diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,36889 @@
+{
+  "best_metric": 1.0,
+  "best_model_checkpoint": "./wavbert_fongbe/checkpoint-5200",
+  "epoch": 1.8220042046250877,
+  "eval_steps": 100,
+  "global_step": 5200,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": Infinity,
+      "learning_rate": 0.0,
+      "loss": 9.8297,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": Infinity,
+      "learning_rate": 0.0,
+      "loss": 9.9301,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 34.74740982055664,
+      "learning_rate": 6e-07,
+      "loss": 10.394,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 25.30831527709961,
+      "learning_rate": 1.2e-06,
+      "loss": 8.5656,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 38.40359115600586,
+      "learning_rate": 1.8e-06,
+      "loss": 11.2407,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 28.476762771606445,
+      "learning_rate": 2.4e-06,
+      "loss": 8.109,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": Infinity,
+      "learning_rate": 2.4e-06,
+      "loss": 11.0636,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 34.81722640991211,
+      "learning_rate": 2.9999999999999997e-06,
+      "loss": 9.1192,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 53.64858627319336,
+      "learning_rate": 3.6e-06,
+      "loss": 11.2103,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 34.31063461303711,
+      "learning_rate": 4.2e-06,
+      "loss": 7.7964,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 40.947654724121094,
+      "learning_rate": 4.8e-06,
+      "loss": 8.9611,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 54.53398513793945,
+      "learning_rate": 5.399999999999999e-06,
+      "loss": 8.7025,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 74.66500091552734,
+      "learning_rate": 5.999999999999999e-06,
+      "loss": 9.4326,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 44.239418029785156,
+      "learning_rate": 6.599999999999999e-06,
+      "loss": 6.5448,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 43.60188293457031,
+      "learning_rate": 7.2e-06,
+      "loss": 5.5535,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 49.674495697021484,
+      "learning_rate": 7.799999999999998e-06,
+      "loss": 5.7704,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 70.16291046142578,
+      "learning_rate": 8.4e-06,
+      "loss": 5.6929,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 37.02568817138672,
+      "learning_rate": 8.999999999999999e-06,
+      "loss": 3.9512,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.776915550231934,
+      "learning_rate": 9.6e-06,
+      "loss": 3.7522,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 10.268403053283691,
+      "learning_rate": 1.02e-05,
+      "loss": 3.557,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 21.770320892333984,
+      "learning_rate": 1.0799999999999998e-05,
+      "loss": 3.8904,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 12.886055946350098,
+      "learning_rate": 1.14e-05,
+      "loss": 3.4933,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 20.950607299804688,
+      "learning_rate": 1.1999999999999999e-05,
+      "loss": 3.4173,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 22.54509925842285,
+      "learning_rate": 1.26e-05,
+      "loss": 3.5335,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 13.497519493103027,
+      "learning_rate": 1.3199999999999997e-05,
+      "loss": 2.9166,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 11.36432933807373,
+      "learning_rate": 1.3799999999999998e-05,
+      "loss": 3.6862,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 21.900163650512695,
+      "learning_rate": 1.44e-05,
+      "loss": 3.3244,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": NaN,
+      "learning_rate": 1.44e-05,
+      "loss": 3.1803,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.995499610900879,
+      "learning_rate": 1.4999999999999999e-05,
+      "loss": 3.4045,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 11.834275245666504,
+      "learning_rate": 1.5599999999999996e-05,
+      "loss": 3.1981,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.6730780601501465,
+      "learning_rate": 1.6199999999999997e-05,
+      "loss": 3.2896,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 14.79286003112793,
+      "learning_rate": 1.68e-05,
+      "loss": 3.1652,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.442844867706299,
+      "learning_rate": 1.74e-05,
+      "loss": 3.102,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.1123127937316895,
+      "learning_rate": 1.7999999999999997e-05,
+      "loss": 3.1152,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.981231212615967,
+      "learning_rate": 1.8599999999999998e-05,
+      "loss": 3.1492,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.901692867279053,
+      "learning_rate": 1.92e-05,
+      "loss": 3.1973,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 17.36743927001953,
+      "learning_rate": 1.98e-05,
+      "loss": 3.0872,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 11.805011749267578,
+      "learning_rate": 2.04e-05,
+      "loss": 2.8694,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.926337242126465,
+      "learning_rate": 2.1e-05,
+      "loss": 3.5693,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.315157890319824,
+      "learning_rate": 2.1599999999999996e-05,
+      "loss": 2.776,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.084774017333984,
+      "learning_rate": 2.2199999999999998e-05,
+      "loss": 2.8658,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 10.755606651306152,
+      "learning_rate": 2.28e-05,
+      "loss": 2.7053,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.2747626304626465,
+      "learning_rate": 2.34e-05,
+      "loss": 2.6037,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 9.70805835723877,
+      "learning_rate": 2.3999999999999997e-05,
+      "loss": 3.1741,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 18.0622615814209,
+      "learning_rate": 2.4599999999999998e-05,
+      "loss": 3.0041,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 14.248910903930664,
+      "learning_rate": 2.52e-05,
+      "loss": 2.6704,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.843019485473633,
+      "learning_rate": 2.5799999999999997e-05,
+      "loss": 2.5436,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 10.0279541015625,
+      "learning_rate": 2.6399999999999995e-05,
+      "loss": 2.5453,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 47.676048278808594,
+      "learning_rate": 2.6999999999999996e-05,
+      "loss": 2.4351,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 10.947311401367188,
+      "learning_rate": 2.7599999999999997e-05,
+      "loss": 2.8235,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.891237735748291,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 3.0562,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.135880947113037,
+      "learning_rate": 2.88e-05,
+      "loss": 3.0568,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.591911315917969,
+      "learning_rate": 2.94e-05,
+      "loss": 2.8868,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.592108726501465,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 2.8324,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.499967098236084,
+      "learning_rate": 3.06e-05,
+      "loss": 2.9309,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 13.009428977966309,
+      "learning_rate": 3.119999999999999e-05,
+      "loss": 2.9151,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 10.796736717224121,
+      "learning_rate": 3.1799999999999994e-05,
+      "loss": 2.9887,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.898942947387695,
+      "learning_rate": 3.2399999999999995e-05,
+      "loss": 2.7634,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.915682792663574,
+      "learning_rate": 3.2999999999999996e-05,
+      "loss": 2.3548,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.942079067230225,
+      "learning_rate": 3.36e-05,
+      "loss": 2.8065,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 9.648048400878906,
+      "learning_rate": 3.42e-05,
+      "loss": 2.7106,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 11.939984321594238,
+      "learning_rate": 3.48e-05,
+      "loss": 2.6683,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 8.364795684814453,
+      "learning_rate": 3.539999999999999e-05,
+      "loss": 3.1089,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.097488880157471,
+      "learning_rate": 3.5999999999999994e-05,
+      "loss": 2.5706,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.490532398223877,
+      "learning_rate": 3.6599999999999995e-05,
+      "loss": 2.7647,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.321822166442871,
+      "learning_rate": 3.7199999999999996e-05,
+      "loss": 2.9021,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 8.508000373840332,
+      "learning_rate": 3.78e-05,
+      "loss": 2.4467,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 8.116604804992676,
+      "learning_rate": 3.84e-05,
+      "loss": 2.2756,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 11.946988105773926,
+      "learning_rate": 3.9e-05,
+      "loss": 2.9973,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 11.239936828613281,
+      "learning_rate": 3.96e-05,
+      "loss": 2.6493,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 9.772220611572266,
+      "learning_rate": 4.02e-05,
+      "loss": 2.5128,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 12.785487174987793,
+      "learning_rate": 4.08e-05,
+      "loss": 1.8853,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 8.324564933776855,
+      "learning_rate": 4.14e-05,
+      "loss": 2.7886,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 11.687192916870117,
+      "learning_rate": 4.2e-05,
+      "loss": 1.838,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 9.597622871398926,
+      "learning_rate": 4.259999999999999e-05,
+      "loss": 1.8491,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 7.367631912231445,
+      "learning_rate": 4.319999999999999e-05,
+      "loss": 2.6484,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.516895771026611,
+      "learning_rate": 4.3799999999999994e-05,
+      "loss": 2.1669,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 7.077077388763428,
+      "learning_rate": 4.4399999999999995e-05,
+      "loss": 2.3611,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.426537990570068,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 1.8982,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.794361114501953,
+      "learning_rate": 4.56e-05,
+      "loss": 1.7709,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.885998725891113,
+      "learning_rate": 4.62e-05,
+      "loss": 1.7069,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 7.63657283782959,
+      "learning_rate": 4.68e-05,
+      "loss": 1.6549,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.502758979797363,
+      "learning_rate": 4.7399999999999993e-05,
+      "loss": 1.5249,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 7.67425537109375,
+      "learning_rate": 4.7999999999999994e-05,
+      "loss": 1.6269,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.933468818664551,
+      "learning_rate": 4.8599999999999995e-05,
+      "loss": 1.535,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.353681564331055,
+      "learning_rate": 4.9199999999999997e-05,
+      "loss": 1.2357,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 8.016544342041016,
+      "learning_rate": 4.98e-05,
+      "loss": 1.2392,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.43305778503418,
+      "learning_rate": 5.04e-05,
+      "loss": 1.5339,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 8.496820449829102,
+      "learning_rate": 5.1e-05,
+      "loss": 1.4414,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 12.498025894165039,
+      "learning_rate": 5.1599999999999994e-05,
+      "loss": 1.0707,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.985462188720703,
+      "learning_rate": 5.2199999999999995e-05,
+      "loss": 1.5268,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 10.915468215942383,
+      "learning_rate": 5.279999999999999e-05,
+      "loss": 1.4841,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 12.632097244262695,
+      "learning_rate": 5.339999999999999e-05,
+      "loss": 2.197,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 11.778656959533691,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 1.1379,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 11.089698791503906,
+      "learning_rate": 5.459999999999999e-05,
+      "loss": 1.3323,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 8.403290748596191,
+      "learning_rate": 5.519999999999999e-05,
+      "loss": 1.0058,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 12.170100212097168,
+      "learning_rate": 5.5799999999999994e-05,
+      "loss": 1.3709,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 13.481610298156738,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 1.5149,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 8.465336799621582,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 1.1589,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 8.697178840637207,
+      "learning_rate": 5.76e-05,
+      "loss": 1.338,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 1.2086584568023682,
+      "eval_runtime": 51.6602,
+      "eval_samples_per_second": 41.967,
+      "eval_steps_per_second": 10.492,
+      "eval_wer": 0.8778758000345961,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 9.466093063354492,
+      "learning_rate": 5.82e-05,
+      "loss": 1.9736,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.502716064453125,
+      "learning_rate": 5.88e-05,
+      "loss": 1.2357,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.822862148284912,
+      "learning_rate": 5.94e-05,
+      "loss": 0.969,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.4939680099487305,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 0.8472,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.335927963256836,
+      "learning_rate": 6.0599999999999996e-05,
+      "loss": 1.232,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.711673736572266,
+      "learning_rate": 6.12e-05,
+      "loss": 1.0268,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.31601619720459,
+      "learning_rate": 6.18e-05,
+      "loss": 0.9878,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 7.08448600769043,
+      "learning_rate": 6.239999999999999e-05,
+      "loss": 1.2664,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.570302486419678,
+      "learning_rate": 6.299999999999999e-05,
+      "loss": 0.5733,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 44.946353912353516,
+      "learning_rate": 6.359999999999999e-05,
+      "loss": 1.6091,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 14.037195205688477,
+      "learning_rate": 6.419999999999999e-05,
+      "loss": 1.5667,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 9.980315208435059,
+      "learning_rate": 6.479999999999999e-05,
+      "loss": 1.8256,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 13.4607572555542,
+      "learning_rate": 6.539999999999999e-05,
+      "loss": 1.2149,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 10.761940002441406,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 1.4465,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 8.355905532836914,
+      "learning_rate": 6.659999999999999e-05,
+      "loss": 0.873,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 8.082406997680664,
+      "learning_rate": 6.72e-05,
+      "loss": 0.8309,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 7.375432968139648,
+      "learning_rate": 6.78e-05,
+      "loss": 1.1447,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 13.326135635375977,
+      "learning_rate": 6.84e-05,
+      "loss": 0.9311,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 8.355464935302734,
+      "learning_rate": 6.9e-05,
+      "loss": 0.9014,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 16.993934631347656,
+      "learning_rate": 6.96e-05,
+      "loss": 1.152,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 7.030590057373047,
+      "learning_rate": 7.02e-05,
+      "loss": 0.6491,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 10.766969680786133,
+      "learning_rate": 7.079999999999999e-05,
+      "loss": 1.738,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 16.502492904663086,
+      "learning_rate": 7.139999999999999e-05,
+      "loss": 1.0539,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 10.595362663269043,
+      "learning_rate": 7.199999999999999e-05,
+      "loss": 1.0724,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 7.935666084289551,
+      "learning_rate": 7.259999999999999e-05,
+      "loss": 0.795,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 7.293322563171387,
+      "learning_rate": 7.319999999999999e-05,
+      "loss": 1.279,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 14.84443187713623,
+      "learning_rate": 7.379999999999999e-05,
+      "loss": 1.993,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 8.164588928222656,
+      "learning_rate": 7.439999999999999e-05,
+      "loss": 1.0304,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 7.177072048187256,
+      "learning_rate": 7.5e-05,
+      "loss": 1.1268,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 8.665708541870117,
+      "learning_rate": 7.56e-05,
+      "loss": 0.911,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 7.278268337249756,
+      "learning_rate": 7.62e-05,
+      "loss": 0.8871,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.9616780281066895,
+      "learning_rate": 7.68e-05,
+      "loss": 0.7107,
+      "step": 132
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 10.167497634887695,
+      "learning_rate": 7.74e-05,
+      "loss": 1.962,
+      "step": 133
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 17.286006927490234,
+      "learning_rate": 7.8e-05,
+      "loss": 2.3033,
+      "step": 134
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.155051231384277,
+      "learning_rate": 7.86e-05,
+      "loss": 0.7892,
+      "step": 135
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 13.249787330627441,
+      "learning_rate": 7.92e-05,
+      "loss": 0.6655,
+      "step": 136
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 13.192652702331543,
+      "learning_rate": 7.98e-05,
+      "loss": 0.9772,
+      "step": 137
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.216400623321533,
+      "learning_rate": 8.04e-05,
+      "loss": 1.4718,
+      "step": 138
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 9.961997985839844,
+      "learning_rate": 8.1e-05,
+      "loss": 1.1065,
+      "step": 139
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 9.072415351867676,
+      "learning_rate": 8.16e-05,
+      "loss": 1.6112,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 8.392301559448242,
+      "learning_rate": 8.22e-05,
+      "loss": 1.284,
+      "step": 141
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 7.51064920425415,
+      "learning_rate": 8.28e-05,
+      "loss": 1.2114,
+      "step": 142
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 19.81893539428711,
+      "learning_rate": 8.34e-05,
+      "loss": 0.909,
+      "step": 143
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 7.77775239944458,
+      "learning_rate": 8.4e-05,
+      "loss": 1.4994,
+      "step": 144
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.78767204284668,
+      "learning_rate": 8.459999999999998e-05,
+      "loss": 0.6961,
+      "step": 145
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 11.081228256225586,
+      "learning_rate": 8.519999999999998e-05,
+      "loss": 1.8639,
+      "step": 146
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 11.193565368652344,
+      "learning_rate": 8.579999999999998e-05,
+      "loss": 1.6451,
+      "step": 147
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 8.930937767028809,
+      "learning_rate": 8.639999999999999e-05,
+      "loss": 0.9557,
+      "step": 148
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 20.329538345336914,
+      "learning_rate": 8.699999999999999e-05,
+      "loss": 0.9914,
+      "step": 149
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": NaN,
+      "learning_rate": 8.699999999999999e-05,
+      "loss": 0.2003,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 8.920183181762695,
+      "learning_rate": 8.759999999999999e-05,
+      "loss": 2.1297,
+      "step": 151
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.7198567390441895,
+      "learning_rate": 8.819999999999999e-05,
+      "loss": 1.0552,
+      "step": 152
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.2832512855529785,
+      "learning_rate": 8.879999999999999e-05,
+      "loss": 1.0677,
+      "step": 153
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 7.0598530769348145,
+      "learning_rate": 8.939999999999999e-05,
+      "loss": 0.9668,
+      "step": 154
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 7.0460052490234375,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 0.917,
+      "step": 155
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.30267333984375,
+      "learning_rate": 9.059999999999999e-05,
+      "loss": 0.8791,
+      "step": 156
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 15.747892379760742,
+      "learning_rate": 9.12e-05,
+      "loss": 1.0952,
+      "step": 157
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.930239200592041,
+      "learning_rate": 9.18e-05,
+      "loss": 0.7907,
+      "step": 158
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 9.977339744567871,
+      "learning_rate": 9.24e-05,
+      "loss": 1.6181,
+      "step": 159
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.140440464019775,
+      "learning_rate": 9.3e-05,
+      "loss": 0.8012,
+      "step": 160
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 8.982709884643555,
+      "learning_rate": 9.36e-05,
+      "loss": 0.8466,
+      "step": 161
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.337733268737793,
+      "learning_rate": 9.419999999999999e-05,
+      "loss": 0.9134,
+      "step": 162
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.364322662353516,
+      "learning_rate": 9.479999999999999e-05,
+      "loss": 0.4464,
+      "step": 163
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 13.642416954040527,
+      "learning_rate": 9.539999999999999e-05,
+      "loss": 1.9781,
+      "step": 164
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.412764549255371,
+      "learning_rate": 9.599999999999999e-05,
+      "loss": 0.7324,
+      "step": 165
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 7.758469581604004,
+      "learning_rate": 9.659999999999999e-05,
+      "loss": 1.5962,
+      "step": 166
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.5638041496276855,
+      "learning_rate": 9.719999999999999e-05,
+      "loss": 1.0662,
+      "step": 167
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.04262638092041,
+      "learning_rate": 9.779999999999999e-05,
+      "loss": 1.106,
+      "step": 168
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.401612758636475,
+      "learning_rate": 9.839999999999999e-05,
+      "loss": 1.0118,
+      "step": 169
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 7.277603626251221,
+      "learning_rate": 9.9e-05,
+      "loss": 1.5725,
+      "step": 170
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 7.579121112823486,
+      "learning_rate": 9.96e-05,
+      "loss": 1.0403,
+      "step": 171
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 16.09749412536621,
+      "learning_rate": 0.0001002,
+      "loss": 1.1833,
+      "step": 172
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 10.291646957397461,
+      "learning_rate": 0.0001008,
+      "loss": 2.9008,
+      "step": 173
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 7.372079372406006,
+      "learning_rate": 0.0001014,
+      "loss": 0.779,
+      "step": 174
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 15.55083179473877,
+      "learning_rate": 0.000102,
+      "loss": 1.0366,
+      "step": 175
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 10.298460006713867,
+      "learning_rate": 0.0001026,
+      "loss": 1.7753,
+      "step": 176
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 7.792703151702881,
+      "learning_rate": 0.00010319999999999999,
+      "loss": 1.3867,
+      "step": 177
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.284561634063721,
+      "learning_rate": 0.00010379999999999999,
+      "loss": 1.2554,
+      "step": 178
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 8.849788665771484,
+      "learning_rate": 0.00010439999999999999,
+      "loss": 1.087,
+      "step": 179
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.7863664627075195,
+      "learning_rate": 0.00010499999999999999,
+      "loss": 1.003,
+      "step": 180
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.806325912475586,
+      "learning_rate": 0.00010559999999999998,
+      "loss": 0.7003,
+      "step": 181
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 8.477960586547852,
+      "learning_rate": 0.00010619999999999998,
+      "loss": 0.9081,
+      "step": 182
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 7.785075664520264,
+      "learning_rate": 0.00010679999999999998,
+      "loss": 1.1227,
+      "step": 183
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.257397174835205,
+      "learning_rate": 0.00010739999999999998,
+      "loss": 0.8996,
+      "step": 184
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 8.024823188781738,
+      "learning_rate": 0.00010799999999999998,
+      "loss": 1.2054,
+      "step": 185
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 7.594734191894531,
+      "learning_rate": 0.00010859999999999998,
+      "loss": 1.2464,
+      "step": 186
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 9.327756881713867,
+      "learning_rate": 0.00010919999999999998,
+      "loss": 1.1473,
+      "step": 187
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 15.395828247070312,
+      "learning_rate": 0.00010979999999999999,
+      "loss": 1.7228,
+      "step": 188
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 8.292728424072266,
+      "learning_rate": 0.00011039999999999999,
+      "loss": 1.3052,
+      "step": 189
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 6.423582077026367,
+      "learning_rate": 0.00011099999999999999,
+      "loss": 0.7601,
+      "step": 190
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 8.61807918548584,
+      "learning_rate": 0.00011159999999999999,
+      "loss": 0.6694,
+      "step": 191
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 10.768646240234375,
+      "learning_rate": 0.00011219999999999999,
+      "loss": 1.2744,
+      "step": 192
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 6.7909836769104,
+      "learning_rate": 0.00011279999999999999,
+      "loss": 0.8817,
+      "step": 193
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 9.817007064819336,
+      "learning_rate": 0.00011339999999999999,
+      "loss": 1.3541,
+      "step": 194
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 6.27230978012085,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 0.5326,
+      "step": 195
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 9.24206256866455,
+      "learning_rate": 0.0001146,
+      "loss": 0.8079,
+      "step": 196
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 7.788369655609131,
+      "learning_rate": 0.0001152,
+      "loss": 0.6113,
+      "step": 197
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 16.040983200073242,
+      "learning_rate": 0.0001158,
+      "loss": 1.1879,
+      "step": 198
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 8.447440147399902,
+      "learning_rate": 0.0001164,
+      "loss": 1.5101,
+      "step": 199
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 12.108735084533691,
+      "learning_rate": 0.000117,
+      "loss": 1.9649,
+      "step": 200
+    },
+    {
+      "epoch": 0.07,
+      "eval_loss": 1.0153309106826782,
+      "eval_runtime": 50.5715,
+      "eval_samples_per_second": 42.87,
+      "eval_steps_per_second": 10.718,
+      "eval_wer": 0.805137519460301,
+      "step": 200
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 8.759153366088867,
+      "learning_rate": 0.0001176,
+      "loss": 1.3547,
+      "step": 201
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 6.493929386138916,
+      "learning_rate": 0.0001182,
+      "loss": 2.0605,
+      "step": 202
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 6.16959285736084,
+      "learning_rate": 0.0001188,
+      "loss": 1.6697,
+      "step": 203
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 6.1440348625183105,
+      "learning_rate": 0.0001194,
+      "loss": 1.4012,
+      "step": 204
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 8.175420761108398,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 1.164,
+      "step": 205
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 7.481700420379639,
+      "learning_rate": 0.00012059999999999999,
+      "loss": 1.0755,
+      "step": 206
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.793044567108154,
+      "learning_rate": 0.00012119999999999999,
+      "loss": 0.8223,
+      "step": 207
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 7.168550968170166,
+      "learning_rate": 0.00012179999999999999,
+      "loss": 1.309,
+      "step": 208
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 9.77776050567627,
+      "learning_rate": 0.0001224,
+      "loss": 1.1608,
+      "step": 209
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 6.302064418792725,
+      "learning_rate": 0.00012299999999999998,
+      "loss": 0.7877,
+      "step": 210
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.7911152839660645,
+      "learning_rate": 0.0001236,
+      "loss": 0.7164,
+      "step": 211
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 6.755216121673584,
+      "learning_rate": 0.00012419999999999998,
+      "loss": 1.3267,
+      "step": 212
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.61589241027832,
+      "learning_rate": 0.00012479999999999997,
+      "loss": 1.1473,
+      "step": 213
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 6.926543712615967,
+      "learning_rate": 0.00012539999999999999,
+      "loss": 1.4776,
+      "step": 214
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 30.6265869140625,
+      "learning_rate": 0.00012599999999999997,
+      "loss": 1.4803,
+      "step": 215
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 10.334158897399902,
+      "learning_rate": 0.0001266,
+      "loss": 1.579,
+      "step": 216
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 8.379430770874023,
+      "learning_rate": 0.00012719999999999997,
