      "step": 8040
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.6215925216674805,
+      "learning_rate": 4.797349727230031e-05,
+      "loss": 0.1263,
+      "step": 8045
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 1.8067142963409424,
+      "learning_rate": 4.796886497332135e-05,
+      "loss": 0.125,
+      "step": 8050
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.806132972240448,
+      "learning_rate": 4.7964227610162316e-05,
+      "loss": 0.1327,
+      "step": 8055
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 0.8761575222015381,
+      "learning_rate": 4.7959585183845655e-05,
+      "loss": 0.1142,
+      "step": 8060
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 0.6686119437217712,
+      "learning_rate": 4.795493769539494e-05,
+      "loss": 0.1078,
+      "step": 8065
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 0.6672380566596985,
+      "learning_rate": 4.795028514583485e-05,
+      "loss": 0.116,
+      "step": 8070
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 0.7437089085578918,
+      "learning_rate": 4.7945627536191166e-05,
+      "loss": 0.1132,
+      "step": 8075
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 0.5553317666053772,
+      "learning_rate": 4.7940964867490814e-05,
+      "loss": 0.1372,
+      "step": 8080
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 0.7084664702415466,
+      "learning_rate": 4.793629714076181e-05,
+      "loss": 0.1267,
+      "step": 8085
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.8994629383087158,
+      "learning_rate": 4.79316243570333e-05,
+      "loss": 0.1408,
+      "step": 8090
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.4514738619327545,
+      "learning_rate": 4.792694651733553e-05,
+      "loss": 0.1424,
+      "step": 8095
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 1.3485304117202759,
+      "learning_rate": 4.7922263622699895e-05,
+      "loss": 0.1289,
+      "step": 8100
+    },
+    {
+      "epoch": 3.05,
+      "eval_loss": 0.13453370332717896,
+      "eval_runtime": 354.4436,
+      "eval_samples_per_second": 64.301,
+      "eval_steps_per_second": 8.038,
+      "eval_wer": 0.10386059823969294,
+      "step": 8100
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.6463409662246704,
+      "learning_rate": 4.7917575674158864e-05,
+      "loss": 0.1793,
+      "step": 8105
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.7853609919548035,
+      "learning_rate": 4.791288267274604e-05,
+      "loss": 0.1416,
+      "step": 8110
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 0.8179067373275757,
+      "learning_rate": 4.7908184619496126e-05,
+      "loss": 0.1401,
+      "step": 8115
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 0.8170843720436096,
+      "learning_rate": 4.790348151544497e-05,
+      "loss": 0.122,
+      "step": 8120
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 0.7175580263137817,
+      "learning_rate": 4.78987733616295e-05,
+      "loss": 0.1301,
+      "step": 8125
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 1.9520536661148071,
+      "learning_rate": 4.789406015908776e-05,
+      "loss": 0.1195,
+      "step": 8130
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 0.7440107464790344,
+      "learning_rate": 4.789028596266976e-05,
+      "loss": 0.1392,
+      "step": 8135
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 0.6097704768180847,
+      "learning_rate": 4.7885563675040224e-05,
+      "loss": 0.1227,
+      "step": 8140
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 0.7710959911346436,
+      "learning_rate": 4.7880836341596904e-05,
+      "loss": 0.1234,
+      "step": 8145
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 0.6513006091117859,
+      "learning_rate": 4.787610396338208e-05,
+      "loss": 0.1224,
+      "step": 8150
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 1.1170635223388672,
+      "learning_rate": 4.787136654143916e-05,
+      "loss": 0.1237,
+      "step": 8155
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 0.37960726022720337,
+      "learning_rate": 4.786662407681264e-05,
+      "loss": 0.1322,
+      "step": 8160
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 0.5684589743614197,
+      "learning_rate": 4.786187657054815e-05,
+      "loss": 0.1269,
+      "step": 8165
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.8693246841430664,
+      "learning_rate": 4.785712402369241e-05,
+      "loss": 0.112,
+      "step": 8170
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.3451720178127289,
+      "learning_rate": 4.785236643729327e-05,
+      "loss": 0.1013,
+      "step": 8175
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.6501955389976501,
+      "learning_rate": 4.7847603812399675e-05,
+      "loss": 0.1286,
+      "step": 8180
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.5982517004013062,
+      "learning_rate": 4.784283615006171e-05,
+      "loss": 0.1237,
+      "step": 8185
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.48952797055244446,
+      "learning_rate": 4.7838063451330526e-05,
+      "loss": 0.1235,
+      "step": 8190
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 0.8246931433677673,
+      "learning_rate": 4.783328571725843e-05,
+      "loss": 0.1117,
+      "step": 8195
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 1.2162154912948608,
+      "learning_rate": 4.782850294889879e-05,
+      "loss": 0.1286,
+      "step": 8200
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 0.9021078944206238,
+      "learning_rate": 4.7823715147306147e-05,
+      "loss": 0.1517,
+      "step": 8205
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 0.8552690148353577,
+      "learning_rate": 4.781892231353609e-05,
+      "loss": 0.1324,
+      "step": 8210
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 0.762871265411377,
+      "learning_rate": 4.7814124448645345e-05,
+      "loss": 0.1271,
+      "step": 8215
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 1.538124918937683,
+      "learning_rate": 4.7809321553691764e-05,
+      "loss": 0.1042,
+      "step": 8220
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.7550507187843323,
+      "learning_rate": 4.7804513629734276e-05,
+      "loss": 0.133,
+      "step": 8225
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.80412358045578,
+      "learning_rate": 4.7799700677832935e-05,
+      "loss": 0.1492,
+      "step": 8230
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.8660932779312134,
+      "learning_rate": 4.77948826990489e-05,
+      "loss": 0.1195,
+      "step": 8235
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.5514984726905823,
+      "learning_rate": 4.779005969444446e-05,
+      "loss": 0.1099,
+      "step": 8240
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.5892981290817261,
+      "learning_rate": 4.7785231665082954e-05,
+      "loss": 0.1432,
+      "step": 8245
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 0.8316653370857239,
+      "learning_rate": 4.7780398612028896e-05,
+      "loss": 0.1269,
+      "step": 8250
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 2.3927345275878906,
+      "learning_rate": 4.777556053634788e-05,
+      "loss": 0.1364,
+      "step": 8255
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 0.578448474407196,
+      "learning_rate": 4.777071743910658e-05,
+      "loss": 0.1251,
+      "step": 8260
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 0.8155702948570251,
+      "learning_rate": 4.7765869321372836e-05,
+      "loss": 0.1168,
+      "step": 8265
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 0.6353880763053894,
+      "learning_rate": 4.7761016184215534e-05,
+      "loss": 0.1093,
+      "step": 8270
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.6077370047569275,
+      "learning_rate": 4.7756158028704714e-05,
+      "loss": 0.1222,
+      "step": 8275
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.6857184171676636,
+      "learning_rate": 4.775129485591149e-05,
+      "loss": 0.1234,
+      "step": 8280
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.6472964882850647,
+      "learning_rate": 4.77464266669081e-05,
+      "loss": 0.1216,
+      "step": 8285
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 1.013681411743164,
+      "learning_rate": 4.774155346276789e-05,
+      "loss": 0.1026,
+      "step": 8290
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.648395299911499,
+      "learning_rate": 4.773667524456529e-05,
+      "loss": 0.1221,
+      "step": 8295
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 2.209243059158325,
+      "learning_rate": 4.7731792013375864e-05,
+      "loss": 0.1368,
+      "step": 8300
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 0.7366249561309814,
+      "learning_rate": 4.7726903770276265e-05,
+      "loss": 0.1378,
+      "step": 8305
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 0.6801765561103821,
+      "learning_rate": 4.772201051634426e-05,
+      "loss": 0.1283,
+      "step": 8310
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 1.3412216901779175,
+      "learning_rate": 4.77171122526587e-05,
+      "loss": 0.1235,
+      "step": 8315
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 0.7595944404602051,
+      "learning_rate": 4.771220898029956e-05,
+      "loss": 0.1253,
+      "step": 8320
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.5139328837394714,
+      "learning_rate": 4.770730070034791e-05,
+      "loss": 0.1184,
+      "step": 8325
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.549645185470581,
+      "learning_rate": 4.770238741388595e-05,
+      "loss": 0.1181,
+      "step": 8330
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.8553074598312378,
+      "learning_rate": 4.769746912199694e-05,
+      "loss": 0.1328,
+      "step": 8335
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.5197417140007019,
+      "learning_rate": 4.769254582576528e-05,
+      "loss": 0.1299,
+      "step": 8340
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.6674187779426575,
+      "learning_rate": 4.768761752627644e-05,
+      "loss": 0.1221,
+      "step": 8345
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.8692063093185425,
+      "learning_rate": 4.768268422461704e-05,
+      "loss": 0.1497,
+      "step": 8350
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 0.736126184463501,
+      "learning_rate": 4.767774592187475e-05,
+      "loss": 0.1324,
+      "step": 8355
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 0.8299170732498169,
+      "learning_rate": 4.767280261913839e-05,
+      "loss": 0.1939,
+      "step": 8360
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 0.6357385516166687,
+      "learning_rate": 4.766785431749784e-05,
+      "loss": 0.1207,
+      "step": 8365
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 4.6167073249816895,
+      "learning_rate": 4.7662901018044116e-05,
+      "loss": 0.1369,
+      "step": 8370
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 0.9137371778488159,
+      "learning_rate": 4.765794272186931e-05,
+      "loss": 0.1194,
+      "step": 8375
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 0.6972301006317139,
+      "learning_rate": 4.765297943006665e-05,
+      "loss": 0.1345,
+      "step": 8380
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 0.8508134484291077,
+      "learning_rate": 4.764801114373042e-05,
+      "loss": 0.1496,
+      "step": 8385
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 1.2958019971847534,
+      "learning_rate": 4.764303786395604e-05,
+      "loss": 0.1365,
+      "step": 8390
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 0.7231436967849731,
+      "learning_rate": 4.763805959184002e-05,
+      "loss": 0.1162,
+      "step": 8395
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 0.4796886146068573,
+      "learning_rate": 4.763307632847998e-05,
+      "loss": 0.113,
+      "step": 8400
+    },
+    {
+      "epoch": 3.16,
+      "eval_loss": 0.13149482011795044,
+      "eval_runtime": 353.9857,
+      "eval_samples_per_second": 64.384,
+      "eval_steps_per_second": 8.048,
+      "eval_wer": 0.10075648534180398,
+      "step": 8400
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.6776131987571716,
+      "learning_rate": 4.7628088074974606e-05,
+      "loss": 0.1211,
+      "step": 8405
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.4905591607093811,
+      "learning_rate": 4.7623094832423735e-05,
+      "loss": 0.1259,
+      "step": 8410
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 2.4199767112731934,
+      "learning_rate": 4.761809660192825e-05,
+      "loss": 0.1231,
+      "step": 8415
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.5603888034820557,
+      "learning_rate": 4.761309338459019e-05,
+      "loss": 0.1248,
+      "step": 8420
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.6131569147109985,
+      "learning_rate": 4.760808518151265e-05,
+      "loss": 0.1028,
+      "step": 8425
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.8877180814743042,
+      "learning_rate": 4.760307199379984e-05,
+      "loss": 0.1427,
+      "step": 8430
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 0.5086161494255066,
+      "learning_rate": 4.759805382255708e-05,
+      "loss": 0.1256,
+      "step": 8435
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 1.4496135711669922,
+      "learning_rate": 4.759303066889077e-05,
+      "loss": 0.1254,
+      "step": 8440
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 0.8230924606323242,
+      "learning_rate": 4.7588002533908405e-05,
+      "loss": 0.1075,
+      "step": 8445
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 0.7228221893310547,
+      "learning_rate": 4.7582969418718605e-05,
+      "loss": 0.1334,
+      "step": 8450
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 0.6956369876861572,
+      "learning_rate": 4.757793132443107e-05,
+      "loss": 0.1328,
+      "step": 8455
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.6291519999504089,
+      "learning_rate": 4.757288825215659e-05,
+      "loss": 0.1192,
+      "step": 8460
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.7695654630661011,
+      "learning_rate": 4.756784020300708e-05,
+      "loss": 0.1083,
+      "step": 8465
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.8787385821342468,
+      "learning_rate": 4.756278717809552e-05,
+      "loss": 0.1313,
+      "step": 8470
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.6089791059494019,
+      "learning_rate": 4.7557729178536e-05,
+      "loss": 0.1266,
+      "step": 8475
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 1.156903624534607,
+      "learning_rate": 4.755266620544372e-05,
+      "loss": 0.1187,
+      "step": 8480
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.6923069357872009,
+      "learning_rate": 4.754759825993496e-05,
+      "loss": 0.1401,
+      "step": 8485
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 2.4087345600128174,
+      "learning_rate": 4.75425253431271e-05,
+      "loss": 0.1274,
+      "step": 8490
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.9054912328720093,
+      "learning_rate": 4.753744745613862e-05,
+      "loss": 0.14,
+      "step": 8495
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.8866832852363586,
+      "learning_rate": 4.75323646000891e-05,
+      "loss": 0.1264,
+      "step": 8500
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.829301118850708,
+      "learning_rate": 4.752727677609919e-05,
+      "loss": 0.1161,
+      "step": 8505
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.4981716573238373,
+      "learning_rate": 4.752218398529068e-05,
+      "loss": 0.134,
+      "step": 8510
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 1.0041528940200806,
+      "learning_rate": 4.751708622878641e-05,
+      "loss": 0.1284,
+      "step": 8515
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.7959632873535156,
+      "learning_rate": 4.7511983507710344e-05,
+      "loss": 0.1263,
+      "step": 8520
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.3928009867668152,
+      "learning_rate": 4.750687582318752e-05,
+      "loss": 0.1063,
+      "step": 8525
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 4.577338218688965,
+      "learning_rate": 4.7501763176344106e-05,
+      "loss": 0.1439,
+      "step": 8530
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.6664185523986816,
+      "learning_rate": 4.749664556830731e-05,
+      "loss": 0.1261,
+      "step": 8535
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 0.7969637513160706,
+      "learning_rate": 4.749152300020548e-05,
+      "loss": 0.1187,
+      "step": 8540
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 0.6103954315185547,
+      "learning_rate": 4.7486395473168033e-05,
+      "loss": 0.1042,
+      "step": 8545
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 0.9539281725883484,
+      "learning_rate": 4.7481262988325495e-05,
+      "loss": 0.1256,
+      "step": 8550
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 0.7999578714370728,
+      "learning_rate": 4.747612554680948e-05,
+      "loss": 0.1243,
+      "step": 8555
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 0.9105613827705383,
+      "learning_rate": 4.747098314975267e-05,
+      "loss": 0.1407,
+      "step": 8560
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 0.5544403791427612,
+      "learning_rate": 4.746583579828889e-05,
+      "loss": 0.1383,
+      "step": 8565
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 0.9830417633056641,
+      "learning_rate": 4.746068349355301e-05,
+      "loss": 0.131,
+      "step": 8570
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 0.9632214307785034,
+      "learning_rate": 4.745552623668102e-05,
+      "loss": 0.13,
+      "step": 8575