+      "loss": 1.0805,
+      "step": 217
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.460320472717285,
+      "learning_rate": 0.0001278,
+      "loss": 0.7705,
+      "step": 218
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 7.094217777252197,
+      "learning_rate": 0.00012839999999999998,
+      "loss": 1.061,
+      "step": 219
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 7.368185043334961,
+      "learning_rate": 0.000129,
+      "loss": 1.1021,
+      "step": 220
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 8.67043685913086,
+      "learning_rate": 0.00012959999999999998,
+      "loss": 1.84,
+      "step": 221
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 9.372278213500977,
+      "learning_rate": 0.0001302,
+      "loss": 1.2919,
+      "step": 222
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 10.148419380187988,
+      "learning_rate": 0.00013079999999999998,
+      "loss": 1.3467,
+      "step": 223
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 9.848054885864258,
+      "learning_rate": 0.0001314,
+      "loss": 1.018,
+      "step": 224
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 13.819363594055176,
+      "learning_rate": 0.00013199999999999998,
+      "loss": 1.8879,
+      "step": 225
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 8.070960998535156,
+      "learning_rate": 0.0001326,
+      "loss": 2.0939,
+      "step": 226
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 7.248965263366699,
+      "learning_rate": 0.00013319999999999999,
+      "loss": 1.4703,
+      "step": 227
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.5253472328186035,
+      "learning_rate": 0.0001338,
+      "loss": 1.6318,
+      "step": 228
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 8.237483024597168,
+      "learning_rate": 0.0001344,
+      "loss": 1.053,
+      "step": 229
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.78982400894165,
+      "learning_rate": 0.000135,
+      "loss": 1.1394,
+      "step": 230
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 7.947813510894775,
+      "learning_rate": 0.0001356,
+      "loss": 1.0707,
+      "step": 231
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.178483486175537,
+      "learning_rate": 0.0001362,
+      "loss": 1.246,
+      "step": 232
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.992203712463379,
+      "learning_rate": 0.0001368,
+      "loss": 0.8734,
+      "step": 233
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.598153114318848,
+      "learning_rate": 0.0001374,
+      "loss": 1.5,
+      "step": 234
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.893948078155518,
+      "learning_rate": 0.000138,
+      "loss": 0.902,
+      "step": 235
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": NaN,
+      "learning_rate": 0.000138,
+      "loss": 2.4242,
+      "step": 236
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 10.769316673278809,
+      "learning_rate": 0.0001386,
+      "loss": 1.7384,
+      "step": 237
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 8.3101167678833,
+      "learning_rate": 0.0001392,
+      "loss": 0.9501,
+      "step": 238
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 7.548043251037598,
+      "learning_rate": 0.00013979999999999998,
+      "loss": 0.9354,
+      "step": 239
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 7.125143527984619,
+      "learning_rate": 0.0001404,
+      "loss": 0.5763,
+      "step": 240
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.7880330085754395,
+      "learning_rate": 0.00014099999999999998,
+      "loss": 0.7926,
+      "step": 241
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.474691867828369,
+      "learning_rate": 0.00014159999999999997,
+      "loss": 0.6569,
+      "step": 242
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.943902492523193,
+      "learning_rate": 0.0001422,
+      "loss": 0.5578,
+      "step": 243
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.652454853057861,
+      "learning_rate": 0.00014279999999999997,
+      "loss": 1.0422,
+      "step": 244
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.662099838256836,
+      "learning_rate": 0.0001434,
+      "loss": 0.9684,
+      "step": 245
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 8.081048965454102,
+      "learning_rate": 0.00014399999999999998,
+      "loss": 1.0536,
+      "step": 246
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.202698230743408,
+      "learning_rate": 0.0001446,
+      "loss": 0.9664,
+      "step": 247
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.779542922973633,
+      "learning_rate": 0.00014519999999999998,
+      "loss": 1.0268,
+      "step": 248
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 11.181925773620605,
+      "learning_rate": 0.0001458,
+      "loss": 1.184,
+      "step": 249
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 11.750948905944824,
+      "learning_rate": 0.00014639999999999998,
+      "loss": 1.8029,
+      "step": 250
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 14.2088041305542,
+      "learning_rate": 0.000147,
+      "loss": 2.2772,
+      "step": 251
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 7.257857322692871,
+      "learning_rate": 0.00014759999999999998,
+      "loss": 1.2177,
+      "step": 252
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.847878456115723,
+      "learning_rate": 0.0001482,
+      "loss": 1.4445,
+      "step": 253
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.302583694458008,
+      "learning_rate": 0.00014879999999999998,
+      "loss": 1.1606,
+      "step": 254
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.7541680335998535,
+      "learning_rate": 0.0001494,
+      "loss": 1.2038,
+      "step": 255
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.664234638214111,
+      "learning_rate": 0.00015,
+      "loss": 1.3001,
+      "step": 256
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.489469051361084,
+      "learning_rate": 0.00015059999999999997,
+      "loss": 0.766,
+      "step": 257
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 9.820629119873047,
+      "learning_rate": 0.0001512,
+      "loss": 0.6067,
+      "step": 258
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.636170864105225,
+      "learning_rate": 0.00015179999999999998,
+      "loss": 0.9712,
+      "step": 259
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.902698993682861,
+      "learning_rate": 0.0001524,
+      "loss": 0.8656,
+      "step": 260
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 12.261871337890625,
+      "learning_rate": 0.00015299999999999998,
+      "loss": 1.3325,
+      "step": 261
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 10.92815113067627,
+      "learning_rate": 0.0001536,
+      "loss": 1.4591,
+      "step": 262
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.674909591674805,
+      "learning_rate": 0.00015419999999999998,
+      "loss": 0.5967,
+      "step": 263
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 7.12188196182251,
+      "learning_rate": 0.0001548,
+      "loss": 0.5938,
+      "step": 264
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 7.955028057098389,
+      "learning_rate": 0.00015539999999999998,
+      "loss": 0.7309,
+      "step": 265
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 8.135099411010742,
+      "learning_rate": 0.000156,
+      "loss": 0.9604,
+      "step": 266
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 17.383867263793945,
+      "learning_rate": 0.00015659999999999998,
+      "loss": 0.966,
+      "step": 267
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 7.752403736114502,
+      "learning_rate": 0.0001572,
+      "loss": 1.5528,
+      "step": 268
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 9.291197776794434,
+      "learning_rate": 0.0001578,
+      "loss": 0.7933,
+      "step": 269
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 7.345749855041504,
+      "learning_rate": 0.0001584,
+      "loss": 0.7673,
+      "step": 270
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.077545166015625,
+      "learning_rate": 0.000159,
+      "loss": 0.4076,
+      "step": 271
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 7.044876575469971,
+      "learning_rate": 0.0001596,
+      "loss": 2.2965,
+      "step": 272
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 8.072477340698242,
+      "learning_rate": 0.0001602,
+      "loss": 1.0095,
+      "step": 273
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 7.533301830291748,
+      "learning_rate": 0.0001608,
+      "loss": 1.6276,
+      "step": 274
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 10.649826049804688,
+      "learning_rate": 0.0001614,
+      "loss": 1.1563,
+      "step": 275
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.723114013671875,
+      "learning_rate": 0.000162,
+      "loss": 1.5564,
+      "step": 276
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.508934497833252,
+      "learning_rate": 0.0001626,
+      "loss": 1.1316,
+      "step": 277
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.111855983734131,
+      "learning_rate": 0.0001632,
+      "loss": 1.0925,
+      "step": 278
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 7.4623613357543945,
+      "learning_rate": 0.0001638,
+      "loss": 0.7827,
+      "step": 279
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.680227756500244,
+      "learning_rate": 0.0001644,
+      "loss": 0.9112,
+      "step": 280
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 11.717487335205078,
+      "learning_rate": 0.000165,
+      "loss": 1.1662,
+      "step": 281
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.690805912017822,
+      "learning_rate": 0.0001656,
+      "loss": 1.0872,
+      "step": 282
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 8.276322364807129,
+      "learning_rate": 0.0001662,
+      "loss": 1.0738,
+      "step": 283
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.989705562591553,
+      "learning_rate": 0.0001668,
+      "loss": 1.2949,
+      "step": 284
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.663930416107178,
+      "learning_rate": 0.0001674,
+      "loss": 0.5814,
+      "step": 285
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.9047698974609375,
+      "learning_rate": 0.000168,
+      "loss": 1.0676,
+      "step": 286
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 7.110291957855225,
+      "learning_rate": 0.0001686,
+      "loss": 1.227,
+      "step": 287
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 8.15951156616211,
+      "learning_rate": 0.00016919999999999997,
+      "loss": 0.9983,
+      "step": 288
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 8.399914741516113,
+      "learning_rate": 0.00016979999999999998,
+      "loss": 0.4559,
+      "step": 289
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 7.747352123260498,
+      "learning_rate": 0.00017039999999999997,
+      "loss": 1.0402,
+      "step": 290
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.47703218460083,
+      "learning_rate": 0.00017099999999999998,
+      "loss": 1.167,
+      "step": 291
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.136383533477783,
+      "learning_rate": 0.00017159999999999997,
+      "loss": 0.7194,
+      "step": 292
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.892377853393555,
+      "learning_rate": 0.00017219999999999998,
+      "loss": 0.9781,
+      "step": 293
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 14.462456703186035,
+      "learning_rate": 0.00017279999999999997,
+      "loss": 0.7622,
+      "step": 294
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 7.86848783493042,
+      "learning_rate": 0.00017339999999999996,
+      "loss": 1.2177,
+      "step": 295
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 7.074151992797852,
+      "learning_rate": 0.00017399999999999997,
+      "loss": 1.0429,
+      "step": 296
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.4072675704956055,
+      "learning_rate": 0.00017459999999999996,
+      "loss": 0.6744,
+      "step": 297
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 12.912552833557129,
+      "learning_rate": 0.00017519999999999998,
+      "loss": 0.888,
+      "step": 298
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 10.346461296081543,
+      "learning_rate": 0.00017579999999999996,
+      "loss": 0.7609,
+      "step": 299
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 8.770968437194824,
+      "learning_rate": 0.00017639999999999998,
+      "loss": 0.905,
+      "step": 300
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 1.0725961923599243,
+      "eval_runtime": 51.0006,
+      "eval_samples_per_second": 42.509,
+      "eval_steps_per_second": 10.627,
+      "eval_wer": 0.8373983739837398,
+      "step": 300
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.282247066497803,
+      "learning_rate": 0.00017699999999999997,
+      "loss": 1.1397,
+      "step": 301
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.363534927368164,
+      "learning_rate": 0.00017759999999999998,
+      "loss": 0.9052,
+      "step": 302
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 7.095592021942139,
+      "learning_rate": 0.00017819999999999997,
+      "loss": 0.7081,
+      "step": 303
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.591117858886719,
+      "learning_rate": 0.00017879999999999998,
+      "loss": 1.198,
+      "step": 304
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 7.835860252380371,
+      "learning_rate": 0.00017939999999999997,
+      "loss": 1.4741,
+      "step": 305
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.496611595153809,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 0.7443,
+      "step": 306
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.134020805358887,
+      "learning_rate": 0.00018059999999999997,
+      "loss": 0.6569,
+      "step": 307
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.844152450561523,
+      "learning_rate": 0.00018119999999999999,
+      "loss": 1.3045,
+      "step": 308
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.551060676574707,
+      "learning_rate": 0.00018179999999999997,
+      "loss": 0.564,
+      "step": 309
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.43617582321167,
+      "learning_rate": 0.0001824,
+      "loss": 1.0049,
+      "step": 310
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.678737640380859,
+      "learning_rate": 0.00018299999999999998,
+      "loss": 0.5783,
+      "step": 311
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.169671058654785,
+      "learning_rate": 0.0001836,
+      "loss": 0.6863,
+      "step": 312
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 8.256979942321777,
+      "learning_rate": 0.00018419999999999998,
+      "loss": 1.0964,
+      "step": 313
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 8.771745681762695,
+      "learning_rate": 0.0001848,
+      "loss": 1.2424,
+      "step": 314
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 7.281775951385498,
+      "learning_rate": 0.00018539999999999998,
+      "loss": 2.047,
+      "step": 315
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.94806432723999,
+      "learning_rate": 0.000186,
+      "loss": 1.0021,
+      "step": 316
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 11.512472152709961,
+      "learning_rate": 0.00018659999999999998,
+      "loss": 1.3445,
+      "step": 317
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.457605838775635,
+      "learning_rate": 0.0001872,
+      "loss": 0.9049,
+      "step": 318
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 7.318243026733398,
+      "learning_rate": 0.00018779999999999998,
+      "loss": 1.0203,
+      "step": 319
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.3816118240356445,
+      "learning_rate": 0.00018839999999999997,
+      "loss": 1.0586,
+      "step": 320
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.394790172576904,
+      "learning_rate": 0.00018899999999999999,
+      "loss": 0.9951,
+      "step": 321
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 8.764740943908691,
+      "learning_rate": 0.00018959999999999997,
+      "loss": 0.9491,
+      "step": 322
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 7.136832237243652,
+      "learning_rate": 0.0001902,
+      "loss": 1.068,
+      "step": 323
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 18.01349639892578,
+      "learning_rate": 0.00019079999999999998,
+      "loss": 1.7603,
+      "step": 324
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 7.3441386222839355,
+      "learning_rate": 0.0001914,
+      "loss": 0.8072,
+      "step": 325
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 7.931651592254639,
+      "learning_rate": 0.00019199999999999998,
+      "loss": 1.7458,
+      "step": 326
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.381895542144775,
+      "learning_rate": 0.0001926,
+      "loss": 1.2449,
+      "step": 327
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.183620452880859,
+      "learning_rate": 0.00019319999999999998,
+      "loss": 0.9892,
+      "step": 328
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.490755558013916,
+      "learning_rate": 0.0001938,
+      "loss": 1.1006,
+      "step": 329
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.9131011962890625,
+      "learning_rate": 0.00019439999999999998,
+      "loss": 1.2143,
+      "step": 330
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.422755718231201,
+      "learning_rate": 0.000195,
+      "loss": 1.1383,
+      "step": 331
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.351710796356201,
+      "learning_rate": 0.00019559999999999998,
+      "loss": 0.9472,
+      "step": 332
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.6042375564575195,
+      "learning_rate": 0.0001962,
+      "loss": 0.5255,
+      "step": 333
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.7591094970703125,
+      "learning_rate": 0.00019679999999999999,
+      "loss": 1.0898,
+      "step": 334
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 17.369613647460938,
+      "learning_rate": 0.0001974,
+      "loss": 0.8401,
+      "step": 335
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 8.167240142822266,
+      "learning_rate": 0.000198,
+      "loss": 1.9533,
+      "step": 336
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 7.36649227142334,
+      "learning_rate": 0.0001986,
+      "loss": 1.0694,
+      "step": 337
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 18.003772735595703,
+      "learning_rate": 0.0001992,
+      "loss": 0.8262,
+      "step": 338
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 7.110168933868408,
+      "learning_rate": 0.0001998,
+      "loss": 1.0649,
+      "step": 339
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 11.043972969055176,
+      "learning_rate": 0.0002004,
+      "loss": 0.9792,
+      "step": 340
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.065829753875732,
+      "learning_rate": 0.000201,
+      "loss": 0.5039,
+      "step": 341
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 8.094096183776855,
+      "learning_rate": 0.0002016,
+      "loss": 0.9588,
+      "step": 342
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 9.9978609085083,
+      "learning_rate": 0.0002022,
+      "loss": 1.2725,
+      "step": 343
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 13.194984436035156,
+      "learning_rate": 0.0002028,
+      "loss": 1.3617,
+      "step": 344
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 8.760825157165527,
+      "learning_rate": 0.00020339999999999998,
+      "loss": 1.0096,
+      "step": 345
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 13.242958068847656,
+      "learning_rate": 0.000204,
+      "loss": 1.2532,
+      "step": 346
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 10.69446086883545,
+      "learning_rate": 0.00020459999999999999,
+      "loss": 0.9107,
+      "step": 347
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 14.089339256286621,
+      "learning_rate": 0.0002052,
+      "loss": 1.0958,
+      "step": 348
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 10.158014297485352,
+      "learning_rate": 0.0002058,
+      "loss": 1.3443,
+      "step": 349
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 277.1723937988281,
+      "learning_rate": 0.00020639999999999998,
+      "loss": 1.2569,
+      "step": 350
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 15.648301124572754,
+      "learning_rate": 0.00020699999999999996,
+      "loss": 2.473,
+      "step": 351
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.679229259490967,
+      "learning_rate": 0.00020759999999999998,
+      "loss": 1.3066,
+      "step": 352
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 6.558560371398926,
+      "learning_rate": 0.00020819999999999996,
+      "loss": 1.0691,
+      "step": 353
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.9980814456939697,
+      "learning_rate": 0.00020879999999999998,
+      "loss": 0.9881,
+      "step": 354
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.269549369812012,
+      "learning_rate": 0.00020939999999999997,
+      "loss": 0.9437,
+      "step": 355
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.434672832489014,
+      "learning_rate": 0.00020999999999999998,
+      "loss": 1.139,
+      "step": 356
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.682892084121704,
+      "learning_rate": 0.00021059999999999997,
+      "loss": 0.9274,
+      "step": 357
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.203353404998779,
+      "learning_rate": 0.00021119999999999996,
+      "loss": 0.9735,
+      "step": 358
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.374783039093018,
+      "learning_rate": 0.00021179999999999997,
+      "loss": 0.7179,
+      "step": 359
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 6.496840476989746,
+      "learning_rate": 0.00021239999999999996,
+      "loss": 0.6335,
+      "step": 360
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.820166110992432,
+      "learning_rate": 0.00021299999999999997,
+      "loss": 0.7724,
+      "step": 361
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.941391468048096,
+      "learning_rate": 0.00021359999999999996,
+      "loss": 0.6864,
+      "step": 362
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.761062145233154,
+      "learning_rate": 0.00021419999999999998,
+      "loss": 1.3754,
+      "step": 363
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 8.317700386047363,
+      "learning_rate": 0.00021479999999999996,
+      "loss": 1.2307,
+      "step": 364
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 7.029150009155273,
+      "learning_rate": 0.00021539999999999998,
+      "loss": 0.7201,
+      "step": 365
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 11.40591812133789,
+      "learning_rate": 0.00021599999999999996,
+      "loss": 1.1349,
+      "step": 366
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 7.807318687438965,
+      "learning_rate": 0.00021659999999999998,
+      "loss": 1.4664,
+      "step": 367
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 7.696564197540283,
+      "learning_rate": 0.00021719999999999997,
+      "loss": 0.9169,
+      "step": 368
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 6.1474151611328125,
+      "learning_rate": 0.00021779999999999998,
+      "loss": 1.0351,
+      "step": 369
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.485285758972168,
+      "learning_rate": 0.00021839999999999997,
+      "loss": 0.5278,
+      "step": 370
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.146036624908447,
+      "learning_rate": 0.00021899999999999998,
+      "loss": 0.6379,
+      "step": 371
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 7.9090800285339355,
+      "learning_rate": 0.00021959999999999997,
+      "loss": 1.071,
+      "step": 372
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 9.113584518432617,
+      "learning_rate": 0.00022019999999999999,
+      "loss": 0.8886,
+      "step": 373
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 6.609561920166016,
+      "learning_rate": 0.00022079999999999997,
+      "loss": 0.7262,
+      "step": 374
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 8.190186500549316,
+      "learning_rate": 0.0002214,
+      "loss": 0.7478,
+      "step": 375
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 6.347649097442627,
+      "learning_rate": 0.00022199999999999998,
+      "loss": 1.5713,
+      "step": 376
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 6.180627346038818,
+      "learning_rate": 0.0002226,
+      "loss": 1.8954,
+      "step": 377
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.399271488189697,
+      "learning_rate": 0.00022319999999999998,
+      "loss": 1.382,
+      "step": 378
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 6.500628471374512,
+      "learning_rate": 0.0002238,
+      "loss": 1.1642,
+      "step": 379
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 8.18594741821289,
+      "learning_rate": 0.00022439999999999998,
+      "loss": 1.3459,
+      "step": 380
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.980257272720337,
+      "learning_rate": 0.000225,
+      "loss": 0.6223,
+      "step": 381
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.482236385345459,
+      "learning_rate": 0.00022559999999999998,
+      "loss": 1.2836,
+      "step": 382
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.334828853607178,
+      "learning_rate": 0.00022619999999999997,
+      "loss": 0.9142,
+      "step": 383
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.967797040939331,
+      "learning_rate": 0.00022679999999999998,
+      "loss": 0.5278,
+      "step": 384
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.172309398651123,
+      "learning_rate": 0.00022739999999999997,
+      "loss": 0.6968,
+      "step": 385
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.380971670150757,
+      "learning_rate": 0.00022799999999999999,
+      "loss": 0.6464,
+      "step": 386
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 7.23440408706665,
+      "learning_rate": 0.00022859999999999997,
+      "loss": 1.8373,
+      "step": 387
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 6.774138450622559,
+      "learning_rate": 0.0002292,
+      "loss": 1.0644,
+      "step": 388
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 8.141240119934082,
+      "learning_rate": 0.00022979999999999997,
+      "loss": 1.4621,
+      "step": 389
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 8.437649726867676,
+      "learning_rate": 0.0002304,
+      "loss": 1.0165,
+      "step": 390
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 6.1060791015625,
+      "learning_rate": 0.00023099999999999998,
+      "loss": 1.5574,
+      "step": 391
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 6.36237907409668,
+      "learning_rate": 0.0002316,
+      "loss": 0.8218,
+      "step": 392
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 7.044564247131348,
+      "learning_rate": 0.00023219999999999998,
+      "loss": 1.0601,
+      "step": 393
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.057678699493408,
+      "learning_rate": 0.0002328,
+      "loss": 0.8001,
+      "step": 394
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 6.342691421508789,
+      "learning_rate": 0.00023339999999999998,
+      "loss": 0.5296,
+      "step": 395
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 7.52527379989624,
+      "learning_rate": 0.000234,
+      "loss": 0.7155,
+      "step": 396
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 8.385841369628906,
+      "learning_rate": 0.00023459999999999998,
+      "loss": 0.7998,
+      "step": 397
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 8.12092399597168,
+      "learning_rate": 0.0002352,
+      "loss": 1.0234,
+      "step": 398
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 14.459668159484863,
+      "learning_rate": 0.00023579999999999999,
+      "loss": 0.457,
+      "step": 399
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 9.308672904968262,
+      "learning_rate": 0.0002364,
+      "loss": 2.0688,
+      "step": 400
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 1.1829124689102173,
+      "eval_runtime": 50.8154,
+      "eval_samples_per_second": 42.664,
+      "eval_steps_per_second": 10.666,
+      "eval_wer": 0.8181975436775645,
+      "step": 400
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.8363261222839355,
+      "learning_rate": 0.000237,
+      "loss": 1.5778,
+      "step": 401
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.74521541595459,
+      "learning_rate": 0.0002376,
+      "loss": 0.9614,
+      "step": 402
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.534332752227783,
+      "learning_rate": 0.0002382,
+      "loss": 0.8923,
+      "step": 403
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.4116997718811035,
+      "learning_rate": 0.0002388,
+      "loss": 1.2341,
+      "step": 404
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 6.503213882446289,
+      "learning_rate": 0.0002394,
+      "loss": 1.2941,
+      "step": 405
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 7.444057941436768,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 0.8984,
+      "step": 406
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.511045932769775,
+      "learning_rate": 0.0002406,
+      "loss": 0.7948,
+      "step": 407
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 7.183623313903809,
+      "learning_rate": 0.00024119999999999998,
+      "loss": 1.095,
+      "step": 408
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 6.442593574523926,
+      "learning_rate": 0.0002418,
+      "loss": 1.6099,
+      "step": 409
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.9881575107574463,
+      "learning_rate": 0.00024239999999999998,
+      "loss": 0.8797,
+      "step": 410
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 10.62885570526123,
+      "learning_rate": 0.000243,
+      "loss": 1.0199,
+      "step": 411
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 7.63018798828125,
+      "learning_rate": 0.00024359999999999999,
+      "loss": 0.9954,
+      "step": 412
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 6.816760540008545,
+      "learning_rate": 0.00024419999999999997,
+      "loss": 0.9977,
+      "step": 413
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 15.825918197631836,
+      "learning_rate": 0.0002448,
+      "loss": 3.4223,
+      "step": 414
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.994204998016357,
+      "learning_rate": 0.00024539999999999995,
+      "loss": 1.4219,
+      "step": 415
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 7.240597248077393,
+      "learning_rate": 0.00024599999999999996,
+      "loss": 1.4543,
+      "step": 416
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.940434455871582,
+      "learning_rate": 0.0002466,
+      "loss": 0.6018,