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 0.6463819146156311,
+      "learning_rate": 4.745036402880999e-05,
+      "loss": 0.1319,
+      "step": 8580
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 0.8018662333488464,
+      "learning_rate": 4.744519687107809e-05,
+      "loss": 0.1337,
+      "step": 8585
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 0.8200687766075134,
+      "learning_rate": 4.744002476462456e-05,
+      "loss": 0.119,
+      "step": 8590
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 1.579991102218628,
+      "learning_rate": 4.743484771058977e-05,
+      "loss": 0.1416,
+      "step": 8595
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.734357476234436,
+      "learning_rate": 4.742966571011515e-05,
+      "loss": 0.1428,
+      "step": 8600
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.672691285610199,
+      "learning_rate": 4.742447876434321e-05,
+      "loss": 0.1165,
+      "step": 8605
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.5834132432937622,
+      "learning_rate": 4.741928687441759e-05,
+      "loss": 0.1074,
+      "step": 8610
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 1.23294198513031,
+      "learning_rate": 4.7414090041482995e-05,
+      "loss": 0.1182,
+      "step": 8615
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 0.733536422252655,
+      "learning_rate": 4.7408888266685227e-05,
+      "loss": 0.1305,
+      "step": 8620
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 0.7514683604240417,
+      "learning_rate": 4.740368155117116e-05,
+      "loss": 0.1399,
+      "step": 8625
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 0.6074724793434143,
+      "learning_rate": 4.7398469896088785e-05,
+      "loss": 0.121,
+      "step": 8630
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 1.0427335500717163,
+      "learning_rate": 4.739325330258716e-05,
+      "loss": 0.1331,
+      "step": 8635
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 0.9787009954452515,
+      "learning_rate": 4.738803177181644e-05,
+      "loss": 0.1199,
+      "step": 8640
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 2.3278040885925293,
+      "learning_rate": 4.7382805304927894e-05,
+      "loss": 0.1504,
+      "step": 8645
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.846390962600708,
+      "learning_rate": 4.737757390307382e-05,
+      "loss": 0.1425,
+      "step": 8650
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.5427407026290894,
+      "learning_rate": 4.7372337567407654e-05,
+      "loss": 0.1388,
+      "step": 8655
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.4931872487068176,
+      "learning_rate": 4.736709629908391e-05,
+      "loss": 0.1136,
+      "step": 8660
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.8178187608718872,
+      "learning_rate": 4.736185009925817e-05,
+      "loss": 0.1393,
+      "step": 8665
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 1.123936653137207,
+      "learning_rate": 4.735659896908713e-05,
+      "loss": 0.1485,
+      "step": 8670
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 0.9179394841194153,
+      "learning_rate": 4.7351342909728554e-05,
+      "loss": 0.1469,
+      "step": 8675
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 0.5245729088783264,
+      "learning_rate": 4.7346081922341303e-05,
+      "loss": 0.1157,
+      "step": 8680
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 0.5953699350357056,
+      "learning_rate": 4.734081600808531e-05,
+      "loss": 0.1104,
+      "step": 8685
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 1.0732849836349487,
+      "learning_rate": 4.733554516812162e-05,
+      "loss": 0.1177,
+      "step": 8690
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 0.9140703678131104,
+      "learning_rate": 4.7330269403612336e-05,
+      "loss": 0.1389,
+      "step": 8695
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 1.3813751935958862,
+      "learning_rate": 4.732498871572066e-05,
+      "loss": 0.1311,
+      "step": 8700
+    },
+    {
+      "epoch": 3.28,
+      "eval_loss": 0.12476861476898193,
+      "eval_runtime": 354.14,
+      "eval_samples_per_second": 64.356,
+      "eval_steps_per_second": 8.045,
+      "eval_wer": 0.09726978691019787,
+      "step": 8700
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 0.7783933877944946,
+      "learning_rate": 4.7319703105610895e-05,
+      "loss": 0.108,
+      "step": 8705
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 4.457291126251221,
+      "learning_rate": 4.7314412574448395e-05,
+      "loss": 0.1285,
+      "step": 8710
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 1.8070018291473389,
+      "learning_rate": 4.730911712339964e-05,
+      "loss": 0.132,
+      "step": 8715
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 0.5996901392936707,
+      "learning_rate": 4.730381675363214e-05,
+      "loss": 0.1313,
+      "step": 8720
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 0.6539852023124695,
+      "learning_rate": 4.7298511466314546e-05,
+      "loss": 0.1367,
+      "step": 8725
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 0.520638108253479,
+      "learning_rate": 4.729320126261656e-05,
+      "loss": 0.1063,
+      "step": 8730
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 0.8143773078918457,
+      "learning_rate": 4.728788614370898e-05,
+      "loss": 0.1173,
+      "step": 8735
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 0.9942452907562256,
+      "learning_rate": 4.728256611076368e-05,
+      "loss": 0.121,
+      "step": 8740
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 1.2685693502426147,
+      "learning_rate": 4.727724116495363e-05,
+      "loss": 0.1396,
+      "step": 8745
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 1.8607779741287231,
+      "learning_rate": 4.727191130745286e-05,
+      "loss": 0.1166,
+      "step": 8750
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.916776716709137,
+      "learning_rate": 4.726657653943652e-05,
+      "loss": 0.1358,
+      "step": 8755
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.6906163096427917,
+      "learning_rate": 4.7261236862080805e-05,
+      "loss": 0.1216,
+      "step": 8760
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.8550844192504883,
+      "learning_rate": 4.725589227656301e-05,
+      "loss": 0.1348,
+      "step": 8765
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.681460440158844,
+      "learning_rate": 4.7250542784061515e-05,
+      "loss": 0.1198,
+      "step": 8770
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 12.219022750854492,
+      "learning_rate": 4.7245188385755765e-05,
+      "loss": 0.1377,
+      "step": 8775
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 1.6511499881744385,
+      "learning_rate": 4.723982908282632e-05,
+      "loss": 0.178,
+      "step": 8780
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 1.3881632089614868,
+      "learning_rate": 4.7234464876454784e-05,
+      "loss": 0.1447,
+      "step": 8785
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 0.9571691155433655,
+      "learning_rate": 4.7229095767823865e-05,
+      "loss": 0.1247,
+      "step": 8790
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 1.4274736642837524,
+      "learning_rate": 4.722372175811735e-05,
+      "loss": 0.1413,
+      "step": 8795
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 0.7503607273101807,
+      "learning_rate": 4.721834284852008e-05,
+      "loss": 0.1287,
+      "step": 8800
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 0.5273829102516174,
+      "learning_rate": 4.721295904021802e-05,
+      "loss": 0.1205,
+      "step": 8805
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 1.2348753213882446,
+      "learning_rate": 4.720757033439818e-05,
+      "loss": 0.1261,
+      "step": 8810
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 4.048149108886719,
+      "learning_rate": 4.720217673224868e-05,
+      "loss": 0.129,
+      "step": 8815
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 1.0477153062820435,
+      "learning_rate": 4.719677823495868e-05,
+      "loss": 0.1086,
+      "step": 8820
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 1.6664701700210571,
+      "learning_rate": 4.719137484371846e-05,
+      "loss": 0.1209,
+      "step": 8825
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.531282365322113,
+      "learning_rate": 4.718596655971936e-05,
+      "loss": 0.1162,
+      "step": 8830
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.7920160293579102,
+      "learning_rate": 4.718055338415378e-05,
+      "loss": 0.1203,
+      "step": 8835
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 1.818257451057434,
+      "learning_rate": 4.7175135318215246e-05,
+      "loss": 0.1338,
+      "step": 8840
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 1.2404865026474,
+      "learning_rate": 4.716971236309831e-05,
+      "loss": 0.1362,
+      "step": 8845
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.8169355988502502,
+      "learning_rate": 4.7164284519998644e-05,
+      "loss": 0.1633,
+      "step": 8850
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 1.0071388483047485,
+      "learning_rate": 4.715885179011297e-05,
+      "loss": 0.1415,
+      "step": 8855
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 0.816893994808197,
+      "learning_rate": 4.7153414174639084e-05,
+      "loss": 0.1229,
+      "step": 8860
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 1.6251165866851807,
+      "learning_rate": 4.7147971674775906e-05,
+      "loss": 0.1134,
+      "step": 8865
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 0.8472614884376526,
+      "learning_rate": 4.7142524291723375e-05,
+      "loss": 0.1258,
+      "step": 8870
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 0.712583601474762,
+      "learning_rate": 4.713707202668255e-05,
+      "loss": 0.1632,
+      "step": 8875
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 1.3845807313919067,
+      "learning_rate": 4.713161488085552e-05,
+      "loss": 0.1355,
+      "step": 8880
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.9438744783401489,
+      "learning_rate": 4.712615285544551e-05,
+      "loss": 0.1327,
+      "step": 8885
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.7201677560806274,
+      "learning_rate": 4.712068595165675e-05,
+      "loss": 0.1541,
+      "step": 8890
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.7544935941696167,
+      "learning_rate": 4.711521417069462e-05,
+      "loss": 0.1395,
+      "step": 8895
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.6092690825462341,
+      "learning_rate": 4.710973751376552e-05,
+      "loss": 0.1531,
+      "step": 8900
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.6689764261245728,
+      "learning_rate": 4.710425598207695e-05,
+      "loss": 0.1082,
+      "step": 8905
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 2.6942625045776367,
+      "learning_rate": 4.709876957683748e-05,
+      "loss": 0.1441,
+      "step": 8910
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 1.0281902551651,
+      "learning_rate": 4.709327829925675e-05,
+      "loss": 0.1352,
+      "step": 8915
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 0.6816349625587463,
+      "learning_rate": 4.708778215054549e-05,
+      "loss": 0.1376,
+      "step": 8920
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 0.4717511832714081,
+      "learning_rate": 4.708228113191547e-05,
+      "loss": 0.1303,
+      "step": 8925
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 0.7959825992584229,
+      "learning_rate": 4.707677524457957e-05,
+      "loss": 0.1409,
+      "step": 8930
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 1.0031319856643677,
+      "learning_rate": 4.707126448975173e-05,
+      "loss": 0.1315,
+      "step": 8935
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 1.1377489566802979,
+      "learning_rate": 4.706574886864696e-05,
+      "loss": 0.1429,
+      "step": 8940
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 0.8349043726921082,
+      "learning_rate": 4.706022838248135e-05,
+      "loss": 0.1126,
+      "step": 8945
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 1.7055108547210693,
+      "learning_rate": 4.705470303247205e-05,
+      "loss": 0.1389,
+      "step": 8950
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 1.3086012601852417,
+      "learning_rate": 4.704917281983728e-05,
+      "loss": 0.1442,
+      "step": 8955
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 0.7926598787307739,
+      "learning_rate": 4.7043637745796375e-05,
+      "loss": 0.1426,
+      "step": 8960
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 0.5920445919036865,
+      "learning_rate": 4.703809781156968e-05,
+      "loss": 0.1271,
+      "step": 8965
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 0.6018089056015015,
+      "learning_rate": 4.7032553018378656e-05,
+      "loss": 0.1264,
+      "step": 8970
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 0.7071680426597595,
+      "learning_rate": 4.702700336744581e-05,
+      "loss": 0.1402,
+      "step": 8975
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 0.6931545734405518,
+      "learning_rate": 4.7021448859994735e-05,
+      "loss": 0.1359,
+      "step": 8980
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 0.8551448583602905,
+      "learning_rate": 4.701588949725009e-05,
+      "loss": 0.1324,
+      "step": 8985
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 0.7851433157920837,
+      "learning_rate": 4.701032528043761e-05,
+      "loss": 0.1205,
+      "step": 8990
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 1.1726717948913574,
+      "learning_rate": 4.700475621078408e-05,
+      "loss": 0.1186,
+      "step": 8995
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 0.6591439247131348,
+      "learning_rate": 4.6999182289517386e-05,
+      "loss": 0.1457,
+      "step": 9000
+    },
+    {
+      "epoch": 3.39,
+      "eval_loss": 0.13104431331157684,
+      "eval_runtime": 347.4643,
+      "eval_samples_per_second": 65.592,
+      "eval_steps_per_second": 8.199,
+      "eval_wer": 0.09942053801866191,
+      "step": 9000
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 1.3822407722473145,
+      "learning_rate": 4.6993603517866456e-05,
+      "loss": 0.1318,
+      "step": 9005
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 2.8463807106018066,
+      "learning_rate": 4.6988019897061305e-05,
+      "loss": 0.1343,
+      "step": 9010
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 0.992216169834137,
+      "learning_rate": 4.6982431428333005e-05,
+      "loss": 0.1385,
+      "step": 9015
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 0.7758907079696655,
+      "learning_rate": 4.697683811291371e-05,
+      "loss": 0.1094,
+      "step": 9020
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 6.152056694030762,
+      "learning_rate": 4.697123995203663e-05,
+      "loss": 0.2175,
+      "step": 9025
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 0.9680445194244385,
+      "learning_rate": 4.696563694693605e-05,
+      "loss": 0.1502,
+      "step": 9030
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 0.4121876060962677,
+      "learning_rate": 4.6960029098847326e-05,
+      "loss": 0.1196,
+      "step": 9035
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 0.756770133972168,
+      "learning_rate": 4.6954416409006877e-05,
+      "loss": 0.1171,
+      "step": 9040
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 0.47032052278518677,
+      "learning_rate": 4.694879887865218e-05,
+      "loss": 0.1336,
+      "step": 9045
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 0.6112526655197144,
+      "learning_rate": 4.694317650902179e-05,
+      "loss": 0.1383,
+      "step": 9050
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 0.5650899410247803,
+      "learning_rate": 4.6937549301355346e-05,
+      "loss": 0.1186,
+      "step": 9055
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 0.6806352734565735,
+      "learning_rate": 4.693191725689351e-05,
+      "loss": 0.1189,
+      "step": 9060
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 1.0119138956069946,
+      "learning_rate": 4.6926280376878064e-05,
+      "loss": 0.1665,
+      "step": 9065
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 0.6327919960021973,
+      "learning_rate": 4.692063866255181e-05,
+      "loss": 0.1165,
+      "step": 9070
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 0.9597269892692566,
+      "learning_rate": 4.6914992115158634e-05,
+      "loss": 0.1288,
+      "step": 9075
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 1.064296007156372,
+      "learning_rate": 4.6909340735943496e-05,
+      "loss": 0.145,
+      "step": 9080
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 0.7135687470436096,
+      "learning_rate": 4.690368452615242e-05,
+      "loss": 0.1098,
+      "step": 9085
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 0.8503847122192383,
+      "learning_rate": 4.689802348703246e-05,
+      "loss": 0.1261,
+      "step": 9090
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 0.881354033946991,
+      "learning_rate": 4.6892357619831796e-05,
+      "loss": 0.128,
+      "step": 9095
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 1.258421540260315,
+      "learning_rate": 4.688668692579962e-05,
+      "loss": 0.1283,
+      "step": 9100
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 0.6556168794631958,
+      "learning_rate": 4.688101140618621e-05,
+      "loss": 0.1255,
+      "step": 9105
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 1.089166522026062,
+      "learning_rate": 4.687533106224291e-05,
+      "loss": 0.1139,
+      "step": 9110
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 0.8696060180664062,
+      "learning_rate": 4.686964589522213e-05,
+      "loss": 0.1307,
+      "step": 9115
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 4.420854568481445,
+      "learning_rate": 4.686395590637732e-05,
+      "loss": 0.1245,
+      "step": 9120
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.7838614583015442,
+      "learning_rate": 4.685826109696302e-05,
+      "loss": 0.1311,
+      "step": 9125
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.7102704644203186,
+      "learning_rate": 4.6852561468234825e-05,
+      "loss": 0.1278,
+      "step": 9130
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.8456223011016846,
+      "learning_rate": 4.6846857021449384e-05,
+      "loss": 0.118,
+      "step": 9135
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.5998367667198181,
+      "learning_rate": 4.6841147757864416e-05,
+      "loss": 0.1293,
+      "step": 9140
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 1.2673438787460327,
+      "learning_rate": 4.683543367873871e-05,
+      "loss": 0.1783,
+      "step": 9145
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 1.1556962728500366,
+      "learning_rate": 4.682971478533209e-05,
+      "loss": 0.1244,
+      "step": 9150
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 0.846036970615387,
+      "learning_rate": 4.682399107890547e-05,
+      "loss": 0.1219,
+      "step": 9155
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 0.8751852512359619,
+      "learning_rate": 4.6818262560720817e-05,
+      "loss": 0.135,
+      "step": 9160
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 0.8367753624916077,
+      "learning_rate": 4.6812529232041144e-05,
+      "loss": 0.1478,
+      "step": 9165
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 0.73105788230896,
+      "learning_rate": 4.680679109413055e-05,
+      "loss": 0.1227,
+      "step": 9170
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 1.4446587562561035,
+      "learning_rate": 4.680104814825417e-05,
+      "loss": 0.1132,
+      "step": 9175
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 0.7135963439941406,
+      "learning_rate": 4.679530039567822e-05,
+      "loss": 0.1352,
+      "step": 9180
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 1.1135661602020264,
+      "learning_rate": 4.678954783766996e-05,
+      "loss": 0.1136,
+      "step": 9185
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 1.1743266582489014,
+      "learning_rate": 4.67837904754977e-05,
+      "loss": 0.1365,
+      "step": 9190
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 0.9979383945465088,
+      "learning_rate": 4.6779181127614826e-05,
+      "loss": 0.1377,
+      "step": 9195
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 0.8301486968994141,
+      "learning_rate": 4.677341512114697e-05,
+      "loss": 0.1251,
+      "step": 9200
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 0.6283919811248779,
+      "learning_rate": 4.676764431407207e-05,
+      "loss": 0.1394,
+      "step": 9205
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 0.7997094988822937,
+      "learning_rate": 4.676186870766247e-05,
+      "loss": 0.1117,
+      "step": 9210
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 1.4697551727294922,
+      "learning_rate": 4.675608830319159e-05,
+      "loss": 0.1297,
+      "step": 9215
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 1.7063761949539185,
+      "learning_rate": 4.6750303101933897e-05,
+      "loss": 0.115,
+      "step": 9220
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 0.647357702255249,
+      "learning_rate": 4.67445131051649e-05,
+      "loss": 0.1041,
+      "step": 9225
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 0.6971457004547119,
+      "learning_rate": 4.673871831416118e-05,
+      "loss": 0.1163,
+      "step": 9230
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 0.7368799448013306,
+      "learning_rate": 4.6732918730200396e-05,
+      "loss": 0.1237,
+      "step": 9235
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 0.7850015759468079,
+      "learning_rate": 4.672711435456121e-05,
+      "loss": 0.1225,
+      "step": 9240
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 1.4816852807998657,
+      "learning_rate": 4.672130518852339e-05,
+      "loss": 0.1666,
+      "step": 9245
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 0.8336917161941528,
+      "learning_rate": 4.6715491233367733e-05,
+      "loss": 0.1092,
+      "step": 9250
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 3.449495315551758,
+      "learning_rate": 4.670967249037611e-05,
+      "loss": 0.1117,
+      "step": 9255
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 0.6401161551475525,
+      "learning_rate": 4.670384896083144e-05,
+      "loss": 0.1469,
+      "step": 9260
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 0.5815885663032532,
+      "learning_rate": 4.669802064601769e-05,
+      "loss": 0.1064,
+      "step": 9265
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 0.8685303926467896,
+      "learning_rate": 4.669218754721989e-05,
+      "loss": 0.1059,
+      "step": 9270
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 0.9634084701538086,
+      "learning_rate": 4.6686349665724116e-05,
+      "loss": 0.1296,
+      "step": 9275
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 0.5055758357048035,
+      "learning_rate": 4.668050700281752e-05,
+      "loss": 0.1222,
+      "step": 9280
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 0.6615349650382996,
+      "learning_rate": 4.667465955978827e-05,
+      "loss": 0.112,
+      "step": 9285
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 0.6351237893104553,
+      "learning_rate": 4.666880733792563e-05,
+      "loss": 0.1346,
+      "step": 9290
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 0.8278821706771851,
+      "learning_rate": 4.666295033851991e-05,
+      "loss": 0.1261,
+      "step": 9295
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 0.6576123237609863,
+      "learning_rate": 4.665708856286243e-05,
+      "loss": 0.1241,
+      "step": 9300
+    },
+    {
+      "epoch": 3.5,
+      "eval_loss": 0.1316123753786087,
+      "eval_runtime": 349.5684,
+      "eval_samples_per_second": 65.198,
+      "eval_steps_per_second": 8.15,
+      "eval_wer": 0.09396921117067038,
+      "step": 9300
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 0.5844833850860596,
+      "learning_rate": 4.665122201224562e-05,
+      "loss": 0.1223,
+      "step": 9305
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 0.7872360348701477,
+      "learning_rate": 4.6645350687962934e-05,
+      "loss": 0.1205,
+      "step": 9310
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 0.5193818211555481,
+      "learning_rate": 4.6639474591308884e-05,
+      "loss": 0.1279,
+      "step": 9315
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 0.6526384353637695,
+      "learning_rate": 4.6633593723579014e-05,
+      "loss": 0.1007,
+      "step": 9320
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 0.5502133369445801,
+      "learning_rate": 4.662770808606997e-05,
+      "loss": 0.1397,
+      "step": 9325
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 2.969034194946289,
+      "learning_rate": 4.6621817680079394e-05,
+      "loss": 0.1423,
+      "step": 9330
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.8931148052215576,
+      "learning_rate": 4.6615922506906016e-05,
+      "loss": 0.1349,
+      "step": 9335
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.8049196600914001,
+      "learning_rate": 4.66100225678496e-05,
+      "loss": 0.1176,
+      "step": 9340
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.9487189054489136,
+      "learning_rate": 4.6604117864210975e-05,
+      "loss": 0.1154,
+      "step": 9345
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.6153743267059326,
+      "learning_rate": 4.6598208397291994e-05,
+      "loss": 0.1227,
+      "step": 9350
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.7711120843887329,
+      "learning_rate": 4.659229416839559e-05,
+      "loss": 0.1538,
+      "step": 9355
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.7197700142860413,
+      "learning_rate": 4.658637517882573e-05,
+      "loss": 0.1353,
+      "step": 9360
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 0.6496450304985046,
+      "learning_rate": 4.6580451429887434e-05,
+      "loss": 0.1097,
+      "step": 9365
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 0.5324687361717224,
+      "learning_rate": 4.657452292288677e-05,
+      "loss": 0.1329,
+      "step": 9370
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 0.5129036903381348,
+      "learning_rate": 4.656858965913087e-05,
+      "loss": 0.1252,
+      "step": 9375
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 0.8698485493659973,
+      "learning_rate": 4.6562651639927876e-05,
+      "loss": 0.1403,
+      "step": 9380
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 1.4602028131484985,
+      "learning_rate": 4.655670886658703e-05,
+      "loss": 0.123,
+      "step": 9385
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 0.7693333625793457,
+      "learning_rate": 4.655076134041857e-05,
+      "loss": 0.1407,
+      "step": 9390
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 1.1744771003723145,
+      "learning_rate": 4.654480906273383e-05,
+      "loss": 0.119,
+      "step": 9395
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 0.5806212425231934,
+      "learning_rate": 4.653885203484515e-05,
+      "loss": 0.1153,
+      "step": 9400
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 0.5284711718559265,
+      "learning_rate": 4.6532890258065964e-05,
+      "loss": 0.1281,
+      "step": 9405
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 0.6468020677566528,
+      "learning_rate": 4.652692373371069e-05,
+      "loss": 0.1262,
+      "step": 9410
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 0.7064224481582642,
+      "learning_rate": 4.652095246309487e-05,
+      "loss": 0.101,
+      "step": 9415
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 0.4459165930747986,
+      "learning_rate": 4.6514976447535006e-05,
+      "loss": 0.1245,
+      "step": 9420
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 0.49144649505615234,
+      "learning_rate": 4.650899568834872e-05,
+      "loss": 0.1151,
+      "step": 9425
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 0.44763481616973877,
+      "learning_rate": 4.6503010186854654e-05,
+      "loss": 0.1209,
+      "step": 9430
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 0.6725236177444458,
+      "learning_rate": 4.649701994437248e-05,
+      "loss": 0.2172,
+      "step": 9435
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 0.606113851070404,
+      "learning_rate": 4.649102496222292e-05,
+      "loss": 0.1184,
+      "step": 9440
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 0.7666552662849426,
+      "learning_rate": 4.648502524172777e-05,
+      "loss": 0.1372,
+      "step": 9445
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 0.5229029059410095,
+      "learning_rate": 4.6479020784209836e-05,
+      "loss": 0.1136,
+      "step": 9450
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 1.0143108367919922,
+      "learning_rate": 4.647301159099299e-05,
+      "loss": 0.1149,
+      "step": 9455
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 0.6132538914680481,
+      "learning_rate": 4.646699766340213e-05,
+      "loss": 0.1223,
+      "step": 9460
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 0.6279433369636536,
+      "learning_rate": 4.6460979002763215e-05,
+      "loss": 0.1192,
+      "step": 9465
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 0.4661136269569397,
+      "learning_rate": 4.6454955610403247e-05,
+      "loss": 0.1219,
+      "step": 9470
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 0.5851460099220276,
+      "learning_rate": 4.644892748765025e-05,
+      "loss": 0.1283,
+      "step": 9475
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 0.5985649824142456,
+      "learning_rate": 4.644289463583331e-05,
+      "loss": 0.1351,
+      "step": 9480
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 1.5796538591384888,
+      "learning_rate": 4.643685705628256e-05,
+      "loss": 0.1166,
+      "step": 9485
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 0.4589870870113373,
+      "learning_rate": 4.6430814750329155e-05,
+      "loss": 0.1324,
+      "step": 9490
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 0.6069396138191223,
+      "learning_rate": 4.642476771930532e-05,
+      "loss": 0.1062,
+      "step": 9495
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 1.1090596914291382,
+      "learning_rate": 4.641871596454429e-05,
+      "loss": 0.1267,
+      "step": 9500
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 1.309804916381836,
+      "learning_rate": 4.6412659487380364e-05,
+      "loss": 0.1029,
+      "step": 9505
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 0.5746486783027649,
+      "learning_rate": 4.6406598289148864e-05,
+      "loss": 0.1208,
+      "step": 9510
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 0.6103121638298035,
+      "learning_rate": 4.640053237118619e-05,
+      "loss": 0.1023,
+      "step": 9515
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 0.8701170086860657,
+      "learning_rate": 4.639446173482973e-05,
+      "loss": 0.1236,
+      "step": 9520
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 0.6347059607505798,
+      "learning_rate": 4.6388386381417946e-05,
+      "loss": 0.1116,
+      "step": 9525
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 0.511256754398346,
+      "learning_rate": 4.638230631229035e-05,
+      "loss": 0.1329,
+      "step": 9530
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 0.9881731271743774,
+      "learning_rate": 4.637622152878746e-05,
+      "loss": 0.1139,
+      "step": 9535
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 0.48110178112983704,
+      "learning_rate": 4.637013203225085e-05,
+      "loss": 0.1167,
+      "step": 9540
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 0.4912276864051819,
+      "learning_rate": 4.636403782402314e-05,
+      "loss": 0.1361,
+      "step": 9545
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.4828767776489258,
+      "learning_rate": 4.635793890544798e-05,
+      "loss": 0.1128,
+      "step": 9550
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.5929685831069946,
+      "learning_rate": 4.635183527787007e-05,
+      "loss": 0.1178,
+      "step": 9555
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.9599171876907349,
+      "learning_rate": 4.634572694263512e-05,
+      "loss": 0.1216,
+      "step": 9560
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.6496784687042236,
+      "learning_rate": 4.6339613901089914e-05,
+      "loss": 0.1132,
+      "step": 9565
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.440505713224411,
+      "learning_rate": 4.633349615458225e-05,
+      "loss": 0.1151,
+      "step": 9570
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 1.028018832206726,
+      "learning_rate": 4.632737370446096e-05,
+      "loss": 0.1405,
+      "step": 9575
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 0.5779049396514893,
+      "learning_rate": 4.6321246552075935e-05,
+      "loss": 0.1764,
+      "step": 9580
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 0.7614063024520874,
+      "learning_rate": 4.63151146987781e-05,
+      "loss": 0.1225,
+      "step": 9585
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 0.7574771046638489,
+      "learning_rate": 4.630897814591938e-05,
+      "loss": 0.135,
+      "step": 9590
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 0.7674686908721924,
+      "learning_rate": 4.6302836894852794e-05,
+      "loss": 0.1486,
+      "step": 9595
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.7298499941825867,
+      "learning_rate": 4.6296690946932344e-05,
+      "loss": 0.1248,
+      "step": 9600
+    },
+    {
+      "epoch": 3.62,
+      "eval_loss": 0.1291228085756302,
+      "eval_runtime": 360.2168,
+      "eval_samples_per_second": 63.27,
+      "eval_steps_per_second": 7.909,
+      "eval_wer": 0.09563397524981801,
+      "step": 9600
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.7000705599784851,
+      "learning_rate": 4.62905403035131e-05,
+      "loss": 0.1535,
+      "step": 9605
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 1.0471701622009277,
+      "learning_rate": 4.628438496595116e-05,
+      "loss": 0.1316,
+      "step": 9610
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.8082237839698792,
+      "learning_rate": 4.627822493560363e-05,
+      "loss": 0.1255,
+      "step": 9615
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.4863331615924835,
+      "learning_rate": 4.627206021382871e-05,
+      "loss": 0.1462,
+      "step": 9620
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.48594653606414795,
+      "learning_rate": 4.626589080198558e-05,
+      "loss": 0.1303,
+      "step": 9625
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 0.7101598978042603,
+      "learning_rate": 4.625971670143447e-05,
+      "loss": 0.1218,
+      "step": 9630
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 1.078147053718567,
+      "learning_rate": 4.625353791353666e-05,
+      "loss": 0.1354,
+      "step": 9635
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 0.5419949889183044,
+      "learning_rate": 4.6247354439654445e-05,
+      "loss": 0.112,
+      "step": 9640
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 0.8821031451225281,
+      "learning_rate": 4.624116628115115e-05,
+      "loss": 0.1391,
+      "step": 9645