+      "step": 417
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 10.660022735595703,
+      "learning_rate": 0.0002472,
+      "loss": 1.586,
+      "step": 418
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 6.618219375610352,
+      "learning_rate": 0.00024779999999999995,
+      "loss": 1.0557,
+      "step": 419
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.79208517074585,
+      "learning_rate": 0.00024839999999999997,
+      "loss": 0.8003,
+      "step": 420
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.461696147918701,
+      "learning_rate": 0.000249,
+      "loss": 1.1184,
+      "step": 421
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.876696586608887,
+      "learning_rate": 0.00024959999999999994,
+      "loss": 0.8481,
+      "step": 422
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 12.002443313598633,
+      "learning_rate": 0.00025019999999999996,
+      "loss": 1.8712,
+      "step": 423
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 6.414361476898193,
+      "learning_rate": 0.00025079999999999997,
+      "loss": 1.0893,
+      "step": 424
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": NaN,
+      "learning_rate": 0.00025079999999999997,
+      "loss": 0.2905,
+      "step": 425
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.970865249633789,
+      "learning_rate": 0.0002514,
+      "loss": 1.5082,
+      "step": 426
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.599332332611084,
+      "learning_rate": 0.00025199999999999995,
+      "loss": 1.8596,
+      "step": 427
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.235030651092529,
+      "learning_rate": 0.00025259999999999996,
+      "loss": 1.3534,
+      "step": 428
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.113259792327881,
+      "learning_rate": 0.0002532,
+      "loss": 1.1064,
+      "step": 429
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.495222091674805,
+      "learning_rate": 0.0002538,
+      "loss": 1.2008,
+      "step": 430
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.514638900756836,
+      "learning_rate": 0.00025439999999999995,
+      "loss": 1.0261,
+      "step": 431
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.7902960777282715,
+      "learning_rate": 0.00025499999999999996,
+      "loss": 1.0724,
+      "step": 432
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 8.044351577758789,
+      "learning_rate": 0.0002556,
+      "loss": 1.1708,
+      "step": 433
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.15432071685791,
+      "learning_rate": 0.0002562,
+      "loss": 1.1937,
+      "step": 434
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 7.8752593994140625,
+      "learning_rate": 0.00025679999999999995,
+      "loss": 0.8926,
+      "step": 435
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 11.0122652053833,
+      "learning_rate": 0.00025739999999999997,
+      "loss": 0.9854,
+      "step": 436
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 6.583104133605957,
+      "learning_rate": 0.000258,
+      "loss": 1.3097,
+      "step": 437
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.033292770385742,
+      "learning_rate": 0.0002586,
+      "loss": 1.1236,
+      "step": 438
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 8.683174133300781,
+      "learning_rate": 0.00025919999999999996,
+      "loss": 1.2291,
+      "step": 439
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.946511268615723,
+      "learning_rate": 0.00025979999999999997,
+      "loss": 0.9983,
+      "step": 440
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 12.835111618041992,
+      "learning_rate": 0.0002604,
+      "loss": 1.512,
+      "step": 441
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 8.656193733215332,
+      "learning_rate": 0.000261,
+      "loss": 0.9368,
+      "step": 442
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 7.563745498657227,
+      "learning_rate": 0.00026159999999999996,
+      "loss": 1.2435,
+      "step": 443
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 6.136525630950928,
+      "learning_rate": 0.0002622,
+      "loss": 0.4889,
+      "step": 444
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.84512996673584,
+      "learning_rate": 0.0002628,
+      "loss": 0.8539,
+      "step": 445
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 6.192235946655273,
+      "learning_rate": 0.00026339999999999995,
+      "loss": 1.1632,
+      "step": 446
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 9.75316333770752,
+      "learning_rate": 0.00026399999999999997,
+      "loss": 1.2679,
+      "step": 447
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.024916648864746,
+      "learning_rate": 0.0002646,
+      "loss": 1.1505,
+      "step": 448
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 7.302039623260498,
+      "learning_rate": 0.0002652,
+      "loss": 1.0232,
+      "step": 449
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.526060104370117,
+      "learning_rate": 0.00026579999999999996,
+      "loss": 0.9171,
+      "step": 450
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.972250938415527,
+      "learning_rate": 0.00026639999999999997,
+      "loss": 1.7666,
+      "step": 451
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.7761054039001465,
+      "learning_rate": 0.000267,
+      "loss": 1.3188,
+      "step": 452
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.936556339263916,
+      "learning_rate": 0.0002676,
+      "loss": 0.8506,
+      "step": 453
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.347964286804199,
+      "learning_rate": 0.00026819999999999996,
+      "loss": 1.0834,
+      "step": 454
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.9066038131713867,
+      "learning_rate": 0.0002688,
+      "loss": 1.0587,
+      "step": 455
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 10.331660270690918,
+      "learning_rate": 0.0002694,
+      "loss": 1.6772,
+      "step": 456
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 6.844241142272949,
+      "learning_rate": 0.00027,
+      "loss": 1.0441,
+      "step": 457
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 7.055009365081787,
+      "learning_rate": 0.00027059999999999996,
+      "loss": 1.4951,
+      "step": 458
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.321764945983887,
+      "learning_rate": 0.0002712,
+      "loss": 0.9027,
+      "step": 459
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 6.612902641296387,
+      "learning_rate": 0.0002718,
+      "loss": 0.9197,
+      "step": 460
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 7.686680316925049,
+      "learning_rate": 0.0002724,
+      "loss": 0.8149,
+      "step": 461
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.893856048583984,
+      "learning_rate": 0.00027299999999999997,
+      "loss": 0.7485,
+      "step": 462
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.316108703613281,
+      "learning_rate": 0.0002736,
+      "loss": 0.8778,
+      "step": 463
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 8.641250610351562,
+      "learning_rate": 0.0002742,
+      "loss": 1.4959,
+      "step": 464
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 6.9227447509765625,
+      "learning_rate": 0.0002748,
+      "loss": 0.9793,
+      "step": 465
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 17.23862075805664,
+      "learning_rate": 0.00027539999999999997,
+      "loss": 1.0452,
+      "step": 466
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 6.343098163604736,
+      "learning_rate": 0.000276,
+      "loss": 1.271,
+      "step": 467
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 9.62498664855957,
+      "learning_rate": 0.0002766,
+      "loss": 0.9764,
+      "step": 468
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 7.0901336669921875,
+      "learning_rate": 0.0002772,
+      "loss": 1.3291,
+      "step": 469
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 8.628947257995605,
+      "learning_rate": 0.0002778,
+      "loss": 1.2843,
+      "step": 470
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 6.520139217376709,
+      "learning_rate": 0.0002784,
+      "loss": 0.8886,
+      "step": 471
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 12.442567825317383,
+      "learning_rate": 0.000279,
+      "loss": 1.1537,
+      "step": 472
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 6.3325347900390625,
+      "learning_rate": 0.00027959999999999997,
+      "loss": 1.3338,
+      "step": 473
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 7.008511066436768,
+      "learning_rate": 0.0002802,
+      "loss": 1.2572,
+      "step": 474
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 10.975461959838867,
+      "learning_rate": 0.0002808,
+      "loss": 1.2421,
+      "step": 475
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 6.61250114440918,
+      "learning_rate": 0.00028139999999999996,
+      "loss": 2.1506,
+      "step": 476
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.0594916343688965,
+      "learning_rate": 0.00028199999999999997,
+      "loss": 1.5596,
+      "step": 477
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.184998989105225,
+      "learning_rate": 0.0002826,
+      "loss": 1.3744,
+      "step": 478
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.103187084197998,
+      "learning_rate": 0.00028319999999999994,
+      "loss": 0.8721,
+      "step": 479
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.862779140472412,
+      "learning_rate": 0.00028379999999999996,
+      "loss": 1.2812,
+      "step": 480
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.681532859802246,
+      "learning_rate": 0.0002844,
+      "loss": 0.7297,
+      "step": 481
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.8014369010925293,
+      "learning_rate": 0.000285,
+      "loss": 0.6603,
+      "step": 482
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 6.400700092315674,
+      "learning_rate": 0.00028559999999999995,
+      "loss": 0.8617,
+      "step": 483
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 7.437506675720215,
+      "learning_rate": 0.00028619999999999996,
+      "loss": 0.9853,
+      "step": 484
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 7.865484237670898,
+      "learning_rate": 0.0002868,
+      "loss": 1.0431,
+      "step": 485
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 11.700081825256348,
+      "learning_rate": 0.00028739999999999994,
+      "loss": 2.4413,
+      "step": 486
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 6.08707332611084,
+      "learning_rate": 0.00028799999999999995,
+      "loss": 1.2467,
+      "step": 487
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.360887050628662,
+      "learning_rate": 0.00028859999999999997,
+      "loss": 1.1013,
+      "step": 488
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.402688980102539,
+      "learning_rate": 0.0002892,
+      "loss": 1.5754,
+      "step": 489
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.761510848999023,
+      "learning_rate": 0.00028979999999999994,
+      "loss": 1.1612,
+      "step": 490
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.469000339508057,
+      "learning_rate": 0.00029039999999999996,
+      "loss": 0.9344,
+      "step": 491
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 8.59809684753418,
+      "learning_rate": 0.00029099999999999997,
+      "loss": 1.902,
+      "step": 492
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 10.263930320739746,
+      "learning_rate": 0.0002916,
+      "loss": 0.6227,
+      "step": 493
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.8753981590271,
+      "learning_rate": 0.00029219999999999995,
+      "loss": 0.4528,
+      "step": 494
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 6.5920562744140625,
+      "learning_rate": 0.00029279999999999996,
+      "loss": 0.6706,
+      "step": 495
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 10.204047203063965,
+      "learning_rate": 0.0002934,
+      "loss": 1.4762,
+      "step": 496
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 14.272616386413574,
+      "learning_rate": 0.000294,
+      "loss": 2.0472,
+      "step": 497
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 8.78574275970459,
+      "learning_rate": 0.00029459999999999995,
+      "loss": 1.101,
+      "step": 498
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 6.776339530944824,
+      "learning_rate": 0.00029519999999999997,
+      "loss": 1.1936,
+      "step": 499
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 6.81951379776001,
+      "learning_rate": 0.0002958,
+      "loss": 1.9781,
+      "step": 500
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 1.4941623210906982,
+      "eval_runtime": 50.4072,
+      "eval_samples_per_second": 43.01,
+      "eval_steps_per_second": 10.752,
+      "eval_wer": 0.8929250994637606,
+      "step": 500
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 8.40855884552002,
+      "learning_rate": 0.0002964,
+      "loss": 2.3229,
+      "step": 501
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 9.991337776184082,
+      "learning_rate": 0.00029699999999999996,
+      "loss": 1.5519,
+      "step": 502
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.177870273590088,
+      "learning_rate": 0.00029759999999999997,
+      "loss": 1.3197,
+      "step": 503
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.229679107666016,
+      "learning_rate": 0.0002982,
+      "loss": 1.2668,
+      "step": 504
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.256344795227051,
+      "learning_rate": 0.0002988,
+      "loss": 1.2646,
+      "step": 505
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.3408243656158447,
+      "learning_rate": 0.00029939999999999996,
+      "loss": 0.8331,
+      "step": 506
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.4926090240478516,
+      "learning_rate": 0.0003,
+      "loss": 0.7081,
+      "step": 507
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.9362640380859375,
+      "learning_rate": 0.0002999627883899777,
+      "loss": 1.1574,
+      "step": 508
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.2679216861724854,
+      "learning_rate": 0.0002999255767799553,
+      "loss": 0.546,
+      "step": 509
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 10.322508811950684,
+      "learning_rate": 0.000299888365169933,
+      "loss": 0.9139,
+      "step": 510
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.090631484985352,
+      "learning_rate": 0.0002998511535599107,
+      "loss": 0.5386,
+      "step": 511
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.046153545379639,
+      "learning_rate": 0.0002998139419498883,
+      "loss": 0.7433,
+      "step": 512
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 10.831178665161133,
+      "learning_rate": 0.00029977673033986603,
+      "loss": 1.5149,
+      "step": 513
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 7.8567633628845215,
+      "learning_rate": 0.0002997395187298437,
+      "loss": 1.0058,
+      "step": 514
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.131727695465088,
+      "learning_rate": 0.0002997023071198214,
+      "loss": 0.3814,
+      "step": 515
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 11.242587089538574,
+      "learning_rate": 0.00029966509550979903,
+      "loss": 1.0306,
+      "step": 516
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 8.567095756530762,
+      "learning_rate": 0.0002996278838997767,
+      "loss": 1.1139,
+      "step": 517
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 6.56288480758667,
+      "learning_rate": 0.0002995906722897544,
+      "loss": 1.6086,
+      "step": 518
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 7.899925231933594,
+      "learning_rate": 0.00029955346067973203,
+      "loss": 1.5415,
+      "step": 519
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 6.132180213928223,
+      "learning_rate": 0.00029951624906970973,
+      "loss": 1.2357,
+      "step": 520
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.827673435211182,
+      "learning_rate": 0.00029947903745968743,
+      "loss": 0.7701,
+      "step": 521
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 12.996684074401855,
+      "learning_rate": 0.0002994418258496651,
+      "loss": 1.4232,
+      "step": 522
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 6.13612174987793,
+      "learning_rate": 0.00029940461423964273,
+      "loss": 0.9823,
+      "step": 523
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 6.259670734405518,
+      "learning_rate": 0.0002993674026296204,
+      "loss": 1.2875,
+      "step": 524
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": NaN,
+      "learning_rate": 0.0002993674026296204,
+      "loss": 0.694,
+      "step": 525
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 6.94905424118042,
+      "learning_rate": 0.0002993301910195981,
+      "loss": 1.9648,
+      "step": 526
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.61449933052063,
+      "learning_rate": 0.0002992929794095758,
+      "loss": 1.4221,
+      "step": 527
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.9974653720855713,
+      "learning_rate": 0.00029925576779955344,
+      "loss": 1.28,
+      "step": 528
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.458034992218018,
+      "learning_rate": 0.00029921855618953114,
+      "loss": 0.9629,
+      "step": 529
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.363546371459961,
+      "learning_rate": 0.0002991813445795088,
+      "loss": 0.8967,
+      "step": 530
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.293959140777588,
+      "learning_rate": 0.00029914413296948644,
+      "loss": 1.022,
+      "step": 531
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.35537052154541,
+      "learning_rate": 0.00029910692135946414,
+      "loss": 0.6593,
+      "step": 532
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.888710021972656,
+      "learning_rate": 0.0002990697097494418,
+      "loss": 0.9094,
+      "step": 533
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.332656383514404,
+      "learning_rate": 0.0002990324981394195,
+      "loss": 1.7321,
+      "step": 534
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 8.324007034301758,
+      "learning_rate": 0.00029899528652939714,
+      "loss": 1.4225,
+      "step": 535
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.71950101852417,
+      "learning_rate": 0.0002989580749193748,
+      "loss": 0.626,
+      "step": 536
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.610879421234131,
+      "learning_rate": 0.0002989208633093525,
+      "loss": 1.0425,
+      "step": 537
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.266473770141602,
+      "learning_rate": 0.00029888365169933014,
+      "loss": 1.2166,
+      "step": 538
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.074027061462402,
+      "learning_rate": 0.00029884644008930784,
+      "loss": 0.9753,
+      "step": 539
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.849838733673096,
+      "learning_rate": 0.00029880922847928554,
+      "loss": 1.1584,
+      "step": 540
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.537852764129639,
+      "learning_rate": 0.0002987720168692632,
+      "loss": 0.907,
+      "step": 541
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.0589985847473145,
+      "learning_rate": 0.00029873480525924084,
+      "loss": 0.6799,
+      "step": 542
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.327163219451904,
+      "learning_rate": 0.0002986975936492185,
+      "loss": 1.8219,
+      "step": 543
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 7.557645797729492,
+      "learning_rate": 0.0002986603820391962,
+      "loss": 1.4392,
+      "step": 544
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.596197128295898,
+      "learning_rate": 0.0002986231704291739,
+      "loss": 1.1227,
+      "step": 545
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.67082405090332,
+      "learning_rate": 0.00029858595881915155,
+      "loss": 1.029,
+      "step": 546
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 7.202919006347656,
+      "learning_rate": 0.00029854874720912925,
+      "loss": 0.736,
+      "step": 547
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.42061185836792,
+      "learning_rate": 0.0002985115355991069,
+      "loss": 1.1307,
+      "step": 548
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 7.092513561248779,
+      "learning_rate": 0.00029847432398908455,
+      "loss": 1.2644,
+      "step": 549
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.857559680938721,
+      "learning_rate": 0.00029843711237906225,
+      "loss": 0.8481,
+      "step": 550
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.834782123565674,
+      "learning_rate": 0.0002983999007690399,
+      "loss": 1.6768,
+      "step": 551
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.073868751525879,
+      "learning_rate": 0.0002983626891590176,
+      "loss": 1.9901,
+      "step": 552
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.636265754699707,
+      "learning_rate": 0.00029832547754899525,
+      "loss": 0.855,
+      "step": 553
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.978489398956299,
+      "learning_rate": 0.00029828826593897295,
+      "loss": 1.5895,
+      "step": 554
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.5552754402160645,
+      "learning_rate": 0.0002982510543289506,
+      "loss": 1.0363,
+      "step": 555
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.47123908996582,
+      "learning_rate": 0.00029821384271892825,
+      "loss": 1.305,
+      "step": 556
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.921194076538086,
+      "learning_rate": 0.00029817663110890595,
+      "loss": 1.571,
+      "step": 557
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.359584331512451,
+      "learning_rate": 0.00029813941949888366,
+      "loss": 1.0663,
+      "step": 558
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.973052501678467,
+      "learning_rate": 0.0002981022078888613,
+      "loss": 1.1589,
+      "step": 559
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.5657124519348145,
+      "learning_rate": 0.00029806499627883895,
+      "loss": 0.8758,
+      "step": 560
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.143813371658325,
+      "learning_rate": 0.00029802778466881666,
+      "loss": 0.9989,
+      "step": 561
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.4931013584136963,
+      "learning_rate": 0.0002979905730587943,
+      "loss": 0.5586,
+      "step": 562
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.445590496063232,
+      "learning_rate": 0.000297953361448772,
+      "loss": 1.5847,
+      "step": 563
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 7.393503665924072,
+      "learning_rate": 0.00029791614983874966,
+      "loss": 1.226,
+      "step": 564
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.212005138397217,
+      "learning_rate": 0.00029787893822872736,
+      "loss": 1.1496,
+      "step": 565
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.066720962524414,
+      "learning_rate": 0.000297841726618705,
+      "loss": 1.0976,
+      "step": 566
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.2609734535217285,
+      "learning_rate": 0.00029780451500868266,
+      "loss": 0.8392,
+      "step": 567
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.803570508956909,
+      "learning_rate": 0.00029776730339866036,
+      "loss": 0.7692,
+      "step": 568
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.159245491027832,
+      "learning_rate": 0.000297730091788638,
+      "loss": 0.8532,
+      "step": 569
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.429141998291016,
+      "learning_rate": 0.0002976928801786157,
+      "loss": 0.7445,
+      "step": 570
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 9.767818450927734,
+      "learning_rate": 0.0002976556685685934,
+      "loss": 1.5825,
+      "step": 571
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.809798717498779,
+      "learning_rate": 0.00029761845695857106,
+      "loss": 1.5131,
+      "step": 572
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 20.902917861938477,
+      "learning_rate": 0.0002975812453485487,
+      "loss": 1.8095,
+      "step": 573
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.394655704498291,
+      "learning_rate": 0.0002975440337385264,
+      "loss": 1.2016,
+      "step": 574
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": NaN,
+      "learning_rate": 0.0002975440337385264,
+      "loss": 0.531,
+      "step": 575
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 9.339009284973145,
+      "learning_rate": 0.00029750682212850406,
+      "loss": 3.3416,
+      "step": 576
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.734882354736328,
+      "learning_rate": 0.00029746961051848177,
+      "loss": 2.2444,
+      "step": 577
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.9475908279418945,
+      "learning_rate": 0.0002974323989084594,
+      "loss": 1.2054,
+      "step": 578
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.528080463409424,
+      "learning_rate": 0.00029739518729843706,
+      "loss": 1.3752,
+      "step": 579
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.754209041595459,
+      "learning_rate": 0.00029735797568841477,
+      "loss": 0.7848,
+      "step": 580
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.216021776199341,
+      "learning_rate": 0.0002973207640783924,
+      "loss": 0.6936,
+      "step": 581
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.249885082244873,
+      "learning_rate": 0.0002972835524683701,
+      "loss": 0.8303,
+      "step": 582
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.777367115020752,
+      "learning_rate": 0.00029724634085834777,
+      "loss": 0.6014,
+      "step": 583
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 9.312623023986816,
+      "learning_rate": 0.00029720912924832547,
+      "loss": 1.1901,
+      "step": 584
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.88458776473999,
+      "learning_rate": 0.0002971719176383031,
+      "loss": 1.1543,
+      "step": 585
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.8880615234375,
+      "learning_rate": 0.00029713470602828077,
+      "loss": 1.0548,
+      "step": 586
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.879009246826172,
+      "learning_rate": 0.00029709749441825847,
+      "loss": 1.0961,
+      "step": 587
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.066569805145264,
+      "learning_rate": 0.0002970602828082361,
+      "loss": 0.6876,
+      "step": 588
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.1321892738342285,
+      "learning_rate": 0.0002970230711982138,
+      "loss": 1.0765,
+      "step": 589
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.651081085205078,
+      "learning_rate": 0.0002969858595881915,
+      "loss": 0.6319,
+      "step": 590
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.676162242889404,
+      "learning_rate": 0.00029694864797816917,
+      "loss": 0.8857,
+      "step": 591
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.472525119781494,
+      "learning_rate": 0.0002969114363681468,
+      "loss": 1.2666,
+      "step": 592
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.793959140777588,
+      "learning_rate": 0.0002968742247581245,
+      "loss": 0.8785,
+      "step": 593
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 7.9074931144714355,
+      "learning_rate": 0.00029683701314810217,
+      "loss": 1.4696,
+      "step": 594
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.401146411895752,
+      "learning_rate": 0.0002967998015380799,
+      "loss": 0.5815,
+      "step": 595
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 7.1119513511657715,
+      "learning_rate": 0.0002967625899280575,
+      "loss": 1.0864,
+      "step": 596
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 7.709166049957275,
+      "learning_rate": 0.0002967253783180352,
+      "loss": 1.6677,
+      "step": 597
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 7.580686092376709,
+      "learning_rate": 0.0002966881667080129,
+      "loss": 1.2696,
+      "step": 598
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.200621128082275,
+      "learning_rate": 0.0002966509550979905,
+      "loss": 0.6652,
+      "step": 599
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 8.517735481262207,
+      "learning_rate": 0.0002966137434879682,
+      "loss": 1.5409,
+      "step": 600
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 1.319638729095459,
+      "eval_runtime": 51.1269,
+      "eval_samples_per_second": 42.404,
+      "eval_steps_per_second": 10.601,
+      "eval_wer": 0.909271752291991,
+      "step": 600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 12.722813606262207,
+      "learning_rate": 0.0002965765318779459,
+      "loss": 2.4553,
+      "step": 601
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.727902412414551,
+      "learning_rate": 0.0002965393202679236,
+      "loss": 1.4027,
+      "step": 602
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.360785007476807,
+      "learning_rate": 0.00029650210865790123,
+      "loss": 1.1641,
+      "step": 603
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.100715637207031,
+      "learning_rate": 0.0002964648970478789,
+      "loss": 1.0047,
+      "step": 604
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.119365692138672,
+      "learning_rate": 0.0002964276854378566,
+      "loss": 1.1648,
+      "step": 605
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.546939373016357,
+      "learning_rate": 0.0002963904738278343,
+      "loss": 1.3271,
+      "step": 606
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.981604099273682,
+      "learning_rate": 0.00029635326221781193,
+      "loss": 1.2142,
+      "step": 607
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.839823246002197,
+      "learning_rate": 0.00029631605060778963,
+      "loss": 1.6792,
+      "step": 608
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 7.334500789642334,
+      "learning_rate": 0.0002962788389977673,
+      "loss": 1.2355,
+      "step": 609
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.874765872955322,
+      "learning_rate": 0.00029624162738774493,
+      "loss": 0.9184,
+      "step": 610
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.3265280723571777,
+      "learning_rate": 0.00029620441577772263,
+      "loss": 0.6558,
+      "step": 611
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.5789058208465576,
+      "learning_rate": 0.0002961672041677003,
+      "loss": 0.8824,
+      "step": 612
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.2460458278656006,
+      "learning_rate": 0.000296129992557678,
+      "loss": 1.0423,
+      "step": 613
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 6.92147970199585,
+      "learning_rate": 0.00029609278094765563,
+      "loss": 0.7398,
+      "step": 614