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 0.9303906559944153,
+      "learning_rate": 4.623497343939115e-05,
+      "loss": 0.113,
+      "step": 9650
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 0.618428111076355,
+      "learning_rate": 4.622877591573985e-05,
+      "loss": 0.139,
+      "step": 9655
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 0.6350094079971313,
+      "learning_rate": 4.622257371156367e-05,
+      "loss": 0.1359,
+      "step": 9660
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 0.7334204316139221,
+      "learning_rate": 4.621636682823007e-05,
+      "loss": 0.1388,
+      "step": 9665
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 1.0643165111541748,
+      "learning_rate": 4.6210155267107554e-05,
+      "loss": 0.1184,
+      "step": 9670
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 0.5198789238929749,
+      "learning_rate": 4.6203939029565644e-05,
+      "loss": 0.117,
+      "step": 9675
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 1.4521546363830566,
+      "learning_rate": 4.619771811697491e-05,
+      "loss": 0.1264,
+      "step": 9680
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.5426422357559204,
+      "learning_rate": 4.619149253070692e-05,
+      "loss": 0.1217,
+      "step": 9685
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.49881890416145325,
+      "learning_rate": 4.61852622721343e-05,
+      "loss": 0.1245,
+      "step": 9690
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.4794444739818573,
+      "learning_rate": 4.617902734263071e-05,
+      "loss": 0.117,
+      "step": 9695
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.7690044045448303,
+      "learning_rate": 4.617278774357081e-05,
+      "loss": 0.1167,
+      "step": 9700
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.6047037839889526,
+      "learning_rate": 4.616654347633032e-05,
+      "loss": 0.1226,
+      "step": 9705
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 0.8620999455451965,
+      "learning_rate": 4.6160294542285986e-05,
+      "loss": 0.1287,
+      "step": 9710
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 0.6710473299026489,
+      "learning_rate": 4.615404094281554e-05,
+      "loss": 0.1236,
+      "step": 9715
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 0.597441554069519,
+      "learning_rate": 4.614778267929782e-05,
+      "loss": 0.1272,
+      "step": 9720
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 0.9019821286201477,
+      "learning_rate": 4.614151975311263e-05,
+      "loss": 0.1166,
+      "step": 9725
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 0.6891220808029175,
+      "learning_rate": 4.613525216564081e-05,
+      "loss": 0.106,
+      "step": 9730
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 0.7619643807411194,
+      "learning_rate": 4.612897991826425e-05,
+      "loss": 0.1188,
+      "step": 9735
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 0.5636903643608093,
+      "learning_rate": 4.612270301236587e-05,
+      "loss": 0.1144,
+      "step": 9740
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 0.5604903697967529,
+      "learning_rate": 4.611642144932957e-05,
+      "loss": 0.1297,
+      "step": 9745
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 1.0471038818359375,
+      "learning_rate": 4.611013523054033e-05,
+      "loss": 0.1185,
+      "step": 9750
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 0.7037379741668701,
+      "learning_rate": 4.6103844357384143e-05,
+      "loss": 0.1076,
+      "step": 9755
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 0.42084038257598877,
+      "learning_rate": 4.6097548831248016e-05,
+      "loss": 0.1138,
+      "step": 9760
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 0.5540190935134888,
+      "learning_rate": 4.609124865351999e-05,
+      "loss": 0.1229,
+      "step": 9765
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 0.5398131608963013,
+      "learning_rate": 4.608494382558912e-05,
+      "loss": 0.1235,
+      "step": 9770
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 0.625120222568512,
+      "learning_rate": 4.607863434884551e-05,
+      "loss": 0.1192,
+      "step": 9775
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 0.5883089303970337,
+      "learning_rate": 4.607232022468026e-05,
+      "loss": 0.128,
+      "step": 9780
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 0.8678866624832153,
+      "learning_rate": 4.606600145448552e-05,
+      "loss": 0.1226,
+      "step": 9785
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 0.8574846982955933,
+      "learning_rate": 4.6059678039654446e-05,
+      "loss": 0.1308,
+      "step": 9790
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 0.5269057154655457,
+      "learning_rate": 4.605334998158123e-05,
+      "loss": 0.1142,
+      "step": 9795
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 1.129819631576538,
+      "learning_rate": 4.604701728166109e-05,
+      "loss": 0.1073,
+      "step": 9800
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 0.7982564568519592,
+      "learning_rate": 4.6040679941290256e-05,
+      "loss": 0.1187,
+      "step": 9805
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 0.6812602281570435,
+      "learning_rate": 4.603433796186598e-05,
+      "loss": 0.116,
+      "step": 9810
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 0.5453401803970337,
+      "learning_rate": 4.6027991344786546e-05,
+      "loss": 0.1043,
+      "step": 9815
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 0.5465959906578064,
+      "learning_rate": 4.602164009145126e-05,
+      "loss": 0.1028,
+      "step": 9820
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 0.6713508367538452,
+      "learning_rate": 4.6015284203260446e-05,
+      "loss": 0.1186,
+      "step": 9825
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 2.9205212593078613,
+      "learning_rate": 4.600892368161546e-05,
+      "loss": 0.1036,
+      "step": 9830
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 0.561049222946167,
+      "learning_rate": 4.600255852791865e-05,
+      "loss": 0.1341,
+      "step": 9835
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 0.981108546257019,
+      "learning_rate": 4.5996188743573435e-05,
+      "loss": 0.1092,
+      "step": 9840
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 0.6135011911392212,
+      "learning_rate": 4.598981432998419e-05,
+      "loss": 0.1265,
+      "step": 9845
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 0.6785146594047546,
+      "learning_rate": 4.598343528855639e-05,
+      "loss": 0.1275,
+      "step": 9850
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 0.6356462836265564,
+      "learning_rate": 4.597705162069646e-05,
+      "loss": 0.123,
+      "step": 9855
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 0.7627720236778259,
+      "learning_rate": 4.597066332781187e-05,
+      "loss": 0.1023,
+      "step": 9860
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 1.4823070764541626,
+      "learning_rate": 4.596427041131113e-05,
+      "loss": 0.1236,
+      "step": 9865
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 0.8360366821289062,
+      "learning_rate": 4.595787287260373e-05,
+      "loss": 0.1262,
+      "step": 9870
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 0.5162907242774963,
+      "learning_rate": 4.595147071310022e-05,
+      "loss": 0.1125,
+      "step": 9875
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 0.5380714535713196,
+      "learning_rate": 4.594506393421214e-05,
+      "loss": 0.1324,
+      "step": 9880
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 0.965778112411499,
+      "learning_rate": 4.593865253735206e-05,
+      "loss": 0.1213,
+      "step": 9885
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 1.179997444152832,
+      "learning_rate": 4.593223652393357e-05,
+      "loss": 0.1187,
+      "step": 9890
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 0.9584506154060364,
+      "learning_rate": 4.592581589537126e-05,
+      "loss": 0.1391,
+      "step": 9895
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 0.8409070372581482,
+      "learning_rate": 4.591939065308077e-05,
+      "loss": 0.1211,
+      "step": 9900
+    },
+    {
+      "epoch": 3.73,
+      "eval_loss": 0.12598523497581482,
+      "eval_runtime": 343.4404,
+      "eval_samples_per_second": 66.361,
+      "eval_steps_per_second": 8.295,
+      "eval_wer": 0.09366727880352062,
+      "step": 9900
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 0.6772841811180115,
+      "learning_rate": 4.5912960798478725e-05,
+      "loss": 0.115,
+      "step": 9905
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 2.683145523071289,
+      "learning_rate": 4.590652633298279e-05,
+      "loss": 0.1391,
+      "step": 9910
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 0.33495739102363586,
+      "learning_rate": 4.5900087258011625e-05,
+      "loss": 0.1227,
+      "step": 9915
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 1.172780990600586,
+      "learning_rate": 4.589364357498493e-05,
+      "loss": 0.1233,
+      "step": 9920
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 0.8327208161354065,
+      "learning_rate": 4.588719528532342e-05,
+      "loss": 0.1339,
+      "step": 9925
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 0.4039396345615387,
+      "learning_rate": 4.5880742390448786e-05,
+      "loss": 0.1062,
+      "step": 9930
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 0.7174561619758606,
+      "learning_rate": 4.5874284891783777e-05,
+      "loss": 0.117,
+      "step": 9935
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 0.7369344830513,
+      "learning_rate": 4.586782279075214e-05,
+      "loss": 0.1122,
+      "step": 9940
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.5557066798210144,
+      "learning_rate": 4.586135608877866e-05,
+      "loss": 0.1088,
+      "step": 9945
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.5122894644737244,
+      "learning_rate": 4.58548847872891e-05,
+      "loss": 0.1175,
+      "step": 9950
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.5674937963485718,
+      "learning_rate": 4.5848408887710256e-05,
+      "loss": 0.1085,
+      "step": 9955
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.6235806345939636,
+      "learning_rate": 4.584192839146993e-05,
+      "loss": 0.1142,
+      "step": 9960
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.519451916217804,
+      "learning_rate": 4.583544329999696e-05,
+      "loss": 0.1304,
+      "step": 9965
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 3.5502662658691406,
+      "learning_rate": 4.582895361472117e-05,
+      "loss": 0.1148,
+      "step": 9970
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 1.3351777791976929,
+      "learning_rate": 4.58224593370734e-05,
+      "loss": 0.1168,
+      "step": 9975
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 0.5293915271759033,
+      "learning_rate": 4.5815960468485514e-05,
+      "loss": 0.1253,
+      "step": 9980
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 3.5751657485961914,
+      "learning_rate": 4.580945701039039e-05,
+      "loss": 0.1245,
+      "step": 9985
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 0.8494266867637634,
+      "learning_rate": 4.5802948964221907e-05,
+      "loss": 0.1284,
+      "step": 9990
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 0.9918404817581177,
+      "learning_rate": 4.579643633141496e-05,
+      "loss": 0.1397,
+      "step": 9995
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 0.5575226545333862,
+      "learning_rate": 4.578991911340545e-05,
+      "loss": 0.1166,
+      "step": 10000
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 0.7927202582359314,
+      "learning_rate": 4.578339731163032e-05,
+      "loss": 0.1449,
+      "step": 10005
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 0.4653702974319458,
+      "learning_rate": 4.5776870927527474e-05,
+      "loss": 0.1294,
+      "step": 10010
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 0.9459448456764221,
+      "learning_rate": 4.577033996253585e-05,
+      "loss": 0.1068,
+      "step": 10015
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 0.539118766784668,
+      "learning_rate": 4.57638044180954e-05,
+      "loss": 0.1217,
+      "step": 10020
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 0.6312394142150879,
+      "learning_rate": 4.575726429564709e-05,
+      "loss": 0.1182,
+      "step": 10025
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 0.6760942935943604,
+      "learning_rate": 4.5750719596632885e-05,
+      "loss": 0.1229,
+      "step": 10030
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 0.6139078736305237,
+      "learning_rate": 4.574417032249575e-05,
+      "loss": 0.1176,
+      "step": 10035
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 1.2270928621292114,
+      "learning_rate": 4.573761647467969e-05,
+      "loss": 0.1177,
+      "step": 10040
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 0.4608573615550995,
+      "learning_rate": 4.573105805462968e-05,
+      "loss": 0.1147,
+      "step": 10045
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 0.46734586358070374,
+      "learning_rate": 4.572449506379174e-05,
+      "loss": 0.1282,
+      "step": 10050
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 0.6161500811576843,
+      "learning_rate": 4.571792750361287e-05,
+      "loss": 0.1114,
+      "step": 10055
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 0.4614641070365906,
+      "learning_rate": 4.571135537554108e-05,
+      "loss": 0.1305,
+      "step": 10060
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 0.40924376249313354,
+      "learning_rate": 4.5704778681025396e-05,
+      "loss": 0.1212,
+      "step": 10065
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 0.5856077671051025,
+      "learning_rate": 4.569819742151585e-05,
+      "loss": 0.128,
+      "step": 10070
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 0.8363874554634094,
+      "learning_rate": 4.5691611598463504e-05,
+      "loss": 0.1063,
+      "step": 10075
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.7160053849220276,
+      "learning_rate": 4.568502121332037e-05,
+      "loss": 0.1103,
+      "step": 10080
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.7000710368156433,
+      "learning_rate": 4.56784262675395e-05,
+      "loss": 0.1376,
+      "step": 10085
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.5111421346664429,
+      "learning_rate": 4.567182676257496e-05,
+      "loss": 0.1168,
+      "step": 10090
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.4339483380317688,
+      "learning_rate": 4.5665222699881814e-05,
+      "loss": 0.1124,
+      "step": 10095
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.7811172604560852,
+      "learning_rate": 4.565861408091612e-05,
+      "loss": 0.1035,
+      "step": 10100
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 0.5670338869094849,
+      "learning_rate": 4.565200090713494e-05,
+      "loss": 0.1329,
+      "step": 10105
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 0.5968881249427795,
+      "learning_rate": 4.564538317999637e-05,
+      "loss": 0.1226,
+      "step": 10110
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 0.6024501919746399,
+      "learning_rate": 4.5638760900959454e-05,
+      "loss": 0.1206,
+      "step": 10115
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 0.7764694690704346,
+      "learning_rate": 4.563213407148431e-05,
+      "loss": 0.1229,
+      "step": 10120
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 0.7752511501312256,
+      "learning_rate": 4.5625502693031994e-05,
+      "loss": 0.1398,
+      "step": 10125
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 1.0763044357299805,
+      "learning_rate": 4.561886676706462e-05,
+      "loss": 0.1294,
+      "step": 10130
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 0.8312416672706604,
+      "learning_rate": 4.561222629504526e-05,
+      "loss": 0.1455,
+      "step": 10135
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 1.1743764877319336,
+      "learning_rate": 4.5605581278438e-05,
+      "loss": 0.1279,
+      "step": 10140
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 0.7562772631645203,
+      "learning_rate": 4.559893171870796e-05,
+      "loss": 0.1134,
+      "step": 10145
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 0.5048375725746155,
+      "learning_rate": 4.559227761732121e-05,
+      "loss": 0.1186,
+      "step": 10150
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 0.6762803792953491,
+      "learning_rate": 4.5585618975744864e-05,
+      "loss": 0.1098,
+      "step": 10155
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 0.7154070734977722,
+      "learning_rate": 4.5578955795447017e-05,
+      "loss": 0.1212,
+      "step": 10160
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 0.705720067024231,
+      "learning_rate": 4.5572288077896766e-05,
+      "loss": 0.1643,
+      "step": 10165
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 1.1014642715454102,
+      "learning_rate": 4.556561582456421e-05,
+      "loss": 0.1248,
+      "step": 10170
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 0.48759591579437256,
+      "learning_rate": 4.5558939036920454e-05,
+      "loss": 0.1174,
+      "step": 10175
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 0.6076847910881042,
+      "learning_rate": 4.555225771643759e-05,
+      "loss": 0.1058,
+      "step": 10180
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.6337810754776001,
+      "learning_rate": 4.554557186458872e-05,
+      "loss": 0.1444,
+      "step": 10185
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.7306234240531921,
+      "learning_rate": 4.5538881482847954e-05,
+      "loss": 0.1205,
+      "step": 10190
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.6337497234344482,
+      "learning_rate": 4.553218657269037e-05,
+      "loss": 0.1067,
+      "step": 10195
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.8178172707557678,
+      "learning_rate": 4.552548713559207e-05,
+      "loss": 0.1297,
+      "step": 10200
+    },
+    {
+      "epoch": 3.84,
+      "eval_loss": 0.11812999844551086,
+      "eval_runtime": 388.1242,
+      "eval_samples_per_second": 58.721,
+      "eval_steps_per_second": 7.34,
+      "eval_wer": 0.09233960360002647,
+      "step": 10200
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.6999384164810181,
+      "learning_rate": 4.551878317303015e-05,
+      "loss": 0.1096,
+      "step": 10205
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.8604580760002136,
+      "learning_rate": 4.55120746864827e-05,
+      "loss": 0.1334,
+      "step": 10210
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 0.6813471913337708,
+      "learning_rate": 4.550536167742881e-05,
+      "loss": 0.1139,
+      "step": 10215
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 4.989047050476074,
+      "learning_rate": 4.549864414734856e-05,
+      "loss": 0.1884,
+      "step": 10220
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 0.9823594689369202,
+      "learning_rate": 4.5491922097723036e-05,
+      "loss": 0.1126,
+      "step": 10225
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 0.5847375988960266,
+      "learning_rate": 4.548519553003432e-05,
+      "loss": 0.1323,
+      "step": 10230
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 0.7734892964363098,
+      "learning_rate": 4.5478464445765485e-05,
+      "loss": 0.1452,
+      "step": 10235
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 1.0305038690567017,
+      "learning_rate": 4.54717288464006e-05,
+      "loss": 0.1214,
+      "step": 10240
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 1.6698479652404785,
+      "learning_rate": 4.546498873342473e-05,
+      "loss": 0.114,
+      "step": 10245
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 0.6180856823921204,
+      "learning_rate": 4.545824410832393e-05,
+      "loss": 0.1232,
+      "step": 10250
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 0.6324100494384766,
+      "learning_rate": 4.545149497258528e-05,
+      "loss": 0.1263,
+      "step": 10255
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 0.6736887693405151,
+      "learning_rate": 4.544474132769681e-05,
+      "loss": 0.1184,
+      "step": 10260
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 0.9379294514656067,
+      "learning_rate": 4.543798317514757e-05,
+      "loss": 0.1083,
+      "step": 10265
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 0.7278345227241516,
+      "learning_rate": 4.54312205164276e-05,
+      "loss": 0.1269,
+      "step": 10270
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 1.234625220298767,
+      "learning_rate": 4.5424453353027927e-05,
+      "loss": 0.1364,
+      "step": 10275
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 0.7699393033981323,
+      "learning_rate": 4.541768168644059e-05,
+      "loss": 0.1229,
+      "step": 10280
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 1.213131308555603,
+      "learning_rate": 4.54109055181586e-05,
+      "loss": 0.1385,
+      "step": 10285
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 0.646499752998352,
+      "learning_rate": 4.5404124849675955e-05,
+      "loss": 0.1181,
+      "step": 10290
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 1.1265209913253784,
+      "learning_rate": 4.5397339682487676e-05,
+      "loss": 0.1292,
+      "step": 10295
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 0.9453241229057312,
+      "learning_rate": 4.539055001808975e-05,
+      "loss": 0.1148,
+      "step": 10300
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 0.6833706498146057,
+      "learning_rate": 4.538375585797917e-05,
+      "loss": 0.1207,
+      "step": 10305
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 0.8939727544784546,
+      "learning_rate": 4.537695720365391e-05,
+      "loss": 0.121,
+      "step": 10310
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 0.850491464138031,
+      "learning_rate": 4.537015405661294e-05,
+      "loss": 0.1089,
+      "step": 10315
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 0.5569143295288086,
+      "learning_rate": 4.536334641835621e-05,
+      "loss": 0.1175,
+      "step": 10320
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 0.8940430879592896,
+      "learning_rate": 4.535653429038469e-05,
+      "loss": 0.1051,
+      "step": 10325
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 0.5401516556739807,
+      "learning_rate": 4.534971767420029e-05,
+      "loss": 0.1233,
+      "step": 10330
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 1.5264440774917603,
+      "learning_rate": 4.534289657130597e-05,
+      "loss": 0.1279,
+      "step": 10335
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 0.6563264727592468,
+      "learning_rate": 4.533607098320562e-05,
+      "loss": 0.1424,
+      "step": 10340
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 1.2829557657241821,
+      "learning_rate": 4.532924091140417e-05,
+      "loss": 0.1429,
+      "step": 10345
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 0.8469234704971313,
+      "learning_rate": 4.532240635740751e-05,
+      "loss": 0.1241,
+      "step": 10350
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 0.9134402871131897,
+      "learning_rate": 4.531556732272251e-05,
+      "loss": 0.1316,
+      "step": 10355
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 1.5426002740859985,
+      "learning_rate": 4.5308723808857055e-05,
+      "loss": 0.1394,
+      "step": 10360
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 0.916485071182251,
+      "learning_rate": 4.5301875817320006e-05,
+      "loss": 0.1325,
+      "step": 10365
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 0.6410298347473145,
+      "learning_rate": 4.52950233496212e-05,
+      "loss": 0.1273,
+      "step": 10370
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 3.180185079574585,
+      "learning_rate": 4.528816640727147e-05,
+      "loss": 0.1473,
+      "step": 10375
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 0.5908339023590088,
+      "learning_rate": 4.5281304991782645e-05,
+      "loss": 0.1595,
+      "step": 10380
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 0.9462037086486816,
+      "learning_rate": 4.527443910466753e-05,
+      "loss": 0.1369,
+      "step": 10385
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 1.5562236309051514,
+      "learning_rate": 4.5267568747439924e-05,
+      "loss": 0.131,
+      "step": 10390
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 0.6705333590507507,
+      "learning_rate": 4.5260693921614584e-05,
+      "loss": 0.1146,
+      "step": 10395
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 4.000081539154053,
+      "learning_rate": 4.5253814628707286e-05,
+      "loss": 0.1372,
+      "step": 10400
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 0.5764330625534058,
+      "learning_rate": 4.5246930870234776e-05,
+      "loss": 0.1234,
+      "step": 10405
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 0.5140576362609863,
+      "learning_rate": 4.5240042647714794e-05,
+      "loss": 0.137,
+      "step": 10410
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 1.791358232498169,
+      "learning_rate": 4.523314996266604e-05,
+      "loss": 0.1302,
+      "step": 10415
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 0.8805174827575684,
+      "learning_rate": 4.522625281660823e-05,
+      "loss": 0.108,
+      "step": 10420
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 1.0258464813232422,
+      "learning_rate": 4.5219351211062046e-05,
+      "loss": 0.1307,
+      "step": 10425
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 1.4129974842071533,
+      "learning_rate": 4.5212445147549146e-05,
+      "loss": 0.1353,
+      "step": 10430
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 0.9663375616073608,
+      "learning_rate": 4.520553462759218e-05,
+      "loss": 0.1157,
+      "step": 10435
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 1.0501443147659302,
+      "learning_rate": 4.5198619652714804e-05,
+      "loss": 0.1195,
+      "step": 10440
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 0.5439842939376831,
+      "learning_rate": 4.51917002244416e-05,
+      "loss": 0.1257,
+      "step": 10445
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 0.5388464331626892,
+      "learning_rate": 4.5184776344298194e-05,
+      "loss": 0.1168,
+      "step": 10450
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 0.951256513595581,
+      "learning_rate": 4.517784801381114e-05,
+      "loss": 0.1228,
+      "step": 10455
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 0.8097508549690247,
+      "learning_rate": 4.5170915234508016e-05,
+      "loss": 0.1096,
+      "step": 10460
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 0.8620818853378296,
+      "learning_rate": 4.5163978007917356e-05,
+      "loss": 0.1288,
+      "step": 10465
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 0.8986382484436035,
+      "learning_rate": 4.5157036335568684e-05,
+      "loss": 0.1038,
+      "step": 10470
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 0.8695809245109558,
+      "learning_rate": 4.5150090218992494e-05,
+      "loss": 0.1159,
+      "step": 10475
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 0.6661204695701599,
+      "learning_rate": 4.514313965972027e-05,
+      "loss": 0.1298,
+      "step": 10480
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 0.8178394436836243,
+      "learning_rate": 4.5136184659284475e-05,
+      "loss": 0.121,
+      "step": 10485
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 0.6948071122169495,
+      "learning_rate": 4.5129225219218554e-05,
+      "loss": 0.1266,
+      "step": 10490
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 0.5710222721099854,
+      "learning_rate": 4.5122261341056915e-05,
+      "loss": 0.1269,
+      "step": 10495
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 0.371713250875473,
+      "learning_rate": 4.511529302633496e-05,
+      "loss": 0.1032,
+      "step": 10500
+    },
+    {
+      "epoch": 3.95,
+      "eval_loss": 0.12301083654165268,
+      "eval_runtime": 343.487,
+      "eval_samples_per_second": 66.352,
+      "eval_steps_per_second": 8.294,
+      "eval_wer": 0.0954892131559791,
+      "step": 10500
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 0.5750148296356201,
+      "learning_rate": 4.5108320276589066e-05,
+      "loss": 0.1473,
+      "step": 10505
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 0.9539104104042053,
+      "learning_rate": 4.510134309335658e-05,
+      "loss": 0.143,
+      "step": 10510
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 0.75587397813797,
+      "learning_rate": 4.509575815569399e-05,
+      "loss": 0.141,
+      "step": 10515
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 0.5331276059150696,
+      "learning_rate": 4.50887729960629e-05,
+      "loss": 0.1212,
+      "step": 10520
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 0.7280718684196472,
+      "learning_rate": 4.508178340725502e-05,
+      "loss": 0.1184,
+      "step": 10525
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 1.2351130247116089,
+      "learning_rate": 4.50747893908114e-05,
+      "loss": 0.1369,
+      "step": 10530
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 1.0986759662628174,
+      "learning_rate": 4.5067790948274094e-05,
+      "loss": 0.1263,
+      "step": 10535
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 1.1033953428268433,
+      "learning_rate": 4.506078808118612e-05,
+      "loss": 0.1302,
+      "step": 10540
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 0.7697123885154724,
+      "learning_rate": 4.505378079109146e-05,
+      "loss": 0.1187,
+      "step": 10545
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 1.2046051025390625,
+      "learning_rate": 4.504676907953509e-05,
+      "loss": 0.1178,
+      "step": 10550
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 12.999878883361816,
+      "learning_rate": 4.5039752948062955e-05,
+      "loss": 0.1276,
+      "step": 10555
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 1.3212690353393555,
+      "learning_rate": 4.503273239822197e-05,
+      "loss": 0.1241,
+      "step": 10560
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 0.6658570170402527,
+      "learning_rate": 4.502570743156004e-05,
+      "loss": 0.1363,
+      "step": 10565
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 0.5766234993934631,
+      "learning_rate": 4.5018678049626e-05,
+      "loss": 0.1191,
+      "step": 10570
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 0.8135549426078796,
+      "learning_rate": 4.501164425396973e-05,
+      "loss": 0.1251,
+      "step": 10575
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 1.0841425657272339,
+      "learning_rate": 4.5004606046142014e-05,
+      "loss": 0.0945,
+      "step": 10580
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 0.9046411514282227,
+      "learning_rate": 4.499756342769465e-05,
+      "loss": 0.1268,
+      "step": 10585
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 0.5654231309890747,
+      "learning_rate": 4.499051640018039e-05,
+      "loss": 0.1349,
+      "step": 10590
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 0.8395793437957764,
+      "learning_rate": 4.498346496515297e-05,
+      "loss": 0.1255,
+      "step": 10595
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 0.7574907541275024,
+      "learning_rate": 4.497640912416707e-05,
+      "loss": 0.1216,
+      "step": 10600
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 0.8355883359909058,
+      "learning_rate": 4.49693488787784e-05,
+      "loss": 0.1181,
+      "step": 10605
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.6714770793914795,
+      "learning_rate": 4.496228423054358e-05,
+      "loss": 0.1384,
+      "step": 10610
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.5086207985877991,
+      "learning_rate": 4.4955215181020225e-05,
+      "loss": 0.125,
+      "step": 10615
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.7007457613945007,
+      "learning_rate": 4.494814173176694e-05,
+      "loss": 0.128,
+      "step": 10620
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.6361066102981567,
+      "learning_rate": 4.494106388434325e-05,
+      "loss": 0.1291,
+      "step": 10625
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.6375873684883118,
+      "learning_rate": 4.4933981640309706e-05,
+      "loss": 0.1134,
+      "step": 10630
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 2.894223928451538,
+      "learning_rate": 4.4926895001227796e-05,
+      "loss": 0.1393,
+      "step": 10635
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 0.46987053751945496,
+      "learning_rate": 4.491980396865998e-05,
+      "loss": 0.0992,
+      "step": 10640
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 0.8070619106292725,
+      "learning_rate": 4.491270854416969e-05,
+      "loss": 0.1604,
+      "step": 10645
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 1.4966495037078857,
+      "learning_rate": 4.490560872932132e-05,
+      "loss": 0.1159,
+      "step": 10650
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 0.6059778332710266,
+      "learning_rate": 4.489850452568026e-05,
+      "loss": 0.1202,
+      "step": 10655
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 0.6750738024711609,
+      "learning_rate": 4.489139593481283e-05,
+      "loss": 0.1008,
+      "step": 10660
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 0.897981584072113,
+      "learning_rate": 4.488428295828633e-05,
+      "loss": 0.1406,
+      "step": 10665
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 0.9830949306488037,
+      "learning_rate": 4.487716559766904e-05,
+      "loss": 0.1156,
+      "step": 10670
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 0.5929490923881531,
+      "learning_rate": 4.48700438545302e-05,
+      "loss": 0.1252,
+      "step": 10675
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 1.2048383951187134,
+      "learning_rate": 4.486291773044e-05,
+      "loss": 0.1295,
+      "step": 10680
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 0.4293842017650604,
+      "learning_rate": 4.485578722696962e-05,
+      "loss": 0.1027,
+      "step": 10685
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 1.2220641374588013,
+      "learning_rate": 4.484865234569119e-05,
+      "loss": 0.1179,
+      "step": 10690
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 0.6177108287811279,
+      "learning_rate": 4.4841513088177813e-05,
+      "loss": 0.117,
+      "step": 10695
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 0.5701348781585693,
+      "learning_rate": 4.4834369456003556e-05,
+      "loss": 0.1049,
+      "step": 10700
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 0.8367097973823547,
+      "learning_rate": 4.482722145074345e-05,
+      "loss": 0.0968,
+      "step": 10705
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 0.6506603956222534,
+      "learning_rate": 4.482006907397347e-05,
+      "loss": 0.1041,
+      "step": 10710
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 0.6454862952232361,
+      "learning_rate": 4.48129123272706e-05,
+      "loss": 0.0925,
+      "step": 10715
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 0.9330674409866333,