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.5256998538970947,
+      "learning_rate": 0.00029605556933763334,
+      "loss": 0.6854,
+      "step": 615
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.17782735824585,
+      "learning_rate": 0.000296018357727611,
+      "loss": 1.0298,
+      "step": 616
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 8.018060684204102,
+      "learning_rate": 0.00029598114611758863,
+      "loss": 2.2126,
+      "step": 617
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 7.263321876525879,
+      "learning_rate": 0.00029594393450756634,
+      "loss": 0.9248,
+      "step": 618
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 6.241823196411133,
+      "learning_rate": 0.00029590672289754404,
+      "loss": 1.1756,
+      "step": 619
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.013612747192383,
+      "learning_rate": 0.0002958695112875217,
+      "loss": 0.882,
+      "step": 620
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.5317792892456055,
+      "learning_rate": 0.00029583229967749934,
+      "loss": 1.0661,
+      "step": 621
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 8.019999504089355,
+      "learning_rate": 0.00029579508806747704,
+      "loss": 0.9743,
+      "step": 622
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 10.188430786132812,
+      "learning_rate": 0.0002957578764574547,
+      "loss": 1.3912,
+      "step": 623
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 6.105098724365234,
+      "learning_rate": 0.0002957206648474324,
+      "loss": 0.8787,
+      "step": 624
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 8.475005149841309,
+      "learning_rate": 0.00029568345323741004,
+      "loss": 2.6934,
+      "step": 625
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.846712589263916,
+      "learning_rate": 0.00029564624162738774,
+      "loss": 1.4757,
+      "step": 626
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.2802951335906982,
+      "learning_rate": 0.0002956090300173654,
+      "loss": 0.9426,
+      "step": 627
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.0451314449310303,
+      "learning_rate": 0.00029557181840734304,
+      "loss": 0.8553,
+      "step": 628
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.837835788726807,
+      "learning_rate": 0.00029553460679732074,
+      "loss": 1.0035,
+      "step": 629
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.3272557258605957,
+      "learning_rate": 0.0002954973951872984,
+      "loss": 1.7012,
+      "step": 630
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.1126511096954346,
+      "learning_rate": 0.0002954601835772761,
+      "loss": 0.8654,
+      "step": 631
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.995132923126221,
+      "learning_rate": 0.00029542297196725374,
+      "loss": 0.9708,
+      "step": 632
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.542963027954102,
+      "learning_rate": 0.00029538576035723145,
+      "loss": 1.0861,
+      "step": 633
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.722038984298706,
+      "learning_rate": 0.0002953485487472091,
+      "loss": 1.012,
+      "step": 634
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.468753337860107,
+      "learning_rate": 0.00029531133713718674,
+      "loss": 1.139,
+      "step": 635
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.822729110717773,
+      "learning_rate": 0.00029527412552716445,
+      "loss": 1.1173,
+      "step": 636
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.884435653686523,
+      "learning_rate": 0.00029523691391714215,
+      "loss": 1.0669,
+      "step": 637
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.150437831878662,
+      "learning_rate": 0.0002951997023071198,
+      "loss": 0.871,
+      "step": 638
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.12093448638916,
+      "learning_rate": 0.0002951624906970975,
+      "loss": 0.6693,
+      "step": 639
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 7.429576396942139,
+      "learning_rate": 0.00029512527908707515,
+      "loss": 1.0517,
+      "step": 640
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.6653037071228027,
+      "learning_rate": 0.0002950880674770528,
+      "loss": 0.8529,
+      "step": 641
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.8426244258880615,
+      "learning_rate": 0.0002950508558670305,
+      "loss": 0.7732,
+      "step": 642
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 7.198141574859619,
+      "learning_rate": 0.00029501364425700815,
+      "loss": 1.6585,
+      "step": 643
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 7.202881813049316,
+      "learning_rate": 0.00029497643264698585,
+      "loss": 1.4548,
+      "step": 644
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.224870681762695,
+      "learning_rate": 0.0002949392210369635,
+      "loss": 1.6397,
+      "step": 645
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.892850160598755,
+      "learning_rate": 0.00029490200942694115,
+      "loss": 1.0224,
+      "step": 646
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.112697601318359,
+      "learning_rate": 0.00029486479781691885,
+      "loss": 0.8937,
+      "step": 647
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 10.07210636138916,
+      "learning_rate": 0.0002948275862068965,
+      "loss": 1.0152,
+      "step": 648
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.240150451660156,
+      "learning_rate": 0.0002947903745968742,
+      "loss": 1.0351,
+      "step": 649
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 9.268518447875977,
+      "learning_rate": 0.0002947531629868519,
+      "loss": 0.9157,
+      "step": 650
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.466054916381836,
+      "learning_rate": 0.00029471595137682956,
+      "loss": 1.6634,
+      "step": 651
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.009162425994873,
+      "learning_rate": 0.0002946787397668072,
+      "loss": 1.2699,
+      "step": 652
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.312174320220947,
+      "learning_rate": 0.00029464152815678485,
+      "loss": 1.2239,
+      "step": 653
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.121635675430298,
+      "learning_rate": 0.00029460431654676256,
+      "loss": 0.9369,
+      "step": 654
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.073437213897705,
+      "learning_rate": 0.00029456710493674026,
+      "loss": 1.1005,
+      "step": 655
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.3540802001953125,
+      "learning_rate": 0.0002945298933267179,
+      "loss": 1.0565,
+      "step": 656
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.2247724533081055,
+      "learning_rate": 0.0002944926817166956,
+      "loss": 0.7415,
+      "step": 657
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.133108139038086,
+      "learning_rate": 0.00029445547010667326,
+      "loss": 0.4355,
+      "step": 658
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.775803089141846,
+      "learning_rate": 0.0002944182584966509,
+      "loss": 1.0846,
+      "step": 659
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.800411701202393,
+      "learning_rate": 0.0002943810468866286,
+      "loss": 1.1723,
+      "step": 660
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.831652641296387,
+      "learning_rate": 0.00029434383527660626,
+      "loss": 0.8722,
+      "step": 661
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.7799882888793945,
+      "learning_rate": 0.00029430662366658396,
+      "loss": 1.0813,
+      "step": 662
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 9.425108909606934,
+      "learning_rate": 0.0002942694120565616,
+      "loss": 2.4706,
+      "step": 663
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 7.258337020874023,
+      "learning_rate": 0.0002942322004465393,
+      "loss": 1.0422,
+      "step": 664
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.805200576782227,
+      "learning_rate": 0.00029419498883651696,
+      "loss": 0.8724,
+      "step": 665
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.374764442443848,
+      "learning_rate": 0.0002941577772264946,
+      "loss": 0.8653,
+      "step": 666
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 30.488962173461914,
+      "learning_rate": 0.0002941205656164723,
+      "loss": 2.5592,
+      "step": 667
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 9.162358283996582,
+      "learning_rate": 0.00029408335400645,
+      "loss": 1.6403,
+      "step": 668
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.208140850067139,
+      "learning_rate": 0.00029404614239642767,
+      "loss": 1.0371,
+      "step": 669
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.24813175201416,
+      "learning_rate": 0.0002940089307864053,
+      "loss": 0.6602,
+      "step": 670
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 6.219293594360352,
+      "learning_rate": 0.00029397171917638296,
+      "loss": 1.2641,
+      "step": 671
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 10.14699935913086,
+      "learning_rate": 0.00029393450756636067,
+      "loss": 1.5108,
+      "step": 672
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 8.945874214172363,
+      "learning_rate": 0.00029389729595633837,
+      "loss": 0.8659,
+      "step": 673
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 9.979119300842285,
+      "learning_rate": 0.000293860084346316,
+      "loss": 0.6198,
+      "step": 674
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 10.116988182067871,
+      "learning_rate": 0.0002938228727362937,
+      "loss": 1.1682,
+      "step": 675
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 7.378312587738037,
+      "learning_rate": 0.00029378566112627137,
+      "loss": 2.0674,
+      "step": 676
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 7.4542012214660645,
+      "learning_rate": 0.000293748449516249,
+      "loss": 1.8939,
+      "step": 677
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 6.61478328704834,
+      "learning_rate": 0.0002937112379062267,
+      "loss": 1.6356,
+      "step": 678
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.97426176071167,
+      "learning_rate": 0.00029367402629620437,
+      "loss": 0.8941,
+      "step": 679
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.164785385131836,
+      "learning_rate": 0.0002936368146861821,
+      "loss": 0.6823,
+      "step": 680
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.324411869049072,
+      "learning_rate": 0.0002935996030761598,
+      "loss": 0.6235,
+      "step": 681
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.255089282989502,
+      "learning_rate": 0.0002935623914661374,
+      "loss": 0.9023,
+      "step": 682
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.362627029418945,
+      "learning_rate": 0.0002935251798561151,
+      "loss": 1.3556,
+      "step": 683
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.432969570159912,
+      "learning_rate": 0.0002934879682460927,
+      "loss": 0.8282,
+      "step": 684
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.972658395767212,
+      "learning_rate": 0.0002934507566360704,
+      "loss": 1.0565,
+      "step": 685
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.701770305633545,
+      "learning_rate": 0.00029341354502604813,
+      "loss": 1.2406,
+      "step": 686
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.701326370239258,
+      "learning_rate": 0.0002933763334160258,
+      "loss": 0.7059,
+      "step": 687
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.311460256576538,
+      "learning_rate": 0.0002933391218060034,
+      "loss": 0.9308,
+      "step": 688
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.932095050811768,
+      "learning_rate": 0.00029330191019598113,
+      "loss": 1.8754,
+      "step": 689
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.051765441894531,
+      "learning_rate": 0.0002932646985859588,
+      "loss": 0.7799,
+      "step": 690
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.459011554718018,
+      "learning_rate": 0.0002932274869759365,
+      "loss": 0.8546,
+      "step": 691
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.098698616027832,
+      "learning_rate": 0.00029319027536591413,
+      "loss": 0.5279,
+      "step": 692
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.640591621398926,
+      "learning_rate": 0.00029315306375589183,
+      "loss": 0.9102,
+      "step": 693
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 6.9347004890441895,
+      "learning_rate": 0.0002931158521458695,
+      "loss": 0.7036,
+      "step": 694
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.845668792724609,
+      "learning_rate": 0.00029307864053584713,
+      "loss": 0.5378,
+      "step": 695
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.8146281242370605,
+      "learning_rate": 0.00029304142892582483,
+      "loss": 1.1422,
+      "step": 696
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 6.487931251525879,
+      "learning_rate": 0.0002930042173158025,
+      "loss": 1.0314,
+      "step": 697
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 6.625868797302246,
+      "learning_rate": 0.0002929670057057802,
+      "loss": 1.5303,
+      "step": 698
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 14.07421588897705,
+      "learning_rate": 0.0002929297940957579,
+      "loss": 2.0771,
+      "step": 699
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 7.190564155578613,
+      "learning_rate": 0.00029289258248573554,
+      "loss": 1.1797,
+      "step": 700
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 1.2068666219711304,
+      "eval_runtime": 51.0804,
+      "eval_samples_per_second": 42.443,
+      "eval_steps_per_second": 10.611,
+      "eval_wer": 0.8056564608199274,
+      "step": 700
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 6.680876731872559,
+      "learning_rate": 0.0002928553708757132,
+      "loss": 2.1408,
+      "step": 701
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.9025561809539795,
+      "learning_rate": 0.00029281815926569083,
+      "loss": 1.5205,
+      "step": 702
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.40712833404541,
+      "learning_rate": 0.00029278094765566854,
+      "loss": 1.4955,
+      "step": 703
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.2378780841827393,
+      "learning_rate": 0.00029274373604564624,
+      "loss": 0.9419,
+      "step": 704
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.353944778442383,
+      "learning_rate": 0.0002927065244356239,
+      "loss": 1.0087,
+      "step": 705
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.185000896453857,
+      "learning_rate": 0.0002926693128256016,
+      "loss": 1.2567,
+      "step": 706
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.718811511993408,
+      "learning_rate": 0.00029263210121557924,
+      "loss": 0.8007,
+      "step": 707
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.834765911102295,
+      "learning_rate": 0.0002925948896055569,
+      "loss": 0.9665,
+      "step": 708
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.415790557861328,
+      "learning_rate": 0.0002925576779955346,
+      "loss": 0.6987,
+      "step": 709
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.706892490386963,
+      "learning_rate": 0.00029252046638551224,
+      "loss": 0.7334,
+      "step": 710
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.0536246299743652,
+      "learning_rate": 0.00029248325477548994,
+      "loss": 0.9511,
+      "step": 711
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.1888933181762695,
+      "learning_rate": 0.0002924460431654676,
+      "loss": 1.2429,
+      "step": 712
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.904407262802124,
+      "learning_rate": 0.0002924088315554453,
+      "loss": 1.1109,
+      "step": 713
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.773108005523682,
+      "learning_rate": 0.00029237161994542294,
+      "loss": 0.9709,
+      "step": 714
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.608194351196289,
+      "learning_rate": 0.0002923344083354006,
+      "loss": 0.402,
+      "step": 715
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.155666351318359,
+      "learning_rate": 0.0002922971967253783,
+      "loss": 1.4452,
+      "step": 716
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 11.795857429504395,
+      "learning_rate": 0.000292259985115356,
+      "loss": 3.5267,
+      "step": 717
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.475924491882324,
+      "learning_rate": 0.00029222277350533365,
+      "loss": 0.6192,
+      "step": 718
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.139604568481445,
+      "learning_rate": 0.0002921855618953113,
+      "loss": 0.7032,
+      "step": 719
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.846983909606934,
+      "learning_rate": 0.00029214835028528894,
+      "loss": 1.1969,
+      "step": 720
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.0570387840271,
+      "learning_rate": 0.00029211113867526665,
+      "loss": 0.6138,
+      "step": 721
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.584964275360107,
+      "learning_rate": 0.00029207392706524435,
+      "loss": 0.6882,
+      "step": 722
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 8.10388469696045,
+      "learning_rate": 0.000292036715455222,
+      "loss": 0.3923,
+      "step": 723
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 6.056919097900391,
+      "learning_rate": 0.0002919995038451997,
+      "loss": 1.0729,
+      "step": 724
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 7.191409111022949,
+      "learning_rate": 0.00029196229223517735,
+      "loss": 1.0135,
+      "step": 725
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 6.211596488952637,
+      "learning_rate": 0.000291925080625155,
+      "loss": 2.1364,
+      "step": 726
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.2215120792388916,
+      "learning_rate": 0.0002918878690151327,
+      "loss": 0.9194,
+      "step": 727
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.045462131500244,
+      "learning_rate": 0.00029185065740511035,
+      "loss": 0.7808,
+      "step": 728
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.9927256107330322,
+      "learning_rate": 0.00029181344579508805,
+      "loss": 0.7217,
+      "step": 729
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.7817091941833496,
+      "learning_rate": 0.0002917762341850657,
+      "loss": 0.724,
+      "step": 730
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.1122050285339355,
+      "learning_rate": 0.0002917390225750434,
+      "loss": 0.7772,
+      "step": 731
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.8279852867126465,
+      "learning_rate": 0.00029170181096502105,
+      "loss": 1.2088,
+      "step": 732
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.453472852706909,
+      "learning_rate": 0.0002916645993549987,
+      "loss": 0.6257,
+      "step": 733
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.543291091918945,
+      "learning_rate": 0.0002916273877449764,
+      "loss": 0.7873,
+      "step": 734
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.899169445037842,
+      "learning_rate": 0.0002915901761349541,
+      "loss": 0.7794,
+      "step": 735
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.045598030090332,
+      "learning_rate": 0.00029155296452493176,
+      "loss": 0.4835,
+      "step": 736
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.990403175354004,
+      "learning_rate": 0.0002915157529149094,
+      "loss": 1.0899,
+      "step": 737
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 7.7346110343933105,
+      "learning_rate": 0.0002914785413048871,
+      "loss": 1.0977,
+      "step": 738
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.3284356594085693,
+      "learning_rate": 0.00029144132969486476,
+      "loss": 0.4629,
+      "step": 739
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.3941850662231445,
+      "learning_rate": 0.00029140411808484246,
+      "loss": 0.9461,
+      "step": 740
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.9711191654205322,
+      "learning_rate": 0.0002913669064748201,
+      "loss": 1.1025,
+      "step": 741
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 19.457284927368164,
+      "learning_rate": 0.0002913296948647978,
+      "loss": 2.8168,
+      "step": 742
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.4924750328063965,
+      "learning_rate": 0.00029129248325477546,
+      "loss": 1.3301,
+      "step": 743
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.154421806335449,
+      "learning_rate": 0.0002912552716447531,
+      "loss": 0.8412,
+      "step": 744
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.290253162384033,
+      "learning_rate": 0.0002912180600347308,
+      "loss": 0.9267,
+      "step": 745
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.45554780960083,
+      "learning_rate": 0.00029118084842470846,
+      "loss": 0.6853,
+      "step": 746
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.069665431976318,
+      "learning_rate": 0.00029114363681468616,
+      "loss": 0.6652,
+      "step": 747
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.831662178039551,
+      "learning_rate": 0.00029110642520466387,
+      "loss": 0.7098,
+      "step": 748
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.927679538726807,
+      "learning_rate": 0.0002910692135946415,
+      "loss": 0.7609,
+      "step": 749
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.876534938812256,
+      "learning_rate": 0.00029103200198461916,
+      "loss": 0.7453,
+      "step": 750
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.562718868255615,
+      "learning_rate": 0.00029099479037459687,
+      "loss": 1.7246,
+      "step": 751
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.7739195823669434,
+      "learning_rate": 0.0002909575787645745,
+      "loss": 1.3101,
+      "step": 752
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.139188051223755,
+      "learning_rate": 0.0002909203671545522,
+      "loss": 1.1718,
+      "step": 753
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.7190756797790527,
+      "learning_rate": 0.00029088315554452987,
+      "loss": 0.7607,
+      "step": 754
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.391061544418335,
+      "learning_rate": 0.00029084594393450757,
+      "loss": 0.937,
+      "step": 755
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.2688794136047363,
+      "learning_rate": 0.0002908087323244852,
+      "loss": 0.4747,
+      "step": 756
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.645151376724243,
+      "learning_rate": 0.00029077152071446287,
+      "loss": 0.6512,
+      "step": 757
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 5.204065322875977,
+      "learning_rate": 0.00029073430910444057,
+      "loss": 0.7823,
+      "step": 758
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.213620185852051,
+      "learning_rate": 0.0002906970974944182,
+      "loss": 1.1063,
+      "step": 759
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 7.5675129890441895,
+      "learning_rate": 0.0002906598858843959,
+      "loss": 1.9301,
+      "step": 760
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.451794385910034,
+      "learning_rate": 0.00029062267427437357,
+      "loss": 0.7759,
+      "step": 761
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.535918712615967,
+      "learning_rate": 0.0002905854626643512,
+      "loss": 0.7944,
+      "step": 762
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 5.344732761383057,
+      "learning_rate": 0.0002905482510543289,
+      "loss": 1.0658,
+      "step": 763
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.685420274734497,
+      "learning_rate": 0.0002905110394443066,
+      "loss": 0.5017,
+      "step": 764
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 5.2186055183410645,
+      "learning_rate": 0.00029047382783428427,
+      "loss": 2.056,
+      "step": 765
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 6.418582439422607,
+      "learning_rate": 0.000290436616224262,
+      "loss": 1.7033,
+      "step": 766
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 6.189699649810791,
+      "learning_rate": 0.0002903994046142396,
+      "loss": 1.5274,
+      "step": 767
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 6.971052646636963,
+      "learning_rate": 0.00029036219300421727,
+      "loss": 0.889,
+      "step": 768
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 5.264997959136963,
+      "learning_rate": 0.000290324981394195,
+      "loss": 0.7537,
+      "step": 769
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 9.451274871826172,
+      "learning_rate": 0.0002902877697841726,
+      "loss": 0.6874,
+      "step": 770
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.780513763427734,
+      "learning_rate": 0.0002902505581741503,
+      "loss": 0.9426,
+      "step": 771
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 7.542695999145508,
+      "learning_rate": 0.000290213346564128,
+      "loss": 1.5722,
+      "step": 772
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 9.267386436462402,
+      "learning_rate": 0.0002901761349541057,
+      "loss": 0.9906,
+      "step": 773
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.18009090423584,
+      "learning_rate": 0.00029013892334408333,
+      "loss": 0.9282,
+      "step": 774
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 15.079828262329102,
+      "learning_rate": 0.000290101711734061,
+      "loss": 2.3483,
+      "step": 775
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 6.639895439147949,
+      "learning_rate": 0.0002900645001240387,
+      "loss": 1.7343,
+      "step": 776
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 5.052302360534668,
+      "learning_rate": 0.00029002728851401633,
+      "loss": 1.4465,
+      "step": 777
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.000107765197754,
+      "learning_rate": 0.00028999007690399403,
+      "loss": 0.9901,
+      "step": 778
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.0668177604675293,
+      "learning_rate": 0.0002899528652939717,
+      "loss": 0.8454,
+      "step": 779
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.6663155555725098,
+      "learning_rate": 0.0002899156536839494,
+      "loss": 0.7529,
+      "step": 780
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.555816650390625,
+      "learning_rate": 0.00028987844207392703,
+      "loss": 0.6824,
+      "step": 781
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.055764675140381,
+      "learning_rate": 0.00028984123046390473,
+      "loss": 1.1432,
+      "step": 782
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.053997039794922,
+      "learning_rate": 0.0002898040188538824,
+      "loss": 1.0801,
+      "step": 783
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.9936132431030273,
+      "learning_rate": 0.0002897668072438601,
+      "loss": 1.2037,
+      "step": 784
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.756950378417969,
+      "learning_rate": 0.00028972959563383773,
+      "loss": 1.5041,
+      "step": 785
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.913256883621216,
+      "learning_rate": 0.0002896923840238154,
+      "loss": 0.6958,
+      "step": 786
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.463769912719727,
+      "learning_rate": 0.0002896551724137931,
+      "loss": 1.1124,
+      "step": 787
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.305164813995361,
+      "learning_rate": 0.00028961796080377073,
+      "loss": 1.2586,
+      "step": 788
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.605264663696289,
+      "learning_rate": 0.00028958074919374844,
+      "loss": 0.4304,
+      "step": 789
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.45980167388916,
+      "learning_rate": 0.0002895435375837261,
+      "loss": 1.613,
+      "step": 790
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.24216365814209,
+      "learning_rate": 0.0002895063259737038,
+      "loss": 1.0244,
+      "step": 791
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.666222333908081,
+      "learning_rate": 0.00028946911436368144,
+      "loss": 0.6473,
+      "step": 792
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.788506031036377,
+      "learning_rate": 0.0002894319027536591,
+      "loss": 0.8939,
+      "step": 793
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.17887020111084,
+      "learning_rate": 0.0002893946911436368,
+      "loss": 1.0748,
+      "step": 794
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.128133773803711,
+      "learning_rate": 0.0002893574795336145,
+      "loss": 0.864,
+      "step": 795
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.645290851593018,
+      "learning_rate": 0.00028932026792359214,
+      "loss": 1.656,
+      "step": 796
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.542060375213623,
+      "learning_rate": 0.00028928305631356984,
+      "loss": 1.016,
+      "step": 797
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.639885425567627,
+      "learning_rate": 0.0002892458447035475,
+      "loss": 0.6474,
+      "step": 798
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.543883800506592,
+      "learning_rate": 0.00028920863309352514,
+      "loss": 0.895,
+      "step": 799
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 6.362172603607178,
+      "learning_rate": 0.00028917142148350284,
+      "loss": 2.4732,
+      "step": 800
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 0.987212598323822,
+      "eval_runtime": 50.7349,
+      "eval_samples_per_second": 42.732,
+      "eval_steps_per_second": 10.683,
+      "eval_wer": 0.7189932537623248,
+      "step": 800
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.333866596221924,
+      "learning_rate": 0.0002891342098734805,
+      "loss": 1.3953,
+      "step": 801
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.874297142028809,
+      "learning_rate": 0.0002890969982634582,
+      "loss": 1.6125,
+      "step": 802
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.766399383544922,
+      "learning_rate": 0.00028905978665343584,
+      "loss": 1.0827,
+      "step": 803
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.576672077178955,
+      "learning_rate": 0.0002890225750434135,
+      "loss": 0.8876,
+      "step": 804
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.9895853996276855,
+      "learning_rate": 0.0002889853634333912,
+      "loss": 0.7372,
+      "step": 805
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.91171407699585,
+      "learning_rate": 0.00028894815182336884,
+      "loss": 2.1011,
+      "step": 806
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.903454303741455,
+      "learning_rate": 0.00028891094021334655,
+      "loss": 1.009,
+      "step": 807
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.659839630126953,