+      "learning_rate": 4.4805751212212744e-05,
+      "loss": 0.1346,
+      "step": 10720
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 0.8604361414909363,
+      "learning_rate": 4.4798585730378784e-05,
+      "loss": 0.104,
+      "step": 10725
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 0.5917999148368835,
+      "learning_rate": 4.4791415883348584e-05,
+      "loss": 0.1604,
+      "step": 10730
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 0.5542749166488647,
+      "learning_rate": 4.4784241672702944e-05,
+      "loss": 0.1139,
+      "step": 10735
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 0.7743073105812073,
+      "learning_rate": 4.477706310002362e-05,
+      "loss": 0.1259,
+      "step": 10740
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 0.7721580266952515,
+      "learning_rate": 4.476988016689337e-05,
+      "loss": 0.1085,
+      "step": 10745
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 0.710868239402771,
+      "learning_rate": 4.476269287489588e-05,
+      "loss": 0.1056,
+      "step": 10750
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 1.2893180847167969,
+      "learning_rate": 4.47555012256158e-05,
+      "loss": 0.1086,
+      "step": 10755
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 0.6400284171104431,
+      "learning_rate": 4.4748305220638754e-05,
+      "loss": 0.1147,
+      "step": 10760
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 0.7458717226982117,
+      "learning_rate": 4.474110486155131e-05,
+      "loss": 0.1233,
+      "step": 10765
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 0.6369180083274841,
+      "learning_rate": 4.473390014994101e-05,
+      "loss": 0.1015,
+      "step": 10770
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 0.49289730191230774,
+      "learning_rate": 4.4726691087396346e-05,
+      "loss": 0.1123,
+      "step": 10775
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 0.6343613862991333,
+      "learning_rate": 4.4719477675506785e-05,
+      "loss": 0.1054,
+      "step": 10780
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 0.7233530282974243,
+      "learning_rate": 4.471225991586273e-05,
+      "loss": 0.1157,
+      "step": 10785
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 0.4932377338409424,
+      "learning_rate": 4.470503781005556e-05,
+      "loss": 0.115,
+      "step": 10790
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 0.7427234649658203,
+      "learning_rate": 4.46978113596776e-05,
+      "loss": 0.1008,
+      "step": 10795
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 0.3277433216571808,
+      "learning_rate": 4.469058056632215e-05,
+      "loss": 0.1129,
+      "step": 10800
+    },
+    {
+      "epoch": 4.07,
+      "eval_loss": 0.1181061714887619,
+      "eval_runtime": 327.6949,
+      "eval_samples_per_second": 69.549,
+      "eval_steps_per_second": 8.694,
+      "eval_wer": 0.08912381708689035,
+      "step": 10800
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 0.576930046081543,
+      "learning_rate": 4.468334543158345e-05,
+      "loss": 0.116,
+      "step": 10805
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 0.7138528227806091,
+      "learning_rate": 4.46761059570567e-05,
+      "loss": 0.1064,
+      "step": 10810
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 0.7877904772758484,
+      "learning_rate": 4.4668862144338064e-05,
+      "loss": 0.0996,
+      "step": 10815
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 0.38472509384155273,
+      "learning_rate": 4.466161399502466e-05,
+      "loss": 0.1113,
+      "step": 10820
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 0.5444314479827881,
+      "learning_rate": 4.465436151071456e-05,
+      "loss": 0.1272,
+      "step": 10825
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 2.8608767986297607,
+      "learning_rate": 4.46471046930068e-05,
+      "loss": 0.1273,
+      "step": 10830
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 0.5464024543762207,
+      "learning_rate": 4.4639843543501364e-05,
+      "loss": 0.1015,
+      "step": 10835
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 0.5623424649238586,
+      "learning_rate": 4.463257806379918e-05,
+      "loss": 0.1089,
+      "step": 10840
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 0.7799835801124573,
+      "learning_rate": 4.462530825550216e-05,
+      "loss": 0.1144,
+      "step": 10845
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 1.417871356010437,
+      "learning_rate": 4.461803412021314e-05,
+      "loss": 0.1316,
+      "step": 10850
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 0.9771950244903564,
+      "learning_rate": 4.4610755659535916e-05,
+      "loss": 0.1338,
+      "step": 10855
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 0.6710395216941833,
+      "learning_rate": 4.460347287507526e-05,
+      "loss": 0.1269,
+      "step": 10860
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 0.7745227217674255,
+      "learning_rate": 4.459618576843688e-05,
+      "loss": 0.1288,
+      "step": 10865
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 0.7659354209899902,
+      "learning_rate": 4.4588894341227426e-05,
+      "loss": 0.1047,
+      "step": 10870
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 0.6502729058265686,
+      "learning_rate": 4.458159859505453e-05,
+      "loss": 0.1103,
+      "step": 10875
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 0.534054696559906,
+      "learning_rate": 4.4574298531526746e-05,
+      "loss": 0.0999,
+      "step": 10880
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 1.0408005714416504,
+      "learning_rate": 4.45669941522536e-05,
+      "loss": 0.1073,
+      "step": 10885
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 0.6043410897254944,
+      "learning_rate": 4.4559685458845566e-05,
+      "loss": 0.1192,
+      "step": 10890
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 1.2987059354782104,
+      "learning_rate": 4.455237245291406e-05,
+      "loss": 0.1052,
+      "step": 10895
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 0.7181374430656433,
+      "learning_rate": 4.4545055136071464e-05,
+      "loss": 0.0914,
+      "step": 10900
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 0.707568883895874,
+      "learning_rate": 4.453773350993108e-05,
+      "loss": 0.1098,
+      "step": 10905
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 0.685168445110321,
+      "learning_rate": 4.453040757610721e-05,
+      "loss": 0.1261,
+      "step": 10910
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 0.6559328436851501,
+      "learning_rate": 4.452307733621506e-05,
+      "loss": 0.0941,
+      "step": 10915
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 1.288680911064148,
+      "learning_rate": 4.4515742791870814e-05,
+      "loss": 0.1148,
+      "step": 10920
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 0.7082732319831848,
+      "learning_rate": 4.450840394469158e-05,
+      "loss": 0.0992,
+      "step": 10925
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 0.5491362810134888,
+      "learning_rate": 4.4501060796295434e-05,
+      "loss": 0.108,
+      "step": 10930
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 0.6937622427940369,
+      "learning_rate": 4.4493713348301404e-05,
+      "loss": 0.1204,
+      "step": 10935
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 0.9267261028289795,
+      "learning_rate": 4.448636160232943e-05,
+      "loss": 0.0981,
+      "step": 10940
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 0.8009552359580994,
+      "learning_rate": 4.4479005560000464e-05,
+      "loss": 0.13,
+      "step": 10945
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 0.5700029730796814,
+      "learning_rate": 4.447164522293634e-05,
+      "loss": 0.1093,
+      "step": 10950
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 0.8897039890289307,
+      "learning_rate": 4.446428059275989e-05,
+      "loss": 0.1156,
+      "step": 10955
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 0.6198723912239075,
+      "learning_rate": 4.4456911671094834e-05,
+      "loss": 0.1209,
+      "step": 10960
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 1.261991024017334,
+      "learning_rate": 4.444953845956591e-05,
+      "loss": 0.1138,
+      "step": 10965
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 0.7931584119796753,
+      "learning_rate": 4.4442160959798745e-05,
+      "loss": 0.1036,
+      "step": 10970
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 1.5593605041503906,
+      "learning_rate": 4.443477917341994e-05,
+      "loss": 0.1381,
+      "step": 10975
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 0.41901832818984985,
+      "learning_rate": 4.442739310205702e-05,
+      "loss": 0.107,
+      "step": 10980
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 0.8393665552139282,
+      "learning_rate": 4.442000274733849e-05,
+      "loss": 0.1333,
+      "step": 10985
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 0.6136124730110168,
+      "learning_rate": 4.441260811089375e-05,
+      "loss": 0.1183,
+      "step": 10990
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 0.8086240291595459,
+      "learning_rate": 4.4405209194353185e-05,
+      "loss": 0.1054,
+      "step": 10995
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 0.3283528983592987,
+      "learning_rate": 4.439780599934812e-05,
+      "loss": 0.1047,
+      "step": 11000
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 0.6242526173591614,
+      "learning_rate": 4.439039852751079e-05,
+      "loss": 0.0945,
+      "step": 11005
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 0.9697647094726562,
+      "learning_rate": 4.438298678047441e-05,
+      "loss": 0.1188,
+      "step": 11010
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 0.6296168565750122,
+      "learning_rate": 4.437557075987313e-05,
+      "loss": 0.1135,
+      "step": 11015
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 0.5502399802207947,
+      "learning_rate": 4.4368150467342016e-05,
+      "loss": 0.1029,
+      "step": 11020
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 0.8477868437767029,
+      "learning_rate": 4.436072590451712e-05,
+      "loss": 0.1165,
+      "step": 11025
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 0.6576706767082214,
+      "learning_rate": 4.4353297073035395e-05,
+      "loss": 0.1207,
+      "step": 11030
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 0.8050335645675659,
+      "learning_rate": 4.434586397453475e-05,
+      "loss": 0.1078,
+      "step": 11035
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 0.5364821553230286,
+      "learning_rate": 4.433842661065404e-05,
+      "loss": 0.1225,
+      "step": 11040
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 1.1133692264556885,
+      "learning_rate": 4.4330984983033054e-05,
+      "loss": 0.1224,
+      "step": 11045
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 0.5409312844276428,
+      "learning_rate": 4.432353909331253e-05,
+      "loss": 0.1158,
+      "step": 11050
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 1.0942381620407104,
+      "learning_rate": 4.4316088943134135e-05,
+      "loss": 0.1114,
+      "step": 11055
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 0.7721982598304749,
+      "learning_rate": 4.430863453414048e-05,
+      "loss": 0.1095,
+      "step": 11060
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 0.5871670246124268,
+      "learning_rate": 4.430117586797512e-05,
+      "loss": 0.1203,
+      "step": 11065
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 0.705535888671875,
+      "learning_rate": 4.4293712946282534e-05,
+      "loss": 0.1083,
+      "step": 11070
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 0.7110278010368347,
+      "learning_rate": 4.4286245770708156e-05,
+      "loss": 0.1169,
+      "step": 11075
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 0.5212209820747375,
+      "learning_rate": 4.427877434289835e-05,
+      "loss": 0.1136,
+      "step": 11080
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 1.0680917501449585,
+      "learning_rate": 4.42712986645004e-05,
+      "loss": 0.1158,
+      "step": 11085
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 0.5946283936500549,
+      "learning_rate": 4.426381873716257e-05,
+      "loss": 0.1488,
+      "step": 11090
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 0.7046301364898682,
+      "learning_rate": 4.425633456253403e-05,
+      "loss": 0.133,
+      "step": 11095
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 0.6597256064414978,
+      "learning_rate": 4.4248846142264884e-05,
+      "loss": 0.1254,
+      "step": 11100
+    },
+    {
+      "epoch": 4.18,
+      "eval_loss": 0.12267578393220901,
+      "eval_runtime": 328.2528,
+      "eval_samples_per_second": 69.431,
+      "eval_steps_per_second": 8.679,
+      "eval_wer": 0.09097470385811661,
+      "step": 11100
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 0.7240242958068848,
+      "learning_rate": 4.424135347800619e-05,
+      "loss": 0.1148,
+      "step": 11105
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 0.5504785180091858,
+      "learning_rate": 4.423385657140991e-05,
+      "loss": 0.1011,
+      "step": 11110
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 0.8128982186317444,
+      "learning_rate": 4.4226355424129e-05,
+      "loss": 0.125,
+      "step": 11115
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 0.9528091549873352,
+      "learning_rate": 4.4218850037817284e-05,
+      "loss": 0.1223,
+      "step": 11120
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 0.6302881836891174,
+      "learning_rate": 4.421134041412957e-05,
+      "loss": 0.1031,
+      "step": 11125
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 0.837955892086029,
+      "learning_rate": 4.4203826554721575e-05,
+      "loss": 0.1218,
+      "step": 11130
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 0.8141761422157288,
+      "learning_rate": 4.4196308461249946e-05,
+      "loss": 0.1176,
+      "step": 11135
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 0.601660430431366,
+      "learning_rate": 4.4188786135372285e-05,
+      "loss": 0.1073,
+      "step": 11140
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 0.7040676474571228,
+      "learning_rate": 4.418125957874711e-05,
+      "loss": 0.1064,
+      "step": 11145
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 1.342969298362732,
+      "learning_rate": 4.417372879303389e-05,
+      "loss": 0.1205,
+      "step": 11150
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 1.3055496215820312,
+      "learning_rate": 4.416619377989301e-05,
+      "loss": 0.122,
+      "step": 11155
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 0.4434565305709839,
+      "learning_rate": 4.4158654540985776e-05,
+      "loss": 0.1226,
+      "step": 11160
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 0.5723581314086914,
+      "learning_rate": 4.415111107797445e-05,
+      "loss": 0.0945,
+      "step": 11165
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 1.5700428485870361,
+      "learning_rate": 4.4143563392522224e-05,
+      "loss": 0.1173,
+      "step": 11170
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 0.5924566388130188,
+      "learning_rate": 4.4136011486293205e-05,
+      "loss": 0.1218,
+      "step": 11175
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 1.0328441858291626,
+      "learning_rate": 4.4128455360952445e-05,
+      "loss": 0.1083,
+      "step": 11180
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 0.6595302820205688,
+      "learning_rate": 4.4120895018165915e-05,
+      "loss": 0.1225,
+      "step": 11185
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 0.6577940583229065,
+      "learning_rate": 4.411333045960052e-05,
+      "loss": 0.1056,
+      "step": 11190
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 1.5622667074203491,
+      "learning_rate": 4.4105761686924097e-05,
+      "loss": 0.1088,
+      "step": 11195
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 0.4584384262561798,
+      "learning_rate": 4.409818870180541e-05,
+      "loss": 0.0947,
+      "step": 11200
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 0.3986893594264984,
+      "learning_rate": 4.4090611505914156e-05,
+      "loss": 0.1075,
+      "step": 11205
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 1.67661714553833,
+      "learning_rate": 4.4083030100920953e-05,
+      "loss": 0.1052,
+      "step": 11210
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 0.5704238414764404,
+      "learning_rate": 4.4075444488497353e-05,
+      "loss": 0.1019,
+      "step": 11215
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 0.8068325519561768,
+      "learning_rate": 4.406785467031583e-05,
+      "loss": 0.1079,
+      "step": 11220
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 0.4918566346168518,
+      "learning_rate": 4.4060260648049796e-05,
+      "loss": 0.0934,
+      "step": 11225
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 0.6979566216468811,
+      "learning_rate": 4.405266242337357e-05,
+      "loss": 0.1211,
+      "step": 11230
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 0.6254603862762451,
+      "learning_rate": 4.4045059997962425e-05,
+      "loss": 0.1122,
+      "step": 11235
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 0.9355801343917847,
+      "learning_rate": 4.4037453373492535e-05,
+      "loss": 0.103,
+      "step": 11240
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 0.679449200630188,
+      "learning_rate": 4.4029842551641006e-05,
+      "loss": 0.1419,
+      "step": 11245
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 0.8364554047584534,