+      "learning_rate": 0.00028887372860332425,
+      "loss": 0.8499,
+      "step": 808
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.591513156890869,
+      "learning_rate": 0.0002888365169933019,
+      "loss": 0.9237,
+      "step": 809
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 6.837412357330322,
+      "learning_rate": 0.00028879930538327955,
+      "loss": 1.2858,
+      "step": 810
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.4030585289001465,
+      "learning_rate": 0.0002887620937732572,
+      "loss": 1.3576,
+      "step": 811
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.567882537841797,
+      "learning_rate": 0.0002887248821632349,
+      "loss": 1.4714,
+      "step": 812
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.555302381515503,
+      "learning_rate": 0.0002886876705532126,
+      "loss": 0.4524,
+      "step": 813
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.767725706100464,
+      "learning_rate": 0.00028865045894319025,
+      "loss": 0.4047,
+      "step": 814
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.69365930557251,
+      "learning_rate": 0.00028861324733316795,
+      "loss": 0.835,
+      "step": 815
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 6.788897514343262,
+      "learning_rate": 0.0002885760357231456,
+      "loss": 1.5947,
+      "step": 816
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.98279070854187,
+      "learning_rate": 0.00028853882411312325,
+      "loss": 0.8821,
+      "step": 817
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.854119777679443,
+      "learning_rate": 0.00028850161250310095,
+      "loss": 0.9184,
+      "step": 818
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.863973617553711,
+      "learning_rate": 0.0002884644008930786,
+      "loss": 0.9386,
+      "step": 819
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 7.1155266761779785,
+      "learning_rate": 0.0002884271892830563,
+      "loss": 1.128,
+      "step": 820
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.065074920654297,
+      "learning_rate": 0.00028838997767303395,
+      "loss": 0.7205,
+      "step": 821
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.343176364898682,
+      "learning_rate": 0.00028835276606301166,
+      "loss": 0.6354,
+      "step": 822
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.124356746673584,
+      "learning_rate": 0.0002883155544529893,
+      "loss": 0.6844,
+      "step": 823
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.882658958435059,
+      "learning_rate": 0.00028827834284296695,
+      "loss": 1.1918,
+      "step": 824
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 6.812358379364014,
+      "learning_rate": 0.00028824113123294466,
+      "loss": 1.0781,
+      "step": 825
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.14019250869751,
+      "learning_rate": 0.00028820391962292236,
+      "loss": 1.4099,
+      "step": 826
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.570350646972656,
+      "learning_rate": 0.0002881667080129,
+      "loss": 1.0887,
+      "step": 827
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.517451047897339,
+      "learning_rate": 0.00028812949640287766,
+      "loss": 0.6863,
+      "step": 828
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.128011703491211,
+      "learning_rate": 0.0002880922847928553,
+      "loss": 0.8951,
+      "step": 829
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 6.206776142120361,
+      "learning_rate": 0.000288055073182833,
+      "loss": 0.6493,
+      "step": 830
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.9472367763519287,
+      "learning_rate": 0.0002880178615728107,
+      "loss": 0.6847,
+      "step": 831
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.013580799102783,
+      "learning_rate": 0.00028798064996278836,
+      "loss": 0.553,
+      "step": 832
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.9942123889923096,
+      "learning_rate": 0.00028794343835276606,
+      "loss": 1.1586,
+      "step": 833
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.667693138122559,
+      "learning_rate": 0.0002879062267427437,
+      "loss": 1.1743,
+      "step": 834
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.7603282928466797,
+      "learning_rate": 0.00028786901513272136,
+      "loss": 0.8572,
+      "step": 835
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.591860294342041,
+      "learning_rate": 0.00028783180352269906,
+      "loss": 0.6201,
+      "step": 836
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.3808979988098145,
+      "learning_rate": 0.0002877945919126767,
+      "loss": 0.7408,
+      "step": 837
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.802478313446045,
+      "learning_rate": 0.0002877573803026544,
+      "loss": 0.5719,
+      "step": 838
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.7019598484039307,
+      "learning_rate": 0.0002877201686926321,
+      "loss": 0.7113,
+      "step": 839
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.928816318511963,
+      "learning_rate": 0.00028768295708260977,
+      "loss": 1.0754,
+      "step": 840
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.269992828369141,
+      "learning_rate": 0.0002876457454725874,
+      "loss": 0.5499,
+      "step": 841
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.93079948425293,
+      "learning_rate": 0.00028760853386256506,
+      "loss": 0.9921,
+      "step": 842
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.9162845611572266,
+      "learning_rate": 0.00028757132225254277,
+      "loss": 0.4579,
+      "step": 843
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.143979549407959,
+      "learning_rate": 0.00028753411064252047,
+      "loss": 0.5102,
+      "step": 844
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 7.756710052490234,
+      "learning_rate": 0.0002874968990324981,
+      "loss": 0.9789,
+      "step": 845
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 9.213879585266113,
+      "learning_rate": 0.00028745968742247577,
+      "loss": 1.778,
+      "step": 846
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 6.540843486785889,
+      "learning_rate": 0.00028742247581245347,
+      "loss": 2.657,
+      "step": 847
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.8558125495910645,
+      "learning_rate": 0.0002873852642024311,
+      "loss": 0.6151,
+      "step": 848
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.576344013214111,
+      "learning_rate": 0.0002873480525924088,
+      "loss": 0.6595,
+      "step": 849
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 10.87994384765625,
+      "learning_rate": 0.00028731084098238647,
+      "loss": 0.8652,
+      "step": 850
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.8747811317443848,
+      "learning_rate": 0.0002872736293723642,
+      "loss": 1.7178,
+      "step": 851
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.656328201293945,
+      "learning_rate": 0.0002872364177623418,
+      "loss": 1.8586,
+      "step": 852
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.2285523414611816,
+      "learning_rate": 0.00028719920615231947,
+      "loss": 1.324,
+      "step": 853
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.688818454742432,
+      "learning_rate": 0.0002871619945422972,
+      "loss": 1.2527,
+      "step": 854
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.996300220489502,
+      "learning_rate": 0.0002871247829322748,
+      "loss": 0.8493,
+      "step": 855
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.3291850090026855,
+      "learning_rate": 0.0002870875713222525,
+      "loss": 0.9196,
+      "step": 856
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.986222267150879,
+      "learning_rate": 0.00028705035971223023,
+      "loss": 0.9148,
+      "step": 857
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.2459797859191895,
+      "learning_rate": 0.0002870131481022079,
+      "loss": 0.6929,
+      "step": 858
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.315959930419922,
+      "learning_rate": 0.0002869759364921855,
+      "loss": 0.9603,
+      "step": 859
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.7050576210021973,
+      "learning_rate": 0.0002869387248821632,
+      "loss": 0.9597,
+      "step": 860
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.6986582279205322,
+      "learning_rate": 0.0002869015132721409,
+      "loss": 0.8796,
+      "step": 861
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.9992449283599854,
+      "learning_rate": 0.0002868643016621186,
+      "loss": 0.7877,
+      "step": 862
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.221742630004883,
+      "learning_rate": 0.00028682709005209623,
+      "loss": 1.0128,
+      "step": 863
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.570732593536377,
+      "learning_rate": 0.00028678987844207393,
+      "loss": 0.5447,
+      "step": 864
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.575841903686523,
+      "learning_rate": 0.0002867526668320516,
+      "loss": 1.1387,
+      "step": 865
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.252793788909912,
+      "learning_rate": 0.00028671545522202923,
+      "loss": 0.659,
+      "step": 866
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.882000923156738,
+      "learning_rate": 0.00028667824361200693,
+      "loss": 0.4831,
+      "step": 867
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.8467326164245605,
+      "learning_rate": 0.0002866410320019846,
+      "loss": 1.0795,
+      "step": 868
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 6.496115684509277,
+      "learning_rate": 0.0002866038203919623,
+      "loss": 1.1627,
+      "step": 869
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.330075263977051,
+      "learning_rate": 0.00028656660878193993,
+      "loss": 0.5286,
+      "step": 870
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 7.284979820251465,
+      "learning_rate": 0.0002865293971719176,
+      "loss": 1.3597,
+      "step": 871
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.145815372467041,
+      "learning_rate": 0.0002864921855618953,
+      "loss": 0.5407,
+      "step": 872
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 6.272894382476807,
+      "learning_rate": 0.00028645497395187293,
+      "loss": 1.104,
+      "step": 873
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.079260349273682,
+      "learning_rate": 0.00028641776234185064,
+      "loss": 0.9572,
+      "step": 874
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 9.606317520141602,
+      "learning_rate": 0.00028638055073182834,
+      "loss": 0.9342,
+      "step": 875
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.756997585296631,
+      "learning_rate": 0.000286343339121806,
+      "loss": 1.7561,
+      "step": 876
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.379587173461914,
+      "learning_rate": 0.00028630612751178364,
+      "loss": 0.8307,
+      "step": 877
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.6264638900756836,
+      "learning_rate": 0.0002862689159017613,
+      "loss": 0.7802,
+      "step": 878
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.880773544311523,
+      "learning_rate": 0.000286231704291739,
+      "loss": 0.7768,
+      "step": 879
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.7575883865356445,
+      "learning_rate": 0.0002861944926817167,
+      "loss": 1.4659,
+      "step": 880
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.334516763687134,
+      "learning_rate": 0.00028615728107169434,
+      "loss": 1.3035,
+      "step": 881
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.547193765640259,
+      "learning_rate": 0.00028612006946167204,
+      "loss": 0.5167,
+      "step": 882
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4741361141204834,
+      "learning_rate": 0.0002860828578516497,
+      "loss": 0.582,
+      "step": 883
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.9844954013824463,
+      "learning_rate": 0.00028604564624162734,
+      "loss": 0.5924,
+      "step": 884
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.942441940307617,
+      "learning_rate": 0.00028600843463160504,
+      "loss": 0.6893,
+      "step": 885
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.320768356323242,
+      "learning_rate": 0.0002859712230215827,
+      "loss": 0.8256,
+      "step": 886
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.074968338012695,
+      "learning_rate": 0.0002859340114115604,
+      "loss": 1.0719,
+      "step": 887
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.55584716796875,
+      "learning_rate": 0.00028589679980153804,
+      "loss": 1.0453,
+      "step": 888
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.58350133895874,
+      "learning_rate": 0.00028585958819151575,
+      "loss": 0.4654,
+      "step": 889
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.6324543952941895,
+      "learning_rate": 0.0002858223765814934,
+      "loss": 1.0127,
+      "step": 890
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.809018611907959,
+      "learning_rate": 0.00028578516497147104,
+      "loss": 1.1784,
+      "step": 891
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.3140645027160645,
+      "learning_rate": 0.00028574795336144875,
+      "loss": 0.9088,
+      "step": 892
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.170782566070557,
+      "learning_rate": 0.00028571074175142645,
+      "loss": 1.2303,
+      "step": 893
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.350752592086792,
+      "learning_rate": 0.0002856735301414041,
+      "loss": 0.8323,
+      "step": 894
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.265800476074219,
+      "learning_rate": 0.00028563631853138175,
+      "loss": 0.9239,
+      "step": 895
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.861973762512207,
+      "learning_rate": 0.00028559910692135945,
+      "loss": 0.5933,
+      "step": 896
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.536797046661377,
+      "learning_rate": 0.0002855618953113371,
+      "loss": 0.5246,
+      "step": 897
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 6.882984638214111,
+      "learning_rate": 0.0002855246837013148,
+      "loss": 0.782,
+      "step": 898
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.6057496070861816,
+      "learning_rate": 0.00028548747209129245,
+      "loss": 0.8962,
+      "step": 899
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 10.800071716308594,
+      "learning_rate": 0.00028545026048127015,
+      "loss": 0.6917,
+      "step": 900
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 0.9631061553955078,
+      "eval_runtime": 51.078,
+      "eval_samples_per_second": 42.445,
+      "eval_steps_per_second": 10.611,
+      "eval_wer": 0.7361183186299948,
+      "step": 900
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.3842124938964844,
+      "learning_rate": 0.0002854130488712478,
+      "loss": 1.3777,
+      "step": 901
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.1951818466186523,
+      "learning_rate": 0.00028537583726122545,
+      "loss": 1.0767,
+      "step": 902
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.5450639724731445,
+      "learning_rate": 0.00028533862565120315,
+      "loss": 0.714,
+      "step": 903
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.42718505859375,
+      "learning_rate": 0.0002853014140411808,
+      "loss": 0.5684,
+      "step": 904
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.6511759757995605,
+      "learning_rate": 0.0002852642024311585,
+      "loss": 1.1985,
+      "step": 905
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.286999225616455,
+      "learning_rate": 0.0002852269908211362,
+      "loss": 0.9587,
+      "step": 906
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.237370014190674,
+      "learning_rate": 0.00028518977921111386,
+      "loss": 0.4929,
+      "step": 907
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.4097900390625,
+      "learning_rate": 0.0002851525676010915,
+      "loss": 0.8108,
+      "step": 908
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.5256292819976807,
+      "learning_rate": 0.00028511535599106915,
+      "loss": 1.0193,
+      "step": 909
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.9239373207092285,
+      "learning_rate": 0.00028507814438104686,
+      "loss": 1.0205,
+      "step": 910
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 6.755927085876465,
+      "learning_rate": 0.00028504093277102456,
+      "loss": 1.2051,
+      "step": 911
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.922299385070801,
+      "learning_rate": 0.0002850037211610022,
+      "loss": 0.6658,
+      "step": 912
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.0863943099975586,
+      "learning_rate": 0.00028496650955097986,
+      "loss": 0.8485,
+      "step": 913
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.762369394302368,
+      "learning_rate": 0.00028492929794095756,
+      "loss": 0.8154,
+      "step": 914
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.955242395401001,
+      "learning_rate": 0.0002848920863309352,
+      "loss": 0.8395,
+      "step": 915
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.491308212280273,
+      "learning_rate": 0.0002848548747209129,
+      "loss": 0.7699,
+      "step": 916
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 6.529202938079834,
+      "learning_rate": 0.00028481766311089056,
+      "loss": 2.1599,
+      "step": 917
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.4373815059661865,
+      "learning_rate": 0.00028478045150086826,
+      "loss": 0.3231,
+      "step": 918
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 6.301473140716553,
+      "learning_rate": 0.0002847432398908459,
+      "loss": 0.4571,
+      "step": 919
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.855682373046875,
+      "learning_rate": 0.00028470602828082356,
+      "loss": 0.5254,
+      "step": 920
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.8739993572235107,
+      "learning_rate": 0.00028466881667080126,
+      "loss": 0.5581,
+      "step": 921
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.096916198730469,
+      "learning_rate": 0.0002846316050607789,
+      "loss": 0.8322,
+      "step": 922
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 6.568998336791992,
+      "learning_rate": 0.0002845943934507566,
+      "loss": 0.4273,
+      "step": 923
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 6.560142993927002,
+      "learning_rate": 0.0002845571818407343,
+      "loss": 0.9331,
+      "step": 924
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.5637927055358887,
+      "learning_rate": 0.00028451997023071197,
+      "loss": 0.3868,
+      "step": 925
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 6.0603837966918945,
+      "learning_rate": 0.0002844827586206896,
+      "loss": 1.7794,
+      "step": 926
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 6.264075756072998,
+      "learning_rate": 0.0002844455470106673,
+      "loss": 1.1889,
+      "step": 927
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.174050331115723,
+      "learning_rate": 0.00028440833540064497,
+      "loss": 0.9896,
+      "step": 928
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.8910465240478516,
+      "learning_rate": 0.00028437112379062267,
+      "loss": 0.7981,
+      "step": 929
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.8150928020477295,
+      "learning_rate": 0.0002843339121806003,
+      "loss": 0.9314,
+      "step": 930
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.244769096374512,
+      "learning_rate": 0.000284296700570578,
+      "loss": 0.7626,
+      "step": 931
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.8552732467651367,
+      "learning_rate": 0.00028425948896055567,
+      "loss": 0.6992,
+      "step": 932
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.5984230041503906,
+      "learning_rate": 0.0002842222773505333,
+      "loss": 1.0394,
+      "step": 933
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 7.442532062530518,
+      "learning_rate": 0.000284185065740511,
+      "loss": 1.0194,
+      "step": 934
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.221444606781006,
+      "learning_rate": 0.00028414785413048867,
+      "loss": 1.116,
+      "step": 935
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.7174394130706787,
+      "learning_rate": 0.00028411064252046637,
+      "loss": 1.2957,
+      "step": 936
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.641437530517578,
+      "learning_rate": 0.000284073430910444,
+      "loss": 0.9645,
+      "step": 937
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.738492965698242,
+      "learning_rate": 0.0002840362193004217,
+      "loss": 1.0473,
+      "step": 938
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.242377758026123,
+      "learning_rate": 0.00028399900769039937,
+      "loss": 0.9279,
+      "step": 939
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.320493221282959,
+      "learning_rate": 0.0002839617960803771,
+      "loss": 0.6088,
+      "step": 940
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.363595962524414,
+      "learning_rate": 0.0002839245844703547,
+      "loss": 0.9191,
+      "step": 941
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.4770991802215576,
+      "learning_rate": 0.0002838873728603324,
+      "loss": 0.6281,
+      "step": 942
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 7.542201519012451,
+      "learning_rate": 0.0002838501612503101,
+      "loss": 2.1469,
+      "step": 943
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 7.0690202713012695,
+      "learning_rate": 0.0002838129496402877,
+      "loss": 1.5071,
+      "step": 944
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.570958614349365,
+      "learning_rate": 0.00028377573803026543,
+      "loss": 0.7142,
+      "step": 945
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 6.251632213592529,
+      "learning_rate": 0.0002837385264202431,
+      "loss": 1.2639,
+      "step": 946
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.675748348236084,
+      "learning_rate": 0.0002837013148102208,
+      "loss": 0.9949,
+      "step": 947
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.878772258758545,
+      "learning_rate": 0.00028366410320019843,
+      "loss": 0.9438,
+      "step": 948
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.873948574066162,
+      "learning_rate": 0.00028362689159017613,
+      "loss": 0.8105,
+      "step": 949
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.9942407608032227,
+      "learning_rate": 0.0002835896799801538,
+      "loss": 0.7128,
+      "step": 950
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.542831897735596,
+      "learning_rate": 0.00028355246837013143,
+      "loss": 1.2922,
+      "step": 951
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.6374423503875732,
+      "learning_rate": 0.00028351525676010913,
+      "loss": 0.912,
+      "step": 952
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.016441822052002,
+      "learning_rate": 0.00028347804515008683,
+      "loss": 0.9093,
+      "step": 953
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.8477137088775635,
+      "learning_rate": 0.0002834408335400645,
+      "loss": 0.7698,
+      "step": 954
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.108532428741455,
+      "learning_rate": 0.00028340362193004213,
+      "loss": 1.0006,
+      "step": 955
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.331216335296631,
+      "learning_rate": 0.00028336641032001983,
+      "loss": 1.0243,
+      "step": 956
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.312037467956543,
+      "learning_rate": 0.0002833291987099975,
+      "loss": 0.6803,
+      "step": 957
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.1434242725372314,
+      "learning_rate": 0.0002832919870999752,
+      "loss": 0.3665,
+      "step": 958
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.74772834777832,
+      "learning_rate": 0.00028325477548995283,
+      "loss": 1.1212,
+      "step": 959
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.111370325088501,
+      "learning_rate": 0.00028321756387993054,
+      "loss": 0.4247,
+      "step": 960
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.159219741821289,
+      "learning_rate": 0.0002831803522699082,
+      "loss": 0.8041,
+      "step": 961
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.435075759887695,
+      "learning_rate": 0.00028314314065988583,
+      "loss": 2.0447,
+      "step": 962
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.009119987487793,
+      "learning_rate": 0.00028310592904986354,
+      "loss": 0.5044,
+      "step": 963
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.7699248790740967,
+      "learning_rate": 0.0002830687174398412,
+      "loss": 0.3857,
+      "step": 964
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.6067960262298584,
+      "learning_rate": 0.0002830315058298189,
+      "loss": 0.3094,
+      "step": 965
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.571464776992798,
+      "learning_rate": 0.00028299429421979654,
+      "loss": 0.3871,
+      "step": 966
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 9.788469314575195,
+      "learning_rate": 0.00028295708260977424,
+      "loss": 1.292,
+      "step": 967
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.8277058601379395,
+      "learning_rate": 0.0002829198709997519,
+      "loss": 0.5882,
+      "step": 968
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.459136486053467,
+      "learning_rate": 0.00028288265938972954,
+      "loss": 1.0885,
+      "step": 969
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.053255558013916,
+      "learning_rate": 0.00028284544777970724,
+      "loss": 0.4897,
+      "step": 970
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.948587417602539,
+      "learning_rate": 0.00028280823616968494,
+      "loss": 1.4657,
+      "step": 971
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 6.1781206130981445,
+      "learning_rate": 0.0002827710245596626,
+      "loss": 0.903,
+      "step": 972
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 7.313625812530518,
+      "learning_rate": 0.0002827338129496403,
+      "loss": 2.1275,
+      "step": 973
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.631922721862793,
+      "learning_rate": 0.00028269660133961794,
+      "loss": 1.1911,
+      "step": 974
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 11.282684326171875,
+      "learning_rate": 0.0002826593897295956,
+      "loss": 2.9956,
+      "step": 975
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 7.1073317527771,
+      "learning_rate": 0.0002826221781195733,
+      "loss": 2.037,
+      "step": 976
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 18.51368522644043,
+      "learning_rate": 0.00028258496650955094,
+      "loss": 2.3501,
+      "step": 977
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.6810758113861084,
+      "learning_rate": 0.00028254775489952865,
+      "loss": 1.236,
+      "step": 978
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.3135673999786377,
+      "learning_rate": 0.0002825105432895063,
+      "loss": 0.8256,
+      "step": 979
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.1581242084503174,
+      "learning_rate": 0.000282473331679484,
+      "loss": 0.7498,
+      "step": 980
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.0917932987213135,
+      "learning_rate": 0.00028243612006946165,
+      "loss": 0.7558,
+      "step": 981
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.050999164581299,
+      "learning_rate": 0.0002823989084594393,
+      "loss": 0.7646,
+      "step": 982
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.006269454956055,
+      "learning_rate": 0.000282361696849417,
+      "loss": 0.5423,
+      "step": 983
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.775635242462158,
+      "learning_rate": 0.0002823244852393947,
+      "loss": 0.5903,
+      "step": 984
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.8768789768218994,
+      "learning_rate": 0.00028228727362937235,
+      "loss": 0.8378,
+      "step": 985
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.105539321899414,
+      "learning_rate": 0.00028225006201935,
+      "loss": 0.7609,
+      "step": 986
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.927220344543457,
+      "learning_rate": 0.00028221285040932765,
+      "loss": 0.5829,
+      "step": 987
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.363030433654785,
+      "learning_rate": 0.00028217563879930535,
+      "loss": 0.5635,
+      "step": 988
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.495129108428955,
+      "learning_rate": 0.00028213842718928305,
+      "loss": 0.8921,
+      "step": 989
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.773660659790039,
+      "learning_rate": 0.0002821012155792607,
+      "loss": 1.5913,
+      "step": 990
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.911088228225708,
+      "learning_rate": 0.0002820640039692384,
+      "loss": 1.5924,
+      "step": 991
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.539827823638916,
+      "learning_rate": 0.00028202679235921605,
+      "loss": 0.8482,
+      "step": 992
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.734384298324585,
+      "learning_rate": 0.0002819895807491937,
+      "loss": 0.8819,
+      "step": 993
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.120278835296631,
+      "learning_rate": 0.0002819523691391714,
+      "loss": 0.8981,
+      "step": 994
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.272894382476807,
+      "learning_rate": 0.00028191515752914905,
+      "loss": 0.7838,
+      "step": 995
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.84804105758667,
+      "learning_rate": 0.00028187794591912676,
+      "loss": 0.46,
+      "step": 996
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 9.430734634399414,
+      "learning_rate": 0.0002818407343091044,
+      "loss": 0.6851,
+      "step": 997
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.4790775775909424,
+      "learning_rate": 0.0002818035226990821,
+      "loss": 0.6983,
+      "step": 998
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.3269646167755127,
+      "learning_rate": 0.00028176631108905976,
+      "loss": 0.2755,
+      "step": 999
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 9.969862937927246,
+      "learning_rate": 0.0002817290994790374,
+      "loss": 1.5394,
+      "step": 1000
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 0.8975178599357605,
+      "eval_runtime": 51.1763,
+      "eval_samples_per_second": 42.363,
+      "eval_steps_per_second": 10.591,
+      "eval_wer": 0.7244421380384016,
+      "step": 1000
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 8.002113342285156,
+      "learning_rate": 0.0002816918878690151,
+      "loss": 1.4136,
+      "step": 1001
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.763845920562744,
+      "learning_rate": 0.0002816546762589928,
+      "loss": 0.6863,
+      "step": 1002
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.5052077770233154,
+      "learning_rate": 0.00028161746464897046,
+      "loss": 1.2483,
+      "step": 1003
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.184037685394287,
+      "learning_rate": 0.0002815802530389481,
+      "loss": 0.7632,
+      "step": 1004
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.645569086074829,
+      "learning_rate": 0.0002815430414289258,
+      "loss": 0.5727,
+      "step": 1005
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.4027628898620605,
+      "learning_rate": 0.00028150582981890346,
+      "loss": 0.5045,
+      "step": 1006
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.2467432022094727,
+      "learning_rate": 0.00028146861820888116,
+      "loss": 0.4922,
+      "step": 1007
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.64123272895813,
+      "learning_rate": 0.0002814314065988588,
+      "loss": 0.5161,
+      "step": 1008
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.1751201152801514,
+      "learning_rate": 0.0002813941949888365,
+      "loss": 1.003,
+      "step": 1009
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.5006515979766846,
+      "learning_rate": 0.00028135698337881416,
+      "loss": 0.6148,
+      "step": 1010
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.372243881225586,
+      "learning_rate": 0.0002813197717687918,
+      "loss": 0.5718,
+      "step": 1011
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.657968044281006,
+      "learning_rate": 0.0002812825601587695,
+      "loss": 1.1868,
+      "step": 1012
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.172140598297119,
+      "learning_rate": 0.00028124534854874716,
+      "loss": 0.7887,
+      "step": 1013
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 4.085210800170898,
+      "learning_rate": 0.00028120813693872487,
+      "loss": 0.9341,
+      "step": 1014
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.6052396297454834,
+      "learning_rate": 0.00028117092532870257,
+      "loss": 0.6269,
+      "step": 1015
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.70512580871582,
+      "learning_rate": 0.0002811337137186802,
+      "loss": 1.2359,
+      "step": 1016
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 8.653592109680176,
+      "learning_rate": 0.00028109650210865787,
+      "loss": 1.07,
+      "step": 1017
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.807731628417969,
+      "learning_rate": 0.0002810592904986355,
+      "loss": 0.8944,
+      "step": 1018
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.5197043418884277,
+      "learning_rate": 0.0002810220788886132,
+      "loss": 0.9289,
+      "step": 1019
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.4789111614227295,
+      "learning_rate": 0.0002809848672785909,
+      "loss": 0.9998,
+      "step": 1020
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.096599102020264,
+      "learning_rate": 0.00028094765566856857,
+      "loss": 0.939,
+      "step": 1021
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.344417095184326,
+      "learning_rate": 0.0002809104440585463,
+      "loss": 1.0038,
+      "step": 1022
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.7382874488830566,
+      "learning_rate": 0.0002808732324485239,
+      "loss": 1.4407,
+      "step": 1023
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.9947118759155273,
+      "learning_rate": 0.00028083602083850157,
+      "loss": 0.5359,
+      "step": 1024
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.5792839527130127,
+      "learning_rate": 0.0002807988092284793,
+      "loss": 0.5672,
+      "step": 1025
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.1083927154541016,
+      "learning_rate": 0.0002807615976184569,
+      "loss": 0.9157,
+      "step": 1026
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.818694829940796,
+      "learning_rate": 0.0002807243860084346,
+      "loss": 0.9535,
+      "step": 1027
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.34951114654541,
+      "learning_rate": 0.0002806871743984123,
+      "loss": 1.2982,
+      "step": 1028
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.1415767669677734,
+      "learning_rate": 0.0002806499627883899,
+      "loss": 0.6431,
+      "step": 1029
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.301403284072876,
+      "learning_rate": 0.0002806127511783676,
+      "loss": 1.0773,
+      "step": 1030
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.992616653442383,
+      "learning_rate": 0.0002805755395683453,
+      "loss": 0.9013,
+      "step": 1031
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.0520200729370117,
+      "learning_rate": 0.000280538327958323,
+      "loss": 0.5579,
+      "step": 1032
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.932624340057373,
+      "learning_rate": 0.0002805011163483007,
+      "loss": 1.0126,
+      "step": 1033
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.583620309829712,
+      "learning_rate": 0.00028046390473827833,
+      "loss": 0.5483,
+      "step": 1034
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.1285436153411865,
+      "learning_rate": 0.000280426693128256,
+      "loss": 0.4759,
+      "step": 1035
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.9035234451293945,
+      "learning_rate": 0.0002803894815182336,
+      "loss": 0.3353,
+      "step": 1036
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.205357789993286,
+      "learning_rate": 0.00028035226990821133,
+      "loss": 1.0646,
+      "step": 1037
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 4.966174602508545,
+      "learning_rate": 0.00028031505829818903,
+      "loss": 2.3098,
+      "step": 1038
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.033513069152832,
+      "learning_rate": 0.0002802778466881667,
+      "loss": 0.9838,
+      "step": 1039
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.171358823776245,
+      "learning_rate": 0.0002802406350781444,
+      "loss": 0.5542,
+      "step": 1040
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.6895203590393066,
+      "learning_rate": 0.00028020342346812203,
+      "loss": 0.8648,
+      "step": 1041
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.115908145904541,
+      "learning_rate": 0.0002801662118580997,
+      "loss": 0.8138,
+      "step": 1042
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.41919469833374,
+      "learning_rate": 0.0002801290002480774,
+      "loss": 0.6559,
+      "step": 1043
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.736941337585449,
+      "learning_rate": 0.00028009178863805503,
+      "loss": 0.7366,
+      "step": 1044
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.1285319328308105,
+      "learning_rate": 0.00028005457702803274,
+      "loss": 0.5219,
+      "step": 1045
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.831568241119385,
+      "learning_rate": 0.0002800173654180104,
+      "loss": 0.5053,
+      "step": 1046
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.898710250854492,
+      "learning_rate": 0.0002799801538079881,
+      "loss": 1.0036,
+      "step": 1047
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.523413181304932,
+      "learning_rate": 0.00027994294219796574,
+      "loss": 1.398,
+      "step": 1048
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.2347447872161865,
+      "learning_rate": 0.0002799057305879434,
+      "loss": 0.4556,
+      "step": 1049
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.337017059326172,
+      "learning_rate": 0.0002798685189779211,
+      "loss": 1.328,
+      "step": 1050
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.164899826049805,
+      "learning_rate": 0.0002798313073678988,
+      "loss": 1.2521,
+      "step": 1051
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6211090087890625,
+      "learning_rate": 0.00027979409575787644,
+      "loss": 1.0136,
+      "step": 1052
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.821180820465088,
+      "learning_rate": 0.0002797568841478541,
+      "loss": 0.9512,
+      "step": 1053
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.4610302448272705,
+      "learning_rate": 0.00027971967253783174,
+      "loss": 0.7071,
+      "step": 1054
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.7207629680633545,
+      "learning_rate": 0.00027968246092780944,
+      "loss": 0.6617,
+      "step": 1055
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.068587779998779,
+      "learning_rate": 0.00027964524931778714,
+      "loss": 0.7244,
+      "step": 1056
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.698821544647217,
+      "learning_rate": 0.0002796080377077648,
+      "loss": 0.8169,
+      "step": 1057
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.56290864944458,
+      "learning_rate": 0.0002795708260977425,
+      "loss": 0.5039,
+      "step": 1058
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.410244941711426,
+      "learning_rate": 0.00027953361448772014,
+      "loss": 0.7284,
+      "step": 1059
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4827516078948975,
+      "learning_rate": 0.0002794964028776978,
+      "loss": 0.675,
+      "step": 1060
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.349926471710205,
+      "learning_rate": 0.0002794591912676755,
+      "loss": 0.3923,
+      "step": 1061
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.622032165527344,
+      "learning_rate": 0.00027942197965765314,
+      "loss": 0.7249,
+      "step": 1062
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.842290878295898,
+      "learning_rate": 0.00027938476804763085,
+      "loss": 0.5359,
+      "step": 1063
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.4208180904388428,
+      "learning_rate": 0.00027934755643760855,
+      "loss": 1.2509,
+      "step": 1064
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.2167344093322754,
+      "learning_rate": 0.0002793103448275862,
+      "loss": 0.3897,
+      "step": 1065
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 6.12382698059082,
+      "learning_rate": 0.00027927313321756385,
+      "loss": 0.8148,
+      "step": 1066
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.50907301902771,
+      "learning_rate": 0.0002792359216075415,
+      "loss": 0.3775,
+      "step": 1067
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.680488109588623,
+      "learning_rate": 0.0002791987099975192,
+      "loss": 0.7939,
+      "step": 1068
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.077605247497559,
+      "learning_rate": 0.0002791614983874969,
+      "loss": 0.8747,
+      "step": 1069
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 8.638588905334473,
+      "learning_rate": 0.00027912428677747455,
+      "loss": 1.3101,
+      "step": 1070
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 6.577147006988525,
+      "learning_rate": 0.0002790870751674522,
+      "loss": 2.0627,
+      "step": 1071
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.956910133361816,
+      "learning_rate": 0.0002790498635574299,
+      "loss": 1.1331,
+      "step": 1072
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 8.133764266967773,
+      "learning_rate": 0.00027901265194740755,
+      "loss": 1.0287,
+      "step": 1073
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 11.1182279586792,
+      "learning_rate": 0.00027897544033738525,
+      "loss": 1.0364,
+      "step": 1074
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.266451835632324,
+      "learning_rate": 0.0002789382287273629,
+      "loss": 0.6508,
+      "step": 1075
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 14.238346099853516,
+      "learning_rate": 0.0002789010171173406,
+      "loss": 2.2819,
+      "step": 1076
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.7446582317352295,
+      "learning_rate": 0.00027886380550731825,
+      "loss": 0.922,
+      "step": 1077
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.616486072540283,
+      "learning_rate": 0.0002788265938972959,
+      "loss": 1.1808,
+      "step": 1078
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.3517651557922363,
+      "learning_rate": 0.0002787893822872736,
+      "loss": 1.3126,
+      "step": 1079
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.9327385425567627,
+      "learning_rate": 0.00027875217067725125,
+      "loss": 0.805,
+      "step": 1080
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.411552906036377,
+      "learning_rate": 0.00027871495906722896,
+      "loss": 0.8779,
+      "step": 1081
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.122033596038818,
+      "learning_rate": 0.00027867774745720666,
+      "loss": 1.3039,
+      "step": 1082
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.9304003715515137,
+      "learning_rate": 0.0002786405358471843,
+      "loss": 1.1372,
+      "step": 1083
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.756955862045288,
+      "learning_rate": 0.00027860332423716196,
+      "loss": 1.0457,
+      "step": 1084
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.7756242752075195,
+      "learning_rate": 0.00027856611262713966,
+      "loss": 0.8883,
+      "step": 1085
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.649970054626465,
+      "learning_rate": 0.0002785289010171173,
+      "loss": 2.0304,
+      "step": 1086
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.984282970428467,
+      "learning_rate": 0.000278491689407095,
+      "loss": 1.1931,
+      "step": 1087
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.8974289894104,
+      "learning_rate": 0.00027845447779707266,
+      "loss": 0.7087,
+      "step": 1088
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.27323579788208,
+      "learning_rate": 0.00027841726618705036,
+      "loss": 0.4356,
+      "step": 1089
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 6.953647136688232,
+      "learning_rate": 0.000278380054577028,
+      "loss": 1.3964,
+      "step": 1090
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.8305001258850098,
+      "learning_rate": 0.00027834284296700566,
+      "loss": 0.6397,
+      "step": 1091
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.174875736236572,
+      "learning_rate": 0.00027830563135698336,
+      "loss": 0.6587,
+      "step": 1092
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.3889546394348145,
+      "learning_rate": 0.000278268419746961,
+      "loss": 0.6804,
+      "step": 1093
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 6.435478210449219,
+      "learning_rate": 0.0002782312081369387,
+      "loss": 1.0459,
+      "step": 1094
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 9.19431209564209,
+      "learning_rate": 0.00027819399652691636,
+      "loss": 1.5923,
+      "step": 1095
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.4626758098602295,
+      "learning_rate": 0.000278156784916894,
+      "loss": 0.7921,
+      "step": 1096
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.5386457443237305,
+      "learning_rate": 0.0002781195733068717,
+      "loss": 0.9378,
+      "step": 1097
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.50201940536499,
+      "learning_rate": 0.00027808236169684936,
+      "loss": 1.7841,
+      "step": 1098
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.9906821250915527,
+      "learning_rate": 0.00027804515008682707,
+      "loss": 0.7195,
+      "step": 1099
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 6.178668022155762,
+      "learning_rate": 0.00027800793847680477,
+      "loss": 0.5866,
+      "step": 1100
+    },
+    {
+      "epoch": 0.39,
+      "eval_loss": 0.858989417552948,
+      "eval_runtime": 50.9765,
+      "eval_samples_per_second": 42.529,
+      "eval_steps_per_second": 10.632,
+      "eval_wer": 0.7215014703338523,
+      "step": 1100
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.665261268615723,
+      "learning_rate": 0.0002779707268667824,
+      "loss": 1.8917,
+      "step": 1101
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.5688321590423584,
+      "learning_rate": 0.00027793351525676007,
+      "loss": 1.1618,
+      "step": 1102
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.149432420730591,
+      "learning_rate": 0.00027789630364673777,
+      "loss": 0.7372,
+      "step": 1103
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.077228307723999,
+      "learning_rate": 0.0002778590920367154,
+      "loss": 0.7426,
+      "step": 1104
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.8044614791870117,
+      "learning_rate": 0.0002778218804266931,
+      "loss": 1.0178,
+      "step": 1105
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.4295260906219482,
+      "learning_rate": 0.00027778466881667077,
+      "loss": 0.6801,
+      "step": 1106
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.1818017959594727,
+      "learning_rate": 0.00027774745720664847,
+      "loss": 0.5587,
+      "step": 1107
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.3339715003967285,
+      "learning_rate": 0.0002777102455966261,
+      "loss": 0.7365,
+      "step": 1108
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.128452301025391,
+      "learning_rate": 0.00027767303398660377,
+      "loss": 1.1199,
+      "step": 1109
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.8575356006622314,
+      "learning_rate": 0.00027763582237658147,
+      "loss": 1.1287,
+      "step": 1110
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.410104751586914,
+      "learning_rate": 0.0002775986107665591,
+      "loss": 0.6495,
+      "step": 1111
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.475276947021484,
+      "learning_rate": 0.0002775613991565368,
+      "loss": 0.7845,
+      "step": 1112
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.5406651496887207,
+      "learning_rate": 0.00027752418754651447,
+      "loss": 0.476,
+      "step": 1113
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.599978446960449,
+      "learning_rate": 0.0002774869759364922,
+      "loss": 0.9411,
+      "step": 1114
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.8049709796905518,
+      "learning_rate": 0.0002774497643264698,
+      "loss": 0.7746,
+      "step": 1115
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 5.1293182373046875,
+      "learning_rate": 0.00027741255271644753,
+      "loss": 1.4263,
+      "step": 1116
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.558226585388184,
+      "learning_rate": 0.0002773753411064252,
+      "loss": 0.8449,
+      "step": 1117
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.3060882091522217,
+      "learning_rate": 0.0002773381294964029,
+      "loss": 0.7272,
+      "step": 1118
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 6.07926607131958,
+      "learning_rate": 0.00027730091788638053,
+      "loss": 0.9118,
+      "step": 1119
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.133337020874023,
+      "learning_rate": 0.0002772637062763582,
+      "loss": 0.483,
+      "step": 1120
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 5.365363121032715,
+      "learning_rate": 0.0002772264946663359,
+      "loss": 0.8083,
+      "step": 1121
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.9955897331237793,
+      "learning_rate": 0.00027718928305631353,
+      "loss": 1.1863,
+      "step": 1122
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.5983023643493652,
+      "learning_rate": 0.00027715207144629123,
+      "loss": 0.7021,
+      "step": 1123
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.3443710803985596,
+      "learning_rate": 0.0002771148598362689,
+      "loss": 0.6462,
+      "step": 1124
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.6084046363830566,
+      "learning_rate": 0.0002770776482262466,
+      "loss": 0.723,
+      "step": 1125
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.003911256790161,
+      "learning_rate": 0.00027704043661622423,
+      "loss": 1.1826,
+      "step": 1126
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.219900608062744,
+      "learning_rate": 0.0002770032250062019,
+      "loss": 1.0835,
+      "step": 1127
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.942505121231079,
+      "learning_rate": 0.0002769660133961796,
+      "loss": 0.9503,
+      "step": 1128
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.2380802631378174,
+      "learning_rate": 0.0002769288017861573,
+      "loss": 0.8017,
+      "step": 1129
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.9053637981414795,
+      "learning_rate": 0.00027689159017613493,
+      "loss": 0.6966,
+      "step": 1130
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.158212900161743,
+      "learning_rate": 0.00027685437856611264,
+      "loss": 0.4872,
+      "step": 1131
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.566776990890503,
+      "learning_rate": 0.0002768171669560903,
+      "loss": 0.7114,
+      "step": 1132
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0422821044921875,
+      "learning_rate": 0.00027677995534606793,
+      "loss": 0.6142,
+      "step": 1133
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.843910813331604,
+      "learning_rate": 0.00027674274373604564,
+      "loss": 0.4941,
+      "step": 1134
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.197638034820557,
+      "learning_rate": 0.0002767055321260233,
+      "loss": 1.1024,
+      "step": 1135
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.642378568649292,
+      "learning_rate": 0.000276668320516001,
+      "loss": 0.5026,
+      "step": 1136
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.609485626220703,
+      "learning_rate": 0.00027663110890597864,
+      "loss": 1.2151,
+      "step": 1137
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2409887313842773,
+      "learning_rate": 0.0002765938972959563,
+      "loss": 0.3389,
+      "step": 1138
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.53881311416626,
+      "learning_rate": 0.000276556685685934,
+      "loss": 0.9275,
+      "step": 1139
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 5.820657730102539,
+      "learning_rate": 0.00027651947407591164,
+      "loss": 0.7122,
+      "step": 1140
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.550056457519531,
+      "learning_rate": 0.00027648226246588934,
+      "loss": 1.9592,
+      "step": 1141
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.5938735008239746,
+      "learning_rate": 0.000276445050855867,
+      "loss": 0.765,
+      "step": 1142
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.5738070011138916,
+      "learning_rate": 0.0002764078392458447,
+      "loss": 0.6004,
+      "step": 1143
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.4456381797790527,
+      "learning_rate": 0.00027637062763582234,
+      "loss": 0.5931,
+      "step": 1144
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 7.033105850219727,
+      "learning_rate": 0.0002763334160258,
+      "loss": 0.8942,
+      "step": 1145
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.166602611541748,
+      "learning_rate": 0.0002762962044157777,
+      "loss": 0.5482,
+      "step": 1146
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.609311103820801,
+      "learning_rate": 0.0002762589928057554,
+      "loss": 0.9272,
+      "step": 1147
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.320896625518799,
+      "learning_rate": 0.00027622178119573304,
+      "loss": 0.6604,
+      "step": 1148
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 6.4265875816345215,
+      "learning_rate": 0.00027618456958571075,
+      "loss": 1.238,
+      "step": 1149
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 8.044411659240723,
+      "learning_rate": 0.0002761473579756884,
+      "loss": 0.9712,
+      "step": 1150
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 7.890890121459961,
+      "learning_rate": 0.00027611014636566604,
+      "loss": 2.561,
+      "step": 1151
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.3083295822143555,
+      "learning_rate": 0.00027607293475564375,
+      "loss": 1.3886,
+      "step": 1152
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.2230350971221924,
+      "learning_rate": 0.0002760357231456214,
+      "loss": 0.9359,
+      "step": 1153
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.595873832702637,
+      "learning_rate": 0.0002759985115355991,
+      "loss": 1.0819,
+      "step": 1154
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.82143497467041,
+      "learning_rate": 0.00027596129992557675,
+      "loss": 0.9526,
+      "step": 1155
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.152099132537842,
+      "learning_rate": 0.00027592408831555445,
+      "loss": 0.5482,
+      "step": 1156
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.9585154056549072,
+      "learning_rate": 0.0002758868767055321,
+      "loss": 1.1054,
+      "step": 1157
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.825305700302124,
+      "learning_rate": 0.00027584966509550975,
+      "loss": 0.7547,
+      "step": 1158
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.47904634475708,
+      "learning_rate": 0.00027581245348548745,
+      "loss": 0.7835,
+      "step": 1159
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.429239511489868,
+      "learning_rate": 0.00027577524187546515,
+      "loss": 1.1264,
+      "step": 1160
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.2548816204071045,
+      "learning_rate": 0.0002757380302654428,
+      "loss": 0.7843,
+      "step": 1161
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 5.171494007110596,
+      "learning_rate": 0.00027570081865542045,
+      "loss": 0.8388,
+      "step": 1162
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.407280921936035,
+      "learning_rate": 0.00027566360704539815,
+      "loss": 0.5067,
+      "step": 1163
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2985994815826416,
+      "learning_rate": 0.0002756263954353758,
+      "loss": 0.7189,
+      "step": 1164
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.708470106124878,
+      "learning_rate": 0.0002755891838253535,
+      "loss": 0.7342,
+      "step": 1165
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.204012393951416,
+      "learning_rate": 0.00027555197221533115,
+      "loss": 0.5299,
+      "step": 1166
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.998049259185791,
+      "learning_rate": 0.00027551476060530886,
+      "loss": 0.3688,
+      "step": 1167
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.720385789871216,
+      "learning_rate": 0.0002754775489952865,
+      "loss": 0.7139,
+      "step": 1168
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.8262248039245605,
+      "learning_rate": 0.00027544033738526415,
+      "loss": 0.4921,
+      "step": 1169
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.7257118225097656,
+      "learning_rate": 0.00027540312577524186,
+      "loss": 0.7639,
+      "step": 1170
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 5.001168251037598,
+      "learning_rate": 0.0002753659141652195,
+      "loss": 0.6115,
+      "step": 1171
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.162776470184326,
+      "learning_rate": 0.0002753287025551972,
+      "loss": 0.2777,
+      "step": 1172
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 5.089084148406982,
+      "learning_rate": 0.0002752914909451749,
+      "loss": 1.4753,
+      "step": 1173
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.25437068939209,
+      "learning_rate": 0.00027525427933515256,
+      "loss": 0.2112,
+      "step": 1174
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 5.870993137359619,
+      "learning_rate": 0.0002752170677251302,
+      "loss": 0.6987,
+      "step": 1175
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.5580074787139893,
+      "learning_rate": 0.00027517985611510786,
+      "loss": 0.8806,
+      "step": 1176
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 5.758315563201904,
+      "learning_rate": 0.00027514264450508556,
+      "loss": 1.0525,
+      "step": 1177
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.0043349266052246,
+      "learning_rate": 0.00027510543289506326,
+      "loss": 1.0657,
+      "step": 1178
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.536125183105469,
+      "learning_rate": 0.0002750682212850409,
+      "loss": 1.1375,
+      "step": 1179
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 22.37661361694336,
+      "learning_rate": 0.00027503100967501856,
+      "loss": 4.7936,
+      "step": 1180
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.035590171813965,
+      "learning_rate": 0.00027499379806499626,
+      "loss": 0.9853,
+      "step": 1181
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.8873214721679688,
+      "learning_rate": 0.0002749565864549739,
+      "loss": 0.5476,
+      "step": 1182
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.164041042327881,
+      "learning_rate": 0.0002749193748449516,
+      "loss": 0.3647,
+      "step": 1183
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.414952039718628,
+      "learning_rate": 0.00027488216323492926,
+      "loss": 1.0755,
+      "step": 1184
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.234614849090576,
+      "learning_rate": 0.00027484495162490697,
+      "loss": 1.3185,
+      "step": 1185
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.8845980167388916,
+      "learning_rate": 0.0002748077400148846,
+      "loss": 0.7329,
+      "step": 1186
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.165452718734741,
+      "learning_rate": 0.00027477052840486226,
+      "loss": 0.4976,
+      "step": 1187
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.6230342388153076,
+      "learning_rate": 0.00027473331679483997,
+      "loss": 1.2895,
+      "step": 1188
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.947673797607422,
+      "learning_rate": 0.0002746961051848176,
+      "loss": 0.7188,
+      "step": 1189
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.158074140548706,
+      "learning_rate": 0.0002746588935747953,
+      "loss": 0.608,
+      "step": 1190
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.541929006576538,
+      "learning_rate": 0.000274621681964773,
+      "loss": 0.6903,
+      "step": 1191
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1511118412017822,
+      "learning_rate": 0.00027458447035475067,
+      "loss": 0.3223,
+      "step": 1192
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.088685989379883,
+      "learning_rate": 0.0002745472587447283,
+      "loss": 0.9799,
+      "step": 1193
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.290481090545654,
+      "learning_rate": 0.00027451004713470597,
+      "loss": 0.7953,
+      "step": 1194
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.309513092041016,
+      "learning_rate": 0.00027447283552468367,
+      "loss": 0.7685,
+      "step": 1195