+      "learning_rate": 4.402222753408589e-05,
+      "loss": 0.211,
+      "step": 11250
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 0.6407322287559509,
+      "learning_rate": 4.401460832250613e-05,
+      "loss": 0.1203,
+      "step": 11255
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 0.5196644067764282,
+      "learning_rate": 4.4006984918581634e-05,
+      "loss": 0.1101,
+      "step": 11260
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 0.6516141295433044,
+      "learning_rate": 4.3999357323993185e-05,
+      "loss": 0.1253,
+      "step": 11265
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 0.532505989074707,
+      "learning_rate": 4.3991725540422534e-05,
+      "loss": 0.1196,
+      "step": 11270
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 0.7241432070732117,
+      "learning_rate": 4.3984089569552334e-05,
+      "loss": 0.1106,
+      "step": 11275
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 0.5014824867248535,
+      "learning_rate": 4.3976449413066156e-05,
+      "loss": 0.1187,
+      "step": 11280
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 0.6922964453697205,
+      "learning_rate": 4.3968805072648524e-05,
+      "loss": 0.13,
+      "step": 11285
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 0.6099381446838379,
+      "learning_rate": 4.396115654998484e-05,
+      "loss": 0.1074,
+      "step": 11290
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 0.7293393611907959,
+      "learning_rate": 4.395350384676147e-05,
+      "loss": 0.1127,
+      "step": 11295
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 0.6668974161148071,
+      "learning_rate": 4.394584696466567e-05,
+      "loss": 0.1223,
+      "step": 11300
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 3.558957576751709,
+      "learning_rate": 4.393818590538565e-05,
+      "loss": 0.1442,
+      "step": 11305
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 0.6252570748329163,
+      "learning_rate": 4.3930520670610496e-05,
+      "loss": 0.1142,
+      "step": 11310
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 0.9166604280471802,
+      "learning_rate": 4.3922851262030264e-05,
+      "loss": 0.1339,
+      "step": 11315
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 0.7122131586074829,
+      "learning_rate": 4.3915177681335894e-05,
+      "loss": 0.1238,
+      "step": 11320
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 0.7663761973381042,
+      "learning_rate": 4.3907499930219263e-05,
+      "loss": 0.1101,
+      "step": 11325
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 0.5663048624992371,
+      "learning_rate": 4.3899818010373164e-05,
+      "loss": 0.1138,
+      "step": 11330
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 1.463631272315979,
+      "learning_rate": 4.38921319234913e-05,
+      "loss": 0.1034,
+      "step": 11335
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 0.7809505462646484,
+      "learning_rate": 4.388444167126831e-05,
+      "loss": 0.108,
+      "step": 11340
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 0.6337461471557617,
+      "learning_rate": 4.387674725539974e-05,
+      "loss": 0.1186,
+      "step": 11345
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 0.9854856729507446,
+      "learning_rate": 4.3869048677582056e-05,
+      "loss": 0.1316,
+      "step": 11350
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 0.9039980173110962,
+      "learning_rate": 4.3861345939512644e-05,
+      "loss": 0.1008,
+      "step": 11355
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 0.48134496808052063,
+      "learning_rate": 4.385363904288981e-05,
+      "loss": 0.0996,
+      "step": 11360
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 0.8735333681106567,
+      "learning_rate": 4.384592798941275e-05,
+      "loss": 0.125,
+      "step": 11365
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 1.5752341747283936,
+      "learning_rate": 4.3838212780781624e-05,
+      "loss": 0.1218,
+      "step": 11370
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 1.080224633216858,
+      "learning_rate": 4.3830493418697474e-05,
+      "loss": 0.1223,
+      "step": 11375
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 0.8141139149665833,
+      "learning_rate": 4.382276990486226e-05,
+      "loss": 0.1091,
+      "step": 11380
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 0.644242525100708,
+      "learning_rate": 4.381504224097886e-05,
+      "loss": 0.1207,
+      "step": 11385
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 0.9917758703231812,
+      "learning_rate": 4.3807310428751094e-05,
+      "loss": 0.1394,
+      "step": 11390
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 0.5563603639602661,
+      "learning_rate": 4.3799574469883656e-05,
+      "loss": 0.1123,
+      "step": 11395
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 1.4581997394561768,
+      "learning_rate": 4.379183436608218e-05,
+      "loss": 0.1336,
+      "step": 11400
+    },
+    {
+      "epoch": 4.29,
+      "eval_loss": 0.1212385967373848,
+      "eval_runtime": 329.144,
+      "eval_samples_per_second": 69.243,
+      "eval_steps_per_second": 8.656,
+      "eval_wer": 0.08914036132618622,
+      "step": 11400
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 1.101729154586792,
+      "learning_rate": 4.37840901190532e-05,
+      "loss": 0.118,
+      "step": 11405
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 1.2017433643341064,
+      "learning_rate": 4.377634173050417e-05,
+      "loss": 0.1361,
+      "step": 11410
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 0.42585229873657227,
+      "learning_rate": 4.3768589202143454e-05,
+      "loss": 0.1025,
+      "step": 11415
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 0.7223350405693054,
+      "learning_rate": 4.376083253568033e-05,
+      "loss": 0.1065,
+      "step": 11420
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 0.5818420052528381,
+      "learning_rate": 4.3753071732825e-05,
+      "loss": 0.1159,
+      "step": 11425
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 0.7802611589431763,
+      "learning_rate": 4.374530679528856e-05,
+      "loss": 0.1233,
+      "step": 11430
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 1.396031141281128,
+      "learning_rate": 4.373753772478302e-05,
+      "loss": 0.1144,
+      "step": 11435
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 0.5439860224723816,
+      "learning_rate": 4.372976452302132e-05,
+      "loss": 0.1243,
+      "step": 11440
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 0.7264595627784729,
+      "learning_rate": 4.3721987191717285e-05,
+      "loss": 0.1286,
+      "step": 11445
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 1.3122107982635498,
+      "learning_rate": 4.3714205732585666e-05,
+      "loss": 0.124,
+      "step": 11450
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 0.5511311888694763,
+      "learning_rate": 4.370642014734212e-05,
+      "loss": 0.1129,
+      "step": 11455
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 0.6841253638267517,
+      "learning_rate": 4.369863043770322e-05,
+      "loss": 0.1288,
+      "step": 11460
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 0.7595229148864746,
+      "learning_rate": 4.369083660538643e-05,
+      "loss": 0.1105,
+      "step": 11465
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 0.7060916423797607,
+      "learning_rate": 4.368303865211015e-05,
+      "loss": 0.1076,
+      "step": 11470
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 0.6916689276695251,
+      "learning_rate": 4.3675236579593675e-05,
+      "loss": 0.11,
+      "step": 11475
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 0.6126403212547302,
+      "learning_rate": 4.36674303895572e-05,
+      "loss": 0.125,
+      "step": 11480
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 0.6903819441795349,
+      "learning_rate": 4.3659620083721834e-05,
+      "loss": 0.1019,
+      "step": 11485
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 1.33014714717865,
+      "learning_rate": 4.3651805663809585e-05,
+      "loss": 0.1018,
+      "step": 11490
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 3.088515520095825,
+      "learning_rate": 4.3643987131543406e-05,
+      "loss": 0.1152,
+      "step": 11495
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 0.46400344371795654,
+      "learning_rate": 4.363616448864711e-05,
+      "loss": 0.1052,
+      "step": 11500
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 0.7497923374176025,
+      "learning_rate": 4.362833773684544e-05,
+      "loss": 0.1198,
+      "step": 11505
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 1.2450987100601196,
+      "learning_rate": 4.362050687786403e-05,
+      "loss": 0.1203,
+      "step": 11510
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 1.2431817054748535,
+      "learning_rate": 4.361267191342944e-05,
+      "loss": 0.115,
+      "step": 11515
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 0.5462827682495117,
+      "learning_rate": 4.3604832845269113e-05,
+      "loss": 0.1054,
+      "step": 11520
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 0.6362592577934265,
+      "learning_rate": 4.3596989675111435e-05,
+      "loss": 0.1011,
+      "step": 11525
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 0.6811506152153015,
+      "learning_rate": 4.358914240468564e-05,
+      "loss": 0.1124,
+      "step": 11530
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 0.7225905656814575,
+      "learning_rate": 4.358129103572191e-05,
+      "loss": 0.1185,
+      "step": 11535
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 1.4909617900848389,
+      "learning_rate": 4.3573435569951314e-05,
+      "loss": 0.1017,
+      "step": 11540
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 1.002028226852417,
+      "learning_rate": 4.356557600910583e-05,
+      "loss": 0.1381,
+      "step": 11545
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 1.08548903465271,
+      "learning_rate": 4.3557712354918336e-05,
+      "loss": 0.104,
+      "step": 11550
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 0.5311010479927063,
+      "learning_rate": 4.354984460912259e-05,
+      "loss": 0.1141,
+      "step": 11555
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 0.5744786262512207,
+      "learning_rate": 4.354197277345331e-05,
+      "loss": 0.1013,
+      "step": 11560
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 0.802068829536438,
+      "learning_rate": 4.3534096849646054e-05,
+      "loss": 0.1124,
+      "step": 11565
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 0.5143678784370422,
+      "learning_rate": 4.3526216839437326e-05,
+      "loss": 0.1176,
+      "step": 11570
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 0.7047765851020813,
+      "learning_rate": 4.351833274456449e-05,
+      "loss": 0.1114,
+      "step": 11575
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 0.6705645322799683,
+      "learning_rate": 4.351044456676585e-05,
+      "loss": 0.1158,
+      "step": 11580
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 0.5474240779876709,
+      "learning_rate": 4.350255230778058e-05,
+      "loss": 0.1138,
+      "step": 11585
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 0.5242738723754883,
+      "learning_rate": 4.349465596934879e-05,
+      "loss": 0.1172,
+      "step": 11590
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 1.2543015480041504,
+      "learning_rate": 4.348675555321144e-05,
+      "loss": 0.1236,
+      "step": 11595
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 0.47975388169288635,
+      "learning_rate": 4.347885106111044e-05,
+      "loss": 0.1137,
+      "step": 11600
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 0.4253996014595032,
+      "learning_rate": 4.3470942494788556e-05,
+      "loss": 0.1176,
+      "step": 11605
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 0.4526536762714386,
+      "learning_rate": 4.346302985598947e-05,
+      "loss": 0.106,
+      "step": 11610
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 0.9203678369522095,
+      "learning_rate": 4.345511314645776e-05,
+      "loss": 0.1029,
+      "step": 11615
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 0.6806662082672119,
+      "learning_rate": 4.3447192367938926e-05,
+      "loss": 0.0862,
+      "step": 11620
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 0.622639536857605,
+      "learning_rate": 4.343926752217932e-05,
+      "loss": 0.1203,
+      "step": 11625
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 0.5239378809928894,
+      "learning_rate": 4.343133861092622e-05,
+      "loss": 0.1165,
+      "step": 11630
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 0.7566112279891968,
+      "learning_rate": 4.3423405635927795e-05,
+      "loss": 0.0981,
+      "step": 11635
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 1.5923075675964355,
+      "learning_rate": 4.341546859893312e-05,
+      "loss": 0.1014,
+      "step": 11640
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 0.7668944597244263,
+      "learning_rate": 4.3407527501692134e-05,
+      "loss": 0.1259,
+      "step": 11645
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 1.5492472648620605,
+      "learning_rate": 4.33995823459557e-05,
+      "loss": 0.1195,
+      "step": 11650
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 1.8058569431304932,
+      "learning_rate": 4.339163313347557e-05,
+      "loss": 0.1179,
+      "step": 11655
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 1.4584447145462036,
+      "learning_rate": 4.3383679866004404e-05,
+      "loss": 0.1121,
+      "step": 11660
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 1.08311927318573,
+      "learning_rate": 4.337572254529571e-05,
+      "loss": 0.1031,
+      "step": 11665
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 3.169051170349121,
+      "learning_rate": 4.3367761173103924e-05,
+      "loss": 0.1146,
+      "step": 11670
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 0.6249878406524658,
+      "learning_rate": 4.335979575118439e-05,
+      "loss": 0.1137,
+      "step": 11675
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 2.3852591514587402,
+      "learning_rate": 4.335182628129332e-05,
+      "loss": 0.1247,
+      "step": 11680
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 0.8955238461494446,
+      "learning_rate": 4.334385276518781e-05,
+      "loss": 0.1093,
+      "step": 11685
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 0.46420732140541077,
+      "learning_rate": 4.333587520462588e-05,
+      "loss": 0.1074,
+      "step": 11690
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 0.9716922640800476,
+      "learning_rate": 4.332789360136641e-05,
+      "loss": 0.0931,
+      "step": 11695
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 0.7544828653335571,
+      "learning_rate": 4.3319907957169197e-05,
+      "loss": 0.1363,
+      "step": 11700
+    },
+    {
+      "epoch": 4.41,
+      "eval_loss": 0.11756496131420135,
+      "eval_runtime": 327.5851,
+      "eval_samples_per_second": 69.573,
+      "eval_steps_per_second": 8.697,
+      "eval_wer": 0.08956637548805506,
+      "step": 11700
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 0.6067286729812622,
+      "learning_rate": 4.331191827379491e-05,
+      "loss": 0.0954,
+      "step": 11705
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 0.6290103197097778,
+      "learning_rate": 4.330392455300513e-05,
+      "loss": 0.1108,
+      "step": 11710
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 0.6087530255317688,
+      "learning_rate": 4.329592679656229e-05,
+      "loss": 0.1007,
+      "step": 11715
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 0.8399063944816589,
+      "learning_rate": 4.3287925006229756e-05,
+      "loss": 0.1084,
+      "step": 11720
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 0.46039116382598877,
+      "learning_rate": 4.327991918377176e-05,
+      "loss": 0.1074,
+      "step": 11725
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 0.7752964496612549,
+      "learning_rate": 4.3271909330953424e-05,
+      "loss": 0.1002,
+      "step": 11730
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 1.2215250730514526,
+      "learning_rate": 4.326389544954076e-05,
+      "loss": 0.1047,
+      "step": 11735
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 0.9924432039260864,
+      "learning_rate": 4.3255877541300685e-05,
+      "loss": 0.1024,
+      "step": 11740
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 0.5713456869125366,
+      "learning_rate": 4.324785560800097e-05,
+      "loss": 0.1083,
+      "step": 11745
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 1.273412823677063,
+      "learning_rate": 4.323982965141029e-05,
+      "loss": 0.1085,
+      "step": 11750
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 0.6912309527397156,
+      "learning_rate": 4.323179967329824e-05,
+      "loss": 0.1136,
+      "step": 11755
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 0.5459722876548767,
+      "learning_rate": 4.322376567543524e-05,
+      "loss": 0.1203,
+      "step": 11760
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 0.835425615310669,
+      "learning_rate": 4.321572765959264e-05,
+      "loss": 0.1244,
+      "step": 11765
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 0.5291778445243835,
+      "learning_rate": 4.320768562754267e-05,
+      "loss": 0.1034,
+      "step": 11770
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 1.0706766843795776,
+      "learning_rate": 4.319963958105843e-05,
+      "loss": 0.1162,
+      "step": 11775
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 0.6912168860435486,