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.8015859127044678,
+      "learning_rate": 0.0002744356239146614,
+      "loss": 0.5714,
+      "step": 1196
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 5.4302568435668945,
+      "learning_rate": 0.000274398412304639,
+      "loss": 0.6369,
+      "step": 1197
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.336185932159424,
+      "learning_rate": 0.0002743612006946167,
+      "loss": 0.3784,
+      "step": 1198
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.980928659439087,
+      "learning_rate": 0.0002743239890845944,
+      "loss": 0.5137,
+      "step": 1199
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 7.486271381378174,
+      "learning_rate": 0.000274286777474572,
+      "loss": 0.4206,
+      "step": 1200
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 0.7752430438995361,
+      "eval_runtime": 49.918,
+      "eval_samples_per_second": 43.431,
+      "eval_steps_per_second": 10.858,
+      "eval_wer": 0.6498875627054143,
+      "step": 1200
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.420276641845703,
+      "learning_rate": 0.0002742495658645497,
+      "loss": 1.3815,
+      "step": 1201
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.3125431537628174,
+      "learning_rate": 0.0002742123542545274,
+      "loss": 1.3345,
+      "step": 1202
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 6.291903495788574,
+      "learning_rate": 0.0002741751426445051,
+      "loss": 1.1505,
+      "step": 1203
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.3073489665985107,
+      "learning_rate": 0.0002741379310344827,
+      "loss": 0.5659,
+      "step": 1204
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.7754597663879395,
+      "learning_rate": 0.00027410071942446043,
+      "loss": 0.8152,
+      "step": 1205
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.6730542182922363,
+      "learning_rate": 0.0002740635078144381,
+      "loss": 1.41,
+      "step": 1206
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.539772987365723,
+      "learning_rate": 0.0002740262962044157,
+      "loss": 1.0475,
+      "step": 1207
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.224508762359619,
+      "learning_rate": 0.00027398908459439343,
+      "loss": 0.5957,
+      "step": 1208
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.530787467956543,
+      "learning_rate": 0.00027395187298437113,
+      "loss": 0.6799,
+      "step": 1209
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.891000509262085,
+      "learning_rate": 0.0002739146613743488,
+      "loss": 1.2861,
+      "step": 1210
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.0431253910064697,
+      "learning_rate": 0.00027387744976432643,
+      "loss": 0.5401,
+      "step": 1211
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.738537549972534,
+      "learning_rate": 0.0002738402381543041,
+      "loss": 0.8909,
+      "step": 1212
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.205295085906982,
+      "learning_rate": 0.0002738030265442818,
+      "loss": 0.8714,
+      "step": 1213
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5161707401275635,
+      "learning_rate": 0.0002737658149342595,
+      "loss": 0.5855,
+      "step": 1214
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.841719150543213,
+      "learning_rate": 0.00027372860332423713,
+      "loss": 0.9467,
+      "step": 1215
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.4140450954437256,
+      "learning_rate": 0.00027369139171421484,
+      "loss": 1.0405,
+      "step": 1216
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.440001487731934,
+      "learning_rate": 0.0002736541801041925,
+      "loss": 0.5273,
+      "step": 1217
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.5466346740722656,
+      "learning_rate": 0.00027361696849417013,
+      "loss": 0.8691,
+      "step": 1218
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.198976516723633,
+      "learning_rate": 0.00027357975688414784,
+      "loss": 0.9419,
+      "step": 1219
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.357349395751953,
+      "learning_rate": 0.0002735425452741255,
+      "loss": 0.7402,
+      "step": 1220
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.02566385269165,
+      "learning_rate": 0.0002735053336641032,
+      "loss": 0.7176,
+      "step": 1221
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 6.563841819763184,
+      "learning_rate": 0.00027346812205408084,
+      "loss": 2.2978,
+      "step": 1222
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.9487085342407227,
+      "learning_rate": 0.00027343091044405854,
+      "loss": 0.6593,
+      "step": 1223
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.61473274230957,
+      "learning_rate": 0.0002733936988340362,
+      "loss": 1.9731,
+      "step": 1224
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.610717296600342,
+      "learning_rate": 0.00027335648722401384,
+      "loss": 2.6307,
+      "step": 1225
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.044619083404541,
+      "learning_rate": 0.00027331927561399154,
+      "loss": 1.2597,
+      "step": 1226
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.211724758148193,
+      "learning_rate": 0.00027328206400396924,
+      "loss": 1.6048,
+      "step": 1227
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5721964836120605,
+      "learning_rate": 0.0002732448523939469,
+      "loss": 0.686,
+      "step": 1228
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.3949391841888428,
+      "learning_rate": 0.00027320764078392454,
+      "loss": 0.6873,
+      "step": 1229
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 18.452733993530273,
+      "learning_rate": 0.00027317042917390224,
+      "loss": 3.4468,
+      "step": 1230
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.9526307582855225,
+      "learning_rate": 0.0002731332175638799,
+      "loss": 0.6906,
+      "step": 1231
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.902743339538574,
+      "learning_rate": 0.0002730960059538576,
+      "loss": 0.5812,
+      "step": 1232
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.806962013244629,
+      "learning_rate": 0.00027305879434383524,
+      "loss": 0.6684,
+      "step": 1233
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.608047008514404,
+      "learning_rate": 0.00027302158273381295,
+      "loss": 0.8382,
+      "step": 1234
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 8.201254844665527,
+      "learning_rate": 0.0002729843711237906,
+      "loss": 0.645,
+      "step": 1235
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.0138301849365234,
+      "learning_rate": 0.00027294715951376824,
+      "loss": 0.7516,
+      "step": 1236
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.389562606811523,
+      "learning_rate": 0.00027290994790374595,
+      "loss": 1.2125,
+      "step": 1237
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5127413272857666,
+      "learning_rate": 0.0002728727362937236,
+      "loss": 0.4319,
+      "step": 1238
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.1524224281311035,
+      "learning_rate": 0.0002728355246837013,
+      "loss": 0.586,
+      "step": 1239
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.574815273284912,
+      "learning_rate": 0.000272798313073679,
+      "loss": 0.8344,
+      "step": 1240
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.405158996582031,
+      "learning_rate": 0.00027276110146365665,
+      "loss": 0.8623,
+      "step": 1241
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.2911202907562256,
+      "learning_rate": 0.0002727238898536343,
+      "loss": 1.0748,
+      "step": 1242
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.5247392654418945,
+      "learning_rate": 0.00027268667824361195,
+      "loss": 0.5555,
+      "step": 1243
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.708747386932373,
+      "learning_rate": 0.00027264946663358965,
+      "loss": 0.988,
+      "step": 1244
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.7301928997039795,
+      "learning_rate": 0.00027261225502356735,
+      "loss": 0.7501,
+      "step": 1245
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.694626331329346,
+      "learning_rate": 0.000272575043413545,
+      "loss": 1.3788,
+      "step": 1246
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.803793907165527,
+      "learning_rate": 0.0002725378318035227,
+      "loss": 0.8157,
+      "step": 1247
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.017162561416626,
+      "learning_rate": 0.00027250062019350035,
+      "loss": 0.7995,
+      "step": 1248
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.8873093128204346,
+      "learning_rate": 0.000272463408583478,
+      "loss": 1.574,
+      "step": 1249
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.3652150630950928,
+      "learning_rate": 0.0002724261969734557,
+      "loss": 0.3168,
+      "step": 1250
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.411014556884766,
+      "learning_rate": 0.00027238898536343335,
+      "loss": 1.3121,
+      "step": 1251
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.4930648803710938,
+      "learning_rate": 0.00027235177375341106,
+      "loss": 0.7999,
+      "step": 1252
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.38267183303833,
+      "learning_rate": 0.0002723145621433887,
+      "loss": 0.9561,
+      "step": 1253
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.609616756439209,
+      "learning_rate": 0.00027227735053336635,
+      "loss": 0.8172,
+      "step": 1254
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.3476998805999756,
+      "learning_rate": 0.00027224013892334406,
+      "loss": 1.4765,
+      "step": 1255
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.823747158050537,
+      "learning_rate": 0.0002722029273133217,
+      "loss": 0.5941,
+      "step": 1256
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.738922595977783,
+      "learning_rate": 0.0002721657157032994,
+      "loss": 0.8431,
+      "step": 1257
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.6117281913757324,
+      "learning_rate": 0.0002721285040932771,
+      "loss": 0.8236,
+      "step": 1258
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.166642665863037,
+      "learning_rate": 0.00027209129248325476,
+      "loss": 0.8192,
+      "step": 1259
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.9680120944976807,
+      "learning_rate": 0.0002720540808732324,
+      "loss": 0.5717,
+      "step": 1260
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.276177406311035,
+      "learning_rate": 0.0002720168692632101,
+      "loss": 0.9233,
+      "step": 1261
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.2780745029449463,
+      "learning_rate": 0.00027197965765318776,
+      "loss": 0.7867,
+      "step": 1262
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.421667575836182,
+      "learning_rate": 0.00027194244604316546,
+      "loss": 0.8406,
+      "step": 1263
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.3682520389556885,
+      "learning_rate": 0.0002719052344331431,
+      "loss": 0.5538,
+      "step": 1264
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.914537191390991,
+      "learning_rate": 0.0002718680228231208,
+      "loss": 0.7114,
+      "step": 1265
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.220479965209961,
+      "learning_rate": 0.00027183081121309846,
+      "loss": 0.8671,
+      "step": 1266
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.633809566497803,
+      "learning_rate": 0.0002717935996030761,
+      "loss": 1.037,
+      "step": 1267
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.3338539600372314,
+      "learning_rate": 0.0002717563879930538,
+      "loss": 0.5996,
+      "step": 1268
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 6.159158706665039,
+      "learning_rate": 0.00027171917638303146,
+      "loss": 0.692,
+      "step": 1269
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.33358097076416,
+      "learning_rate": 0.00027168196477300917,
+      "loss": 0.6227,
+      "step": 1270
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.7529525756835938,
+      "learning_rate": 0.0002716447531629868,
+      "loss": 0.5452,
+      "step": 1271
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.7959513664245605,
+      "learning_rate": 0.0002716075415529645,
+      "loss": 0.6587,
+      "step": 1272
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.419649600982666,
+      "learning_rate": 0.00027157032994294217,
+      "loss": 0.8046,
+      "step": 1273
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.8088862895965576,
+      "learning_rate": 0.00027153311833291987,
+      "loss": 0.8147,
+      "step": 1274
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.987424373626709,
+      "learning_rate": 0.0002714959067228975,
+      "loss": 0.8777,
+      "step": 1275
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.6485095024108887,
+      "learning_rate": 0.0002714586951128752,
+      "loss": 1.4681,
+      "step": 1276
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.164416313171387,
+      "learning_rate": 0.00027142148350285287,
+      "loss": 0.8455,
+      "step": 1277
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.422079086303711,
+      "learning_rate": 0.0002713842718928305,
+      "loss": 1.0435,
+      "step": 1278
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.755207061767578,
+      "learning_rate": 0.0002713470602828082,
+      "loss": 0.6229,
+      "step": 1279
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.0500385761260986,
+      "learning_rate": 0.00027130984867278587,
+      "loss": 0.6549,
+      "step": 1280
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.116751194000244,
+      "learning_rate": 0.00027127263706276357,
+      "loss": 0.6984,
+      "step": 1281
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.164412498474121,
+      "learning_rate": 0.0002712354254527412,
+      "loss": 0.6526,
+      "step": 1282
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.9012765884399414,
+      "learning_rate": 0.0002711982138427189,
+      "loss": 0.7316,
+      "step": 1283
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.856905937194824,
+      "learning_rate": 0.00027116100223269657,
+      "loss": 0.6951,
+      "step": 1284
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.1080093383789062,
+      "learning_rate": 0.0002711237906226742,
+      "loss": 0.5701,
+      "step": 1285
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.3872121572494507,
+      "learning_rate": 0.0002710865790126519,
+      "loss": 0.3921,
+      "step": 1286
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.119579792022705,
+      "learning_rate": 0.0002710493674026296,
+      "loss": 0.4822,
+      "step": 1287
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.3475329875946045,
+      "learning_rate": 0.0002710121557926073,
+      "loss": 0.4675,
+      "step": 1288
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.7075090408325195,
+      "learning_rate": 0.000270974944182585,
+      "loss": 0.8302,
+      "step": 1289
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.20393180847168,
+      "learning_rate": 0.00027093773257256263,
+      "loss": 1.3705,
+      "step": 1290
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.181398391723633,
+      "learning_rate": 0.0002709005209625403,
+      "loss": 0.5417,
+      "step": 1291
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": NaN,
+      "learning_rate": 0.0002709005209625403,
+      "loss": 0.097,
+      "step": 1292
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.0050837993621826,
+      "learning_rate": 0.000270863309352518,
+      "loss": 0.5245,
+      "step": 1293
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.091763973236084,
+      "learning_rate": 0.00027082609774249563,
+      "loss": 1.1929,
+      "step": 1294
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.52555513381958,
+      "learning_rate": 0.00027078888613247333,
+      "loss": 1.1448,
+      "step": 1295
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.984481334686279,
+      "learning_rate": 0.000270751674522451,
+      "loss": 0.7998,
+      "step": 1296
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 6.075478553771973,
+      "learning_rate": 0.00027071446291242863,
+      "loss": 1.3081,
+      "step": 1297
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 16.068614959716797,
+      "learning_rate": 0.00027067725130240633,
+      "loss": 0.6254,
+      "step": 1298
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.1314620971679688,
+      "learning_rate": 0.000270640039692384,
+      "loss": 0.4411,
+      "step": 1299
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.841050624847412,
+      "learning_rate": 0.0002706028280823617,
+      "loss": 0.5328,
+      "step": 1300
+    },
+    {
+      "epoch": 0.46,
+      "eval_loss": 0.718771755695343,
+      "eval_runtime": 50.1715,
+      "eval_samples_per_second": 43.212,
+      "eval_steps_per_second": 10.803,
+      "eval_wer": 0.6280055353745027,
+      "step": 1300
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.5195000171661377,
+      "learning_rate": 0.00027056561647233933,
+      "loss": 1.2308,
+      "step": 1301
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.1528189182281494,
+      "learning_rate": 0.00027052840486231703,
+      "loss": 1.2859,
+      "step": 1302
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.554877519607544,
+      "learning_rate": 0.0002704911932522947,
+      "loss": 0.6787,
+      "step": 1303
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.0567691326141357,
+      "learning_rate": 0.00027045398164227233,
+      "loss": 0.8472,
+      "step": 1304
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9456758499145508,
+      "learning_rate": 0.00027041677003225003,
+      "loss": 0.5319,
+      "step": 1305
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.8807168006896973,
+      "learning_rate": 0.00027037955842222774,
+      "loss": 0.5714,
+      "step": 1306
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.7060537338256836,
+      "learning_rate": 0.0002703423468122054,
+      "loss": 0.8815,
+      "step": 1307
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.672762870788574,
+      "learning_rate": 0.0002703051352021831,
+      "loss": 0.9812,
+      "step": 1308
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.651334285736084,
+      "learning_rate": 0.00027026792359216074,
+      "loss": 0.58,
+      "step": 1309
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.646411418914795,
+      "learning_rate": 0.0002702307119821384,
+      "loss": 1.1246,
+      "step": 1310
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.932492971420288,
+      "learning_rate": 0.0002701935003721161,
+      "loss": 0.6828,
+      "step": 1311
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.369302749633789,
+      "learning_rate": 0.00027015628876209374,
+      "loss": 1.0535,
+      "step": 1312
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.5227344036102295,
+      "learning_rate": 0.00027011907715207144,
+      "loss": 0.7979,
+      "step": 1313
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.5197794437408447,
+      "learning_rate": 0.0002700818655420491,
+      "loss": 0.818,
+      "step": 1314
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.7741096019744873,
+      "learning_rate": 0.0002700446539320268,
+      "loss": 0.2863,
+      "step": 1315
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.058023929595947,
+      "learning_rate": 0.00027000744232200444,
+      "loss": 0.9948,
+      "step": 1316
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.746575832366943,
+      "learning_rate": 0.0002699702307119821,
+      "loss": 1.0099,
+      "step": 1317
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.513601064682007,
+      "learning_rate": 0.0002699330191019598,
+      "loss": 0.5189,
+      "step": 1318
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.361132860183716,
+      "learning_rate": 0.0002698958074919375,
+      "loss": 0.7626,
+      "step": 1319
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.748187780380249,
+      "learning_rate": 0.00026985859588191514,
+      "loss": 0.4882,
+      "step": 1320
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.661416530609131,
+      "learning_rate": 0.0002698213842718928,
+      "loss": 0.5571,
+      "step": 1321
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 6.057156562805176,
+      "learning_rate": 0.00026978417266187044,
+      "loss": 0.6306,
+      "step": 1322
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.695087432861328,
+      "learning_rate": 0.00026974696105184814,
+      "loss": 0.617,
+      "step": 1323
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.9815876483917236,
+      "learning_rate": 0.00026970974944182585,
+      "loss": 0.4741,
+      "step": 1324
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.866700649261475,
+      "learning_rate": 0.0002696725378318035,
+      "loss": 0.7968,
+      "step": 1325
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.9536242485046387,
+      "learning_rate": 0.0002696353262217812,
+      "loss": 1.1302,
+      "step": 1326
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.9558680057525635,
+      "learning_rate": 0.00026959811461175885,
+      "loss": 0.7709,
+      "step": 1327
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1161468029022217,
+      "learning_rate": 0.0002695609030017365,
+      "loss": 0.8176,
+      "step": 1328
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.3528337478637695,
+      "learning_rate": 0.0002695236913917142,
+      "loss": 1.0837,
+      "step": 1329
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.7077548503875732,
+      "learning_rate": 0.00026948647978169185,
+      "loss": 1.0823,
+      "step": 1330
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.875311851501465,
+      "learning_rate": 0.00026944926817166955,
+      "loss": 0.7376,
+      "step": 1331
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.183072090148926,
+      "learning_rate": 0.0002694120565616472,
+      "loss": 0.9196,
+      "step": 1332
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.7492129802703857,
+      "learning_rate": 0.0002693748449516249,
+      "loss": 0.4203,
+      "step": 1333
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.388514518737793,
+      "learning_rate": 0.00026933763334160255,
+      "loss": 1.483,
+      "step": 1334
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9908661842346191,
+      "learning_rate": 0.0002693004217315802,
+      "loss": 0.3537,
+      "step": 1335
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.7505980730056763,
+      "learning_rate": 0.0002692632101215579,
+      "loss": 0.3952,
+      "step": 1336
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 8.66080379486084,
+      "learning_rate": 0.0002692259985115356,
+      "loss": 2.9207,
+      "step": 1337
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.433154821395874,
+      "learning_rate": 0.00026918878690151325,
+      "loss": 0.8745,
+      "step": 1338
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.7627408504486084,
+      "learning_rate": 0.0002691515752914909,
+      "loss": 0.3728,
+      "step": 1339
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.7169649600982666,
+      "learning_rate": 0.0002691143636814686,
+      "loss": 0.4879,
+      "step": 1340
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.4643096923828125,
+      "learning_rate": 0.00026907715207144625,
+      "loss": 0.4795,
+      "step": 1341
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.340239524841309,
+      "learning_rate": 0.00026903994046142396,
+      "loss": 0.8171,
+      "step": 1342
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.2308506965637207,
+      "learning_rate": 0.0002690027288514016,
+      "loss": 0.7844,
+      "step": 1343
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.7775092124938965,
+      "learning_rate": 0.0002689655172413793,
+      "loss": 0.7076,
+      "step": 1344
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.886669635772705,
+      "learning_rate": 0.00026892830563135696,
+      "loss": 0.5041,
+      "step": 1345
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.971267223358154,
+      "learning_rate": 0.0002688910940213346,
+      "loss": 0.8198,
+      "step": 1346
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.96894907951355,
+      "learning_rate": 0.0002688538824113123,
+      "loss": 0.4911,
+      "step": 1347
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.252577781677246,
+      "learning_rate": 0.00026881667080128996,
+      "loss": 0.7421,
+      "step": 1348
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 5.704499244689941,
+      "learning_rate": 0.00026877945919126766,
+      "loss": 0.8419,
+      "step": 1349
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 6.062288761138916,
+      "learning_rate": 0.00026874224758124536,
+      "loss": 2.6257,
+      "step": 1350
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.9042396545410156,
+      "learning_rate": 0.000268705035971223,
+      "loss": 1.0895,
+      "step": 1351
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.361873149871826,
+      "learning_rate": 0.00026866782436120066,
+      "loss": 1.0075,
+      "step": 1352
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.925755262374878,
+      "learning_rate": 0.0002686306127511783,
+      "loss": 0.5891,
+      "step": 1353
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8617517948150635,
+      "learning_rate": 0.000268593401141156,
+      "loss": 0.6051,
+      "step": 1354
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.268332004547119,
+      "learning_rate": 0.0002685561895311337,
+      "loss": 1.3279,
+      "step": 1355
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.6798906326293945,
+      "learning_rate": 0.00026851897792111136,
+      "loss": 0.7166,
+      "step": 1356
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.4666085243225098,
+      "learning_rate": 0.00026848176631108907,
+      "loss": 0.8236,
+      "step": 1357
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.445189952850342,
+      "learning_rate": 0.0002684445547010667,
+      "loss": 1.0953,
+      "step": 1358
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.491154909133911,
+      "learning_rate": 0.00026840734309104436,
+      "loss": 0.86,
+      "step": 1359
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.873551368713379,
+      "learning_rate": 0.00026837013148102207,
+      "loss": 0.3485,
+      "step": 1360
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.520420551300049,
+      "learning_rate": 0.0002683329198709997,
+      "loss": 0.5189,
+      "step": 1361
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.5249836444854736,
+      "learning_rate": 0.0002682957082609774,
+      "loss": 0.6036,
+      "step": 1362
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 6.196778774261475,
+      "learning_rate": 0.00026825849665095507,
+      "loss": 1.0194,
+      "step": 1363
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.7715699672698975,
+      "learning_rate": 0.0002682212850409327,
+      "loss": 0.6127,
+      "step": 1364
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.2750773429870605,
+      "learning_rate": 0.0002681840734309104,
+      "loss": 0.5814,
+      "step": 1365
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 6.068713188171387,
+      "learning_rate": 0.00026814686182088807,
+      "loss": 0.8995,
+      "step": 1366
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.5809454917907715,
+      "learning_rate": 0.00026810965021086577,
+      "loss": 0.4658,
+      "step": 1367
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 8.29049301147461,
+      "learning_rate": 0.0002680724386008435,
+      "loss": 0.942,
+      "step": 1368
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.825986385345459,
+      "learning_rate": 0.0002680352269908211,
+      "loss": 0.2832,
+      "step": 1369
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.205315113067627,
+      "learning_rate": 0.00026799801538079877,
+      "loss": 0.3227,
+      "step": 1370
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.076242685317993,
+      "learning_rate": 0.0002679608037707764,
+      "loss": 0.6167,
+      "step": 1371
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.471637487411499,
+      "learning_rate": 0.0002679235921607541,
+      "loss": 0.4435,
+      "step": 1372
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 6.81670618057251,
+      "learning_rate": 0.0002678863805507318,
+      "loss": 0.6166,
+      "step": 1373
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 10.90646743774414,
+      "learning_rate": 0.0002678491689407095,
+      "loss": 1.4561,
+      "step": 1374
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 10.379186630249023,
+      "learning_rate": 0.0002678119573306872,
+      "loss": 1.1457,
+      "step": 1375
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 10.573211669921875,
+      "learning_rate": 0.0002677747457206648,
+      "loss": 1.8288,
+      "step": 1376
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.0730042457580566,
+      "learning_rate": 0.0002677375341106425,
+      "loss": 1.1138,
+      "step": 1377
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.4380507469177246,
+      "learning_rate": 0.0002677003225006202,
+      "loss": 0.6603,
+      "step": 1378
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.709672212600708,
+      "learning_rate": 0.0002676631108905978,
+      "loss": 0.7853,
+      "step": 1379
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.5759897232055664,
+      "learning_rate": 0.00026762589928057553,
+      "loss": 0.7393,
+      "step": 1380
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.016786813735962,
+      "learning_rate": 0.0002675886876705532,
+      "loss": 0.9876,
+      "step": 1381
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.079416036605835,
+      "learning_rate": 0.0002675514760605309,
+      "loss": 0.4571,
+      "step": 1382
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.269665241241455,
+      "learning_rate": 0.00026751426445050853,
+      "loss": 0.9098,
+      "step": 1383
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.512655735015869,
+      "learning_rate": 0.0002674770528404862,
+      "loss": 1.1042,
+      "step": 1384
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.725264072418213,
+      "learning_rate": 0.0002674398412304639,
+      "loss": 0.8414,
+      "step": 1385
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.0862860679626465,
+      "learning_rate": 0.0002674026296204416,
+      "loss": 1.6329,
+      "step": 1386
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.442559242248535,
+      "learning_rate": 0.00026736541801041923,
+      "loss": 1.1877,
+      "step": 1387
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.18272066116333,
+      "learning_rate": 0.0002673282064003969,
+      "loss": 0.8395,