+      "learning_rate": 4.31915895219139e-05,
+      "loss": 0.1216,
+      "step": 11780
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 0.6259210705757141,
+      "learning_rate": 4.3183535451883986e-05,
+      "loss": 0.1051,
+      "step": 11785
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 0.9779114723205566,
+      "learning_rate": 4.3175477372744435e-05,
+      "loss": 0.1049,
+      "step": 11790
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 0.9029730558395386,
+      "learning_rate": 4.3167415286271905e-05,
+      "loss": 0.1154,
+      "step": 11795
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 0.8965585231781006,
+      "learning_rate": 4.315934919424391e-05,
+      "loss": 0.1141,
+      "step": 11800
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 3.935770273208618,
+      "learning_rate": 4.315127909843888e-05,
+      "loss": 0.1249,
+      "step": 11805
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 1.3469558954238892,
+      "learning_rate": 4.31432050006361e-05,
+      "loss": 0.1179,
+      "step": 11810
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 1.3287595510482788,
+      "learning_rate": 4.3135126902615755e-05,
+      "loss": 0.126,
+      "step": 11815
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 0.7570639252662659,
+      "learning_rate": 4.3127044806158914e-05,
+      "loss": 0.1373,
+      "step": 11820
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 74.8725814819336,
+      "learning_rate": 4.311895871304749e-05,
+      "loss": 0.1584,
+      "step": 11825
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 0.9372453093528748,
+      "learning_rate": 4.311086862506434e-05,
+      "loss": 0.1173,
+      "step": 11830
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 0.6495732069015503,
+      "learning_rate": 4.310277454399314e-05,
+      "loss": 0.1416,
+      "step": 11835
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 0.5391997694969177,
+      "learning_rate": 4.30946764716185e-05,
+      "loss": 0.1877,
+      "step": 11840
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 0.6120247840881348,
+      "learning_rate": 4.308657440972587e-05,
+      "loss": 0.1217,
+      "step": 11845
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 0.6802361607551575,
+      "learning_rate": 4.30784683601016e-05,
+      "loss": 0.11,
+      "step": 11850
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 0.8464838266372681,
+      "learning_rate": 4.30703583245329e-05,
+      "loss": 0.1161,
+      "step": 11855
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 0.6961728930473328,
+      "learning_rate": 4.30622443048079e-05,
+      "loss": 0.1238,
+      "step": 11860
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 0.8787757158279419,
+      "learning_rate": 4.3054126302715545e-05,
+      "loss": 0.1259,
+      "step": 11865
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 1.5648038387298584,
+      "learning_rate": 4.304600432004573e-05,
+      "loss": 0.122,
+      "step": 11870
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 0.7913205623626709,
+      "learning_rate": 4.3037878358589165e-05,
+      "loss": 0.1142,
+      "step": 11875
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 1.0053131580352783,
+      "learning_rate": 4.302974842013747e-05,
+      "loss": 0.1224,
+      "step": 11880
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 0.8083571791648865,
+      "learning_rate": 4.302161450648315e-05,
+      "loss": 0.1339,
+      "step": 11885
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 0.687061071395874,
+      "learning_rate": 4.3013476619419546e-05,
+      "loss": 0.1134,
+      "step": 11890
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 1.2226988077163696,
+      "learning_rate": 4.300533476074092e-05,
+      "loss": 0.1084,
+      "step": 11895
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 1.0914326906204224,
+      "learning_rate": 4.29988184154415e-05,
+      "loss": 0.1273,
+      "step": 11900
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 0.5646753907203674,
+      "learning_rate": 4.2990669412380115e-05,
+      "loss": 0.1033,
+      "step": 11905
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 0.8335121870040894,
+      "learning_rate": 4.298251644273224e-05,
+      "loss": 0.12,
+      "step": 11910
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 0.7126038074493408,
+      "learning_rate": 4.2974359508295447e-05,
+      "loss": 0.1293,
+      "step": 11915
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 0.6517030596733093,
+      "learning_rate": 4.296619861086817e-05,
+      "loss": 0.1043,
+      "step": 11920
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 1.0076056718826294,
+      "learning_rate": 4.295803375224974e-05,
+      "loss": 0.111,
+      "step": 11925
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 0.6330487132072449,
+      "learning_rate": 4.294986493424033e-05,
+      "loss": 0.1049,
+      "step": 11930
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 3.109189987182617,
+      "learning_rate": 4.294169215864102e-05,
+      "loss": 0.1345,
+      "step": 11935
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 0.8041120767593384,
+      "learning_rate": 4.2933515427253724e-05,
+      "loss": 0.1229,
+      "step": 11940
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 0.5482778549194336,
+      "learning_rate": 4.292533474188127e-05,
+      "loss": 0.0951,
+      "step": 11945
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 0.5020703077316284,
+      "learning_rate": 4.2917150104327327e-05,
+      "loss": 0.121,
+      "step": 11950
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 0.7766263484954834,
+      "learning_rate": 4.290896151639644e-05,
+      "loss": 0.1119,
+      "step": 11955
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 0.8595438599586487,
+      "learning_rate": 4.290076897989404e-05,
+      "loss": 0.0955,
+      "step": 11960
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 0.5140489339828491,
+      "learning_rate": 4.289257249662643e-05,
+      "loss": 0.1286,
+      "step": 11965
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 0.8254638910293579,
+      "learning_rate": 4.2884372068400745e-05,
+      "loss": 0.1261,
+      "step": 11970
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 0.6057325005531311,
+      "learning_rate": 4.2876167697025036e-05,
+      "loss": 0.1251,
+      "step": 11975
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 0.5562911629676819,
+      "learning_rate": 4.28679593843082e-05,
+      "loss": 0.1149,
+      "step": 11980
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 0.8396725654602051,
+      "learning_rate": 4.2859747132060006e-05,
+      "loss": 0.1096,
+      "step": 11985
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 0.40999341011047363,
+      "learning_rate": 4.28515309420911e-05,
+      "loss": 0.0926,
+      "step": 11990
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 0.4702136218547821,
+      "learning_rate": 4.284331081621298e-05,
+      "loss": 0.1093,
+      "step": 11995
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 0.5203172564506531,
+      "learning_rate": 4.2835086756238016e-05,
+      "loss": 0.1138,
+      "step": 12000
+    },
+    {
+      "epoch": 4.52,
+      "eval_loss": 0.11922828108072281,
+      "eval_runtime": 327.0259,
+      "eval_samples_per_second": 69.692,
+      "eval_steps_per_second": 8.712,
+      "eval_wer": 0.09021780491033023,
+      "step": 12000
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 0.7019593715667725,
+      "learning_rate": 4.282685876397947e-05,
+      "loss": 0.1084,
+      "step": 12005
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 0.7474701404571533,
+      "learning_rate": 4.281862684125143e-05,
+      "loss": 0.1037,
+      "step": 12010
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 0.631989598274231,
+      "learning_rate": 4.2810390989868884e-05,
+      "loss": 0.1123,
+      "step": 12015
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 0.6779237389564514,
+      "learning_rate": 4.2803799481351825e-05,
+      "loss": 0.1359,
+      "step": 12020
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 0.7691624760627747,
+      "learning_rate": 4.2795556562967675e-05,
+      "loss": 0.1389,
+      "step": 12025
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 1.1529332399368286,
+      "learning_rate": 4.2787309721015554e-05,
+      "loss": 0.1085,
+      "step": 12030
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 1.0018315315246582,
+      "learning_rate": 4.2779058957313734e-05,
+      "loss": 0.1292,
+      "step": 12035
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 0.7257592678070068,
+      "learning_rate": 4.2770804273681344e-05,
+      "loss": 0.1761,
+      "step": 12040
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 1.0803810358047485,
+      "learning_rate": 4.276254567193837e-05,
+      "loss": 0.121,
+      "step": 12045
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 0.7213515043258667,
+      "learning_rate": 4.2754283153905675e-05,
+      "loss": 0.1204,
+      "step": 12050
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 0.5956887006759644,
+      "learning_rate": 4.274601672140498e-05,
+      "loss": 0.1286,
+      "step": 12055
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 0.541029155254364,
+      "learning_rate": 4.273774637625888e-05,
+      "loss": 0.1137,
+      "step": 12060
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 3.437516212463379,
+      "learning_rate": 4.272947212029081e-05,
+      "loss": 0.1279,
+      "step": 12065
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 0.7367995381355286,
+      "learning_rate": 4.272119395532508e-05,
+      "loss": 0.1196,
+      "step": 12070
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 1.5402261018753052,
+      "learning_rate": 4.271291188318687e-05,
+      "loss": 0.1246,
+      "step": 12075
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 0.7366180419921875,
+      "learning_rate": 4.27046259057022e-05,
+      "loss": 0.1135,
+      "step": 12080
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 1.216825008392334,
+      "learning_rate": 4.269633602469798e-05,
+      "loss": 0.11,
+      "step": 12085
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 0.8633574843406677,
+      "learning_rate": 4.268804224200196e-05,
+      "loss": 0.1173,
+      "step": 12090
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 0.6643964648246765,
+      "learning_rate": 4.267974455944275e-05,
+      "loss": 0.1316,
+      "step": 12095
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 1.0604419708251953,
+      "learning_rate": 4.267144297884982e-05,
+      "loss": 0.1199,
+      "step": 12100
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 0.6547016501426697,
+      "learning_rate": 4.2663137502053514e-05,
+      "loss": 0.1081,
+      "step": 12105
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 2.5705246925354004,
+      "learning_rate": 4.2654828130885015e-05,
+      "loss": 0.1099,
+      "step": 12110
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 1.048218011856079,
+      "learning_rate": 4.264651486717639e-05,
+      "loss": 0.118,
+      "step": 12115
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 0.8607979416847229,
+      "learning_rate": 4.263819771276054e-05,
+      "loss": 0.1208,
+      "step": 12120
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 1.3617023229599,
+      "learning_rate": 4.262987666947123e-05,
+      "loss": 0.1248,
+      "step": 12125
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 0.6462903022766113,
+      "learning_rate": 4.262155173914309e-05,
+      "loss": 0.1315,
+      "step": 12130
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 0.731080949306488,
+      "learning_rate": 4.2613222923611604e-05,
+      "loss": 0.1314,
+      "step": 12135
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 0.6185482740402222,
+      "learning_rate": 4.260489022471311e-05,
+      "loss": 0.1227,
+      "step": 12140
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 0.7105937600135803,
+      "learning_rate": 4.25965536442848e-05,
+      "loss": 0.1255,
+      "step": 12145
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 0.7570763230323792,
+      "learning_rate": 4.2588213184164725e-05,
+      "loss": 0.1108,
+      "step": 12150
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 0.5891075134277344,
+      "learning_rate": 4.25798688461918e-05,
+      "loss": 0.1049,
+      "step": 12155
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 0.7149835824966431,
+      "learning_rate": 4.2571520632205766e-05,
+      "loss": 0.1261,
+      "step": 12160
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 0.5590789914131165,
+      "learning_rate": 4.2563168544047255e-05,
+      "loss": 0.1046,
+      "step": 12165
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 1.0008964538574219,
+      "learning_rate": 4.255481258355773e-05,
+      "loss": 0.1371,
+      "step": 12170
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 0.7180017828941345,
+      "learning_rate": 4.2546452752579536e-05,
+      "loss": 0.1135,
+      "step": 12175
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 1.3704843521118164,
+      "learning_rate": 4.2538089052955824e-05,
+      "loss": 0.1311,
+      "step": 12180
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 0.7218495011329651,
+      "learning_rate": 4.252972148653063e-05,
+      "loss": 0.1165,
+      "step": 12185
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 0.6096352934837341,
+      "learning_rate": 4.2521350055148844e-05,
+      "loss": 0.1225,
+      "step": 12190
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 0.7255779504776001,
+      "learning_rate": 4.25129747606562e-05,
+      "loss": 0.1311,
+      "step": 12195
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 0.6532098054885864,
+      "learning_rate": 4.2504595604899275e-05,
+      "loss": 0.1291,
+      "step": 12200
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 1.2533913850784302,
+      "learning_rate": 4.249621258972552e-05,
+      "loss": 0.1456,
+      "step": 12205
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 1.3645999431610107,
+      "learning_rate": 4.2487825716983214e-05,
+      "loss": 0.1268,
+      "step": 12210
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 1.098724365234375,
+      "learning_rate": 4.2479434988521505e-05,
+      "loss": 0.1166,
+      "step": 12215
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 1.3733069896697998,
+      "learning_rate": 4.247104040619037e-05,
+      "loss": 0.1248,
+      "step": 12220
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 0.5608084201812744,
+      "learning_rate": 4.246264197184066e-05,
+      "loss": 0.1187,
+      "step": 12225
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 0.63069087266922,
+      "learning_rate": 4.245423968732406e-05,
+      "loss": 0.1101,
+      "step": 12230
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 1.7209112644195557,
+      "learning_rate": 4.244583355449311e-05,
+      "loss": 0.1302,
+      "step": 12235
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 0.6770058274269104,
+      "learning_rate": 4.2437423575201187e-05,
+      "loss": 0.1223,
+      "step": 12240
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 0.8090385794639587,
+      "learning_rate": 4.242900975130253e-05,
+      "loss": 0.1271,
+      "step": 12245
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 0.6945059299468994,
+      "learning_rate": 4.242059208465223e-05,
+      "loss": 0.1014,
+      "step": 12250
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 0.6217449307441711,
+      "learning_rate": 4.2412170577106205e-05,
+      "loss": 0.1191,
+      "step": 12255
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 0.6591787338256836,
+      "learning_rate": 4.240374523052123e-05,
+      "loss": 0.1185,
+      "step": 12260
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 0.6990086436271667,
+      "learning_rate": 4.239531604675493e-05,
+      "loss": 0.1158,
+      "step": 12265
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 1.3511993885040283,
+      "learning_rate": 4.238688302766577e-05,
+      "loss": 0.118,
+      "step": 12270
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 0.7292054891586304,
+      "learning_rate": 4.237844617511307e-05,
+      "loss": 0.1198,
+      "step": 12275
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 0.5475161075592041,
+      "learning_rate": 4.237000549095699e-05,
+      "loss": 0.1076,
+      "step": 12280
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 1.855530023574829,
+      "learning_rate": 4.2361560977058524e-05,
+      "loss": 0.1092,
+      "step": 12285
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 0.8871486186981201,
+      "learning_rate": 4.235311263527952e-05,
+      "loss": 0.1202,
+      "step": 12290
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 1.288094162940979,
+      "learning_rate": 4.2344660467482686e-05,
+      "loss": 0.1256,
+      "step": 12295
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 0.96184903383255,
+      "learning_rate": 4.2336204475531555e-05,
+      "loss": 0.1333,
+      "step": 12300
+    },
+    {
+      "epoch": 4.63,
+      "eval_loss": 0.12410614639520645,
+      "eval_runtime": 327.2988,
+      "eval_samples_per_second": 69.634,
+      "eval_steps_per_second": 8.705,
+      "eval_wer": 0.08887978955727616,
+      "step": 12300
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 37170,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 14,
+  "save_steps": 300,
+  "total_flos": 4.124402511035645e+20,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}