+      "step": 1388
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.8753631114959717,
+      "learning_rate": 0.0002672909947903746,
+      "loss": 0.5428,
+      "step": 1389
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.923710823059082,
+      "learning_rate": 0.00026725378318035223,
+      "loss": 1.0107,
+      "step": 1390
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.947282552719116,
+      "learning_rate": 0.00026721657157032994,
+      "loss": 0.4316,
+      "step": 1391
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.441121816635132,
+      "learning_rate": 0.0002671793599603076,
+      "loss": 1.4819,
+      "step": 1392
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.639099597930908,
+      "learning_rate": 0.0002671421483502853,
+      "loss": 0.4009,
+      "step": 1393
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.281423091888428,
+      "learning_rate": 0.00026710493674026294,
+      "loss": 0.8285,
+      "step": 1394
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.369678020477295,
+      "learning_rate": 0.0002670677251302406,
+      "loss": 1.3822,
+      "step": 1395
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.5579185485839844,
+      "learning_rate": 0.0002670305135202183,
+      "loss": 0.4678,
+      "step": 1396
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.262704849243164,
+      "learning_rate": 0.00026699330191019594,
+      "loss": 0.5735,
+      "step": 1397
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.989433526992798,
+      "learning_rate": 0.00026695609030017364,
+      "loss": 0.6595,
+      "step": 1398
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 7.13302755355835,
+      "learning_rate": 0.00026691887869015134,
+      "loss": 1.1522,
+      "step": 1399
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 10.057518005371094,
+      "learning_rate": 0.000266881667080129,
+      "loss": 1.3733,
+      "step": 1400
+    },
+    {
+      "epoch": 0.49,
+      "eval_loss": 0.6848714351654053,
+      "eval_runtime": 50.8029,
+      "eval_samples_per_second": 42.675,
+      "eval_steps_per_second": 10.669,
+      "eval_wer": 0.6019719771665801,
+      "step": 1400
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.617882251739502,
+      "learning_rate": 0.00026684445547010664,
+      "loss": 1.3938,
+      "step": 1401
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1804263591766357,
+      "learning_rate": 0.0002668072438600843,
+      "loss": 0.9542,
+      "step": 1402
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.778125286102295,
+      "learning_rate": 0.000266770032250062,
+      "loss": 0.6878,
+      "step": 1403
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.387033700942993,
+      "learning_rate": 0.0002667328206400397,
+      "loss": 0.6723,
+      "step": 1404
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.006556272506714,
+      "learning_rate": 0.00026669560903001734,
+      "loss": 0.8387,
+      "step": 1405
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9671204090118408,
+      "learning_rate": 0.000266658397419995,
+      "loss": 0.6155,
+      "step": 1406
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.056529998779297,
+      "learning_rate": 0.0002666211858099727,
+      "loss": 0.5556,
+      "step": 1407
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.5942492485046387,
+      "learning_rate": 0.00026658397419995034,
+      "loss": 0.3429,
+      "step": 1408
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 6.031838417053223,
+      "learning_rate": 0.00026654676258992805,
+      "loss": 1.1818,
+      "step": 1409
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.842709541320801,
+      "learning_rate": 0.0002665095509799057,
+      "loss": 0.9802,
+      "step": 1410
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.389888286590576,
+      "learning_rate": 0.0002664723393698834,
+      "loss": 0.4917,
+      "step": 1411
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.805605173110962,
+      "learning_rate": 0.00026643512775986105,
+      "loss": 0.6546,
+      "step": 1412
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.952422142028809,
+      "learning_rate": 0.0002663979161498387,
+      "loss": 0.6714,
+      "step": 1413
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.914050340652466,
+      "learning_rate": 0.0002663607045398164,
+      "loss": 0.6369,
+      "step": 1414
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.893771171569824,
+      "learning_rate": 0.00026632349292979405,
+      "loss": 0.721,
+      "step": 1415
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.831104755401611,
+      "learning_rate": 0.00026628628131977175,
+      "loss": 1.0663,
+      "step": 1416
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.74212384223938,
+      "learning_rate": 0.00026624906970974945,
+      "loss": 0.6219,
+      "step": 1417
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.251964569091797,
+      "learning_rate": 0.0002662118580997271,
+      "loss": 1.5889,
+      "step": 1418
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.620336532592773,
+      "learning_rate": 0.00026617464648970475,
+      "loss": 1.05,
+      "step": 1419
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.674115180969238,
+      "learning_rate": 0.00026613743487968245,
+      "loss": 0.7279,
+      "step": 1420
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.6265783309936523,
+      "learning_rate": 0.0002661002232696601,
+      "loss": 0.4997,
+      "step": 1421
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.726874828338623,
+      "learning_rate": 0.0002660630116596378,
+      "loss": 0.9038,
+      "step": 1422
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.306706428527832,
+      "learning_rate": 0.00026602580004961545,
+      "loss": 1.3036,
+      "step": 1423
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.389394283294678,
+      "learning_rate": 0.00026598858843959316,
+      "loss": 1.1654,
+      "step": 1424
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.7293262481689453,
+      "learning_rate": 0.0002659513768295708,
+      "loss": 0.4846,
+      "step": 1425
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.0907764434814453,
+      "learning_rate": 0.00026591416521954845,
+      "loss": 1.5577,
+      "step": 1426
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.340789318084717,
+      "learning_rate": 0.00026587695360952616,
+      "loss": 0.8854,
+      "step": 1427
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.936422348022461,
+      "learning_rate": 0.0002658397419995038,
+      "loss": 1.3372,
+      "step": 1428
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.7207283973693848,
+      "learning_rate": 0.0002658025303894815,
+      "loss": 0.9362,
+      "step": 1429
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.6525309085845947,
+      "learning_rate": 0.00026576531877945916,
+      "loss": 0.7226,
+      "step": 1430
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.381458282470703,
+      "learning_rate": 0.00026572810716943686,
+      "loss": 0.6479,
+      "step": 1431
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.5098137855529785,
+      "learning_rate": 0.0002656908955594145,
+      "loss": 0.6157,
+      "step": 1432
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.9847960472106934,
+      "learning_rate": 0.00026565368394939216,
+      "loss": 0.5089,
+      "step": 1433
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.012515544891357,
+      "learning_rate": 0.00026561647233936986,
+      "loss": 0.4906,
+      "step": 1434
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.7779381275177,
+      "learning_rate": 0.00026557926072934756,
+      "loss": 0.5565,
+      "step": 1435
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.0392794609069824,
+      "learning_rate": 0.0002655420491193252,
+      "loss": 0.631,
+      "step": 1436
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.07166051864624,
+      "learning_rate": 0.00026550483750930286,
+      "loss": 0.4397,
+      "step": 1437
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.711935520172119,
+      "learning_rate": 0.00026546762589928056,
+      "loss": 0.5008,
+      "step": 1438
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.4669036865234375,
+      "learning_rate": 0.0002654304142892582,
+      "loss": 0.5278,
+      "step": 1439
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.123307228088379,
+      "learning_rate": 0.0002653932026792359,
+      "loss": 0.7766,
+      "step": 1440
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.164463758468628,
+      "learning_rate": 0.00026535599106921356,
+      "loss": 0.6879,
+      "step": 1441
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.237036228179932,
+      "learning_rate": 0.00026531877945919127,
+      "loss": 0.6142,
+      "step": 1442
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.148812770843506,
+      "learning_rate": 0.0002652815678491689,
+      "loss": 0.469,
+      "step": 1443
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.390077590942383,
+      "learning_rate": 0.00026524435623914656,
+      "loss": 0.7129,
+      "step": 1444
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.447877883911133,
+      "learning_rate": 0.00026520714462912427,
+      "loss": 1.0652,
+      "step": 1445
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.019307613372803,
+      "learning_rate": 0.0002651699330191019,
+      "loss": 0.6862,
+      "step": 1446
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.470080375671387,
+      "learning_rate": 0.0002651327214090796,
+      "loss": 0.9423,
+      "step": 1447
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.0855607986450195,
+      "learning_rate": 0.00026509550979905727,
+      "loss": 0.7612,
+      "step": 1448
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.5919671058654785,
+      "learning_rate": 0.00026505829818903497,
+      "loss": 0.7675,
+      "step": 1449
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.587916612625122,
+      "learning_rate": 0.0002650210865790126,
+      "loss": 0.2524,
+      "step": 1450
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 6.112603664398193,
+      "learning_rate": 0.0002649838749689903,
+      "loss": 1.4048,
+      "step": 1451
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.97434139251709,
+      "learning_rate": 0.00026494666335896797,
+      "loss": 1.2854,
+      "step": 1452
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.173609972000122,
+      "learning_rate": 0.00026490945174894567,
+      "loss": 0.5143,
+      "step": 1453
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.1337573528289795,
+      "learning_rate": 0.0002648722401389233,
+      "loss": 0.7012,
+      "step": 1454
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 10.74719524383545,
+      "learning_rate": 0.00026483502852890097,
+      "loss": 2.8888,
+      "step": 1455
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.577202558517456,
+      "learning_rate": 0.0002647978169188787,
+      "loss": 0.5115,
+      "step": 1456
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.5629165172576904,
+      "learning_rate": 0.0002647606053088563,
+      "loss": 0.9478,
+      "step": 1457
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.9556891918182373,
+      "learning_rate": 0.000264723393698834,
+      "loss": 0.5669,
+      "step": 1458
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.905815601348877,
+      "learning_rate": 0.0002646861820888117,
+      "loss": 0.732,
+      "step": 1459
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.5102758407592773,
+      "learning_rate": 0.0002646489704787894,
+      "loss": 0.875,
+      "step": 1460
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.5579817295074463,
+      "learning_rate": 0.000264611758868767,
+      "loss": 0.5173,
+      "step": 1461
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.0412824153900146,
+      "learning_rate": 0.0002645745472587447,
+      "loss": 0.95,
+      "step": 1462
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.4420363903045654,
+      "learning_rate": 0.0002645373356487224,
+      "loss": 0.5622,
+      "step": 1463
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.9669206142425537,
+      "learning_rate": 0.0002645001240387001,
+      "loss": 1.3098,
+      "step": 1464
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.515052080154419,
+      "learning_rate": 0.00026446291242867773,
+      "loss": 0.6694,
+      "step": 1465
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.764845848083496,
+      "learning_rate": 0.00026442570081865543,
+      "loss": 0.4299,
+      "step": 1466
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.278150796890259,
+      "learning_rate": 0.0002643884892086331,
+      "loss": 0.5183,
+      "step": 1467
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.361281394958496,
+      "learning_rate": 0.00026435127759861073,
+      "loss": 0.8904,
+      "step": 1468
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.450416088104248,
+      "learning_rate": 0.00026431406598858843,
+      "loss": 0.5072,
+      "step": 1469
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.9510233402252197,
+      "learning_rate": 0.0002642768543785661,
+      "loss": 0.5038,
+      "step": 1470
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.09872579574585,
+      "learning_rate": 0.0002642396427685438,
+      "loss": 0.8801,
+      "step": 1471
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.2074294090270996,
+      "learning_rate": 0.00026420243115852143,
+      "loss": 0.5852,
+      "step": 1472
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 6.999640941619873,
+      "learning_rate": 0.00026416521954849913,
+      "loss": 0.4915,
+      "step": 1473
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.909586191177368,
+      "learning_rate": 0.0002641280079384768,
+      "loss": 0.811,
+      "step": 1474
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 8.129749298095703,
+      "learning_rate": 0.00026409079632845443,
+      "loss": 2.2248,
+      "step": 1475
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.023890018463135,
+      "learning_rate": 0.00026405358471843213,
+      "loss": 1.2998,
+      "step": 1476
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.378636598587036,
+      "learning_rate": 0.0002640163731084098,
+      "loss": 0.7717,
+      "step": 1477
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.6731226444244385,
+      "learning_rate": 0.0002639791614983875,
+      "loss": 0.7744,
+      "step": 1478
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.031921863555908,
+      "learning_rate": 0.00026394194988836513,
+      "loss": 0.9579,
+      "step": 1479
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.2059433460235596,
+      "learning_rate": 0.0002639047382783428,
+      "loss": 0.7075,
+      "step": 1480
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.67105770111084,
+      "learning_rate": 0.0002638675266683205,
+      "loss": 0.7341,
+      "step": 1481
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.016000986099243,
+      "learning_rate": 0.0002638303150582982,
+      "loss": 0.4283,
+      "step": 1482
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.9530158042907715,
+      "learning_rate": 0.00026379310344827584,
+      "loss": 0.6266,
+      "step": 1483
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.5564568042755127,
+      "learning_rate": 0.00026375589183825354,
+      "loss": 0.7194,
+      "step": 1484
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.650444269180298,
+      "learning_rate": 0.0002637186802282312,
+      "loss": 0.5366,
+      "step": 1485
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.470327854156494,
+      "learning_rate": 0.00026368146861820884,
+      "loss": 0.6349,
+      "step": 1486
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.796874761581421,
+      "learning_rate": 0.00026364425700818654,
+      "loss": 0.6011,
+      "step": 1487
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.449688673019409,
+      "learning_rate": 0.0002636070453981642,
+      "loss": 0.6161,
+      "step": 1488
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.23581600189209,
+      "learning_rate": 0.0002635698337881419,
+      "loss": 0.4338,
+      "step": 1489
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.239893436431885,
+      "learning_rate": 0.00026353262217811954,
+      "loss": 0.9053,
+      "step": 1490
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.8986151218414307,
+      "learning_rate": 0.00026349541056809724,
+      "loss": 0.3529,
+      "step": 1491
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.6103169918060303,
+      "learning_rate": 0.0002634581989580749,
+      "loss": 0.8344,
+      "step": 1492
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.4422993659973145,
+      "learning_rate": 0.00026342098734805254,
+      "loss": 0.6958,
+      "step": 1493
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.637202501296997,
+      "learning_rate": 0.00026338377573803024,
+      "loss": 0.9728,
+      "step": 1494
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.514386177062988,
+      "learning_rate": 0.00026334656412800795,
+      "loss": 0.625,
+      "step": 1495
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.6551239490509033,
+      "learning_rate": 0.0002633093525179856,
+      "loss": 0.7805,
+      "step": 1496
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.161332130432129,
+      "learning_rate": 0.00026327214090796324,
+      "loss": 0.7602,
+      "step": 1497
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.9714465141296387,
+      "learning_rate": 0.00026323492929794095,
+      "loss": 0.5906,
+      "step": 1498
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.682180404663086,
+      "learning_rate": 0.0002631977176879186,
+      "loss": 0.4266,
+      "step": 1499
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.590712785720825,
+      "learning_rate": 0.0002631605060778963,
+      "loss": 0.4957,
+      "step": 1500
+    },
+    {
+      "epoch": 0.53,
+      "eval_loss": 0.735007107257843,
+      "eval_runtime": 50.1513,
+      "eval_samples_per_second": 43.229,
+      "eval_steps_per_second": 10.807,
+      "eval_wer": 0.630254281266217,
+      "step": 1500
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.7293701171875,
+      "learning_rate": 0.00026312329446787395,
+      "loss": 0.8107,
+      "step": 1501
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.134906053543091,
+      "learning_rate": 0.00026308608285785165,
+      "loss": 0.6081,
+      "step": 1502
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.5551223754882812,
+      "learning_rate": 0.0002630488712478293,
+      "loss": 0.7375,
+      "step": 1503
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.314273357391357,
+      "learning_rate": 0.00026301165963780695,
+      "loss": 0.8512,
+      "step": 1504
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.985286235809326,
+      "learning_rate": 0.00026297444802778465,
+      "loss": 0.5388,
+      "step": 1505
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.9923386573791504,
+      "learning_rate": 0.0002629372364177623,
+      "loss": 0.9692,
+      "step": 1506
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.3773350715637207,
+      "learning_rate": 0.00026290002480774,
+      "loss": 0.9575,
+      "step": 1507
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.891386032104492,
+      "learning_rate": 0.0002628628131977177,
+      "loss": 0.8126,
+      "step": 1508
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.9968199729919434,
+      "learning_rate": 0.00026282560158769535,
+      "loss": 0.5441,
+      "step": 1509
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.9973716735839844,
+      "learning_rate": 0.000262788389977673,
+      "loss": 0.618,
+      "step": 1510
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.8595619201660156,
+      "learning_rate": 0.00026275117836765065,
+      "loss": 0.904,
+      "step": 1511
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.169881820678711,
+      "learning_rate": 0.00026271396675762835,
+      "loss": 0.5045,
+      "step": 1512
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.6832542419433594,
+      "learning_rate": 0.00026267675514760606,
+      "loss": 0.3136,
+      "step": 1513
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.8257126808166504,
+      "learning_rate": 0.0002626395435375837,
+      "loss": 0.4304,
+      "step": 1514
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0919361114501953,
+      "learning_rate": 0.0002626023319275614,
+      "loss": 0.3809,
+      "step": 1515
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.3260436058044434,
+      "learning_rate": 0.00026256512031753906,
+      "loss": 0.5717,
+      "step": 1516
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.5611228942871094,
+      "learning_rate": 0.0002625279087075167,
+      "loss": 0.5815,
+      "step": 1517
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.097376346588135,
+      "learning_rate": 0.0002624906970974944,
+      "loss": 0.7399,
+      "step": 1518
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.300809621810913,
+      "learning_rate": 0.00026245348548747206,
+      "loss": 0.7625,
+      "step": 1519
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 6.034483432769775,
+      "learning_rate": 0.00026241627387744976,
+      "loss": 1.3396,
+      "step": 1520
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.987584114074707,
+      "learning_rate": 0.0002623790622674274,
+      "loss": 0.8065,
+      "step": 1521
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.958810567855835,
+      "learning_rate": 0.00026234185065740506,
+      "loss": 0.5565,
+      "step": 1522
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.113748550415039,
+      "learning_rate": 0.00026230463904738276,
+      "loss": 0.4262,
+      "step": 1523
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.8435819149017334,
+      "learning_rate": 0.0002622674274373604,
+      "loss": 0.6072,
+      "step": 1524
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.7314817905426025,
+      "learning_rate": 0.0002622302158273381,
+      "loss": 0.6308,
+      "step": 1525
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.7327356338500977,
+      "learning_rate": 0.0002621930042173158,
+      "loss": 0.904,
+      "step": 1526
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.118708848953247,
+      "learning_rate": 0.00026215579260729346,
+      "loss": 0.784,
+      "step": 1527
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.2013180255889893,
+      "learning_rate": 0.0002621185809972711,
+      "loss": 0.8928,
+      "step": 1528
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.631147861480713,
+      "learning_rate": 0.00026208136938724876,
+      "loss": 0.6634,
+      "step": 1529
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8624056577682495,
+      "learning_rate": 0.00026204415777722646,
+      "loss": 0.3717,
+      "step": 1530
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.32458233833313,
+      "learning_rate": 0.00026200694616720417,
+      "loss": 0.4845,
+      "step": 1531
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.4256484508514404,
+      "learning_rate": 0.0002619697345571818,
+      "loss": 0.6534,
+      "step": 1532
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.6999833583831787,
+      "learning_rate": 0.0002619325229471595,
+      "loss": 0.6113,
+      "step": 1533
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.6705386638641357,
+      "learning_rate": 0.00026189531133713717,
+      "loss": 0.6816,
+      "step": 1534
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.588336706161499,
+      "learning_rate": 0.0002618580997271148,
+      "loss": 0.5221,
+      "step": 1535
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.662912368774414,
+      "learning_rate": 0.0002618208881170925,
+      "loss": 1.0771,
+      "step": 1536
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.598870277404785,
+      "learning_rate": 0.00026178367650707017,
+      "loss": 0.3864,
+      "step": 1537
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.303178310394287,
+      "learning_rate": 0.00026174646489704787,
+      "loss": 0.8527,
+      "step": 1538
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.631732702255249,
+      "learning_rate": 0.0002617092532870255,
+      "loss": 0.2559,
+      "step": 1539
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.9511115550994873,
+      "learning_rate": 0.0002616720416770032,
+      "loss": 1.2759,
+      "step": 1540
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.213435173034668,
+      "learning_rate": 0.00026163483006698087,
+      "loss": 0.9761,
+      "step": 1541
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.040883541107178,
+      "learning_rate": 0.0002615976184569585,
+      "loss": 1.2503,
+      "step": 1542
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.2061383724212646,
+      "learning_rate": 0.0002615604068469362,
+      "loss": 0.4728,
+      "step": 1543
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.8172144889831543,
+      "learning_rate": 0.0002615231952369139,
+      "loss": 0.8298,
+      "step": 1544
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.939277648925781,
+      "learning_rate": 0.0002614859836268916,
+      "loss": 1.1306,
+      "step": 1545
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.502394676208496,
+      "learning_rate": 0.0002614487720168692,
+      "loss": 0.4795,
+      "step": 1546
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.375518321990967,
+      "learning_rate": 0.00026141156040684687,
+      "loss": 0.3351,
+      "step": 1547
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 6.979621410369873,
+      "learning_rate": 0.0002613743487968246,
+      "loss": 1.9284,
+      "step": 1548
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.901841878890991,
+      "learning_rate": 0.0002613371371868023,
+      "loss": 0.4371,
+      "step": 1549
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.348433971405029,
+      "learning_rate": 0.0002612999255767799,
+      "loss": 0.6329,
+      "step": 1550
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.033573865890503,
+      "learning_rate": 0.00026126271396675763,
+      "loss": 1.1795,
+      "step": 1551
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.457150459289551,
+      "learning_rate": 0.0002612255023567353,
+      "loss": 0.5499,
+      "step": 1552
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8844290971755981,
+      "learning_rate": 0.0002611882907467129,
+      "loss": 0.7523,
+      "step": 1553
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.3162200450897217,
+      "learning_rate": 0.00026115107913669063,
+      "loss": 0.884,
+      "step": 1554
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.550788164138794,
+      "learning_rate": 0.0002611138675266683,
+      "loss": 0.7523,
+      "step": 1555
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.001537561416626,
+      "learning_rate": 0.000261076655916646,
+      "loss": 0.5522,
+      "step": 1556
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1897246837615967,
+      "learning_rate": 0.0002610394443066237,
+      "loss": 0.4203,
+      "step": 1557
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.3803975582122803,
+      "learning_rate": 0.00026100223269660133,
+      "loss": 0.9045,
+      "step": 1558
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.0036349296569824,
+      "learning_rate": 0.000260965021086579,
+      "loss": 0.6177,
+      "step": 1559
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.269506454467773,
+      "learning_rate": 0.00026092780947655663,
+      "loss": 1.6503,
+      "step": 1560
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.8251953125,
+      "learning_rate": 0.00026089059786653433,
+      "loss": 0.6453,
+      "step": 1561
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.577118396759033,
+      "learning_rate": 0.00026085338625651204,
+      "loss": 1.0677,
+      "step": 1562
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.7220301628112793,
+      "learning_rate": 0.0002608161746464897,
+      "loss": 1.0965,
+      "step": 1563
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.421391010284424,
+      "learning_rate": 0.00026077896303646733,
+      "loss": 0.4751,
+      "step": 1564
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.9482858180999756,
+      "learning_rate": 0.00026074175142644504,
+      "loss": 0.5636,
+      "step": 1565
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.108609199523926,
+      "learning_rate": 0.0002607045398164227,
+      "loss": 0.6971,
+      "step": 1566
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.993597984313965,
+      "learning_rate": 0.0002606673282064004,
+      "loss": 0.6064,
+      "step": 1567
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.803035259246826,
+      "learning_rate": 0.00026063011659637804,
+      "loss": 0.5257,
+      "step": 1568
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.5908119678497314,
+      "learning_rate": 0.00026059290498635574,
+      "loss": 0.8194,
+      "step": 1569
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1012234687805176,
+      "learning_rate": 0.0002605556933763334,
+      "loss": 0.5269,
+      "step": 1570
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.691065549850464,
+      "learning_rate": 0.00026051848176631104,
+      "loss": 0.638,
+      "step": 1571
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.872591018676758,
+      "learning_rate": 0.00026048127015628874,
+      "loss": 0.8042,
+      "step": 1572
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.684198379516602,
+      "learning_rate": 0.0002604440585462664,
+      "loss": 0.8874,
+      "step": 1573
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.133639812469482,
+      "learning_rate": 0.0002604068469362441,
+      "loss": 0.933,
+      "step": 1574
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.2042012214660645,
+      "learning_rate": 0.0002603696353262218,
+      "loss": 1.4783,
+      "step": 1575
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.421505928039551,
+      "learning_rate": 0.00026033242371619944,
+      "loss": 1.4112,
+      "step": 1576
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.258162975311279,
+      "learning_rate": 0.0002602952121061771,
+      "loss": 1.1223,
+      "step": 1577
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.783501386642456,
+      "learning_rate": 0.00026025800049615474,
+      "loss": 0.6567,
+      "step": 1578
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.9202888011932373,
+      "learning_rate": 0.00026022078888613244,
+      "loss": 1.1697,
+      "step": 1579
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.2230772972106934,
+      "learning_rate": 0.00026018357727611015,
+      "loss": 0.6542,
+      "step": 1580
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.5005552768707275,
+      "learning_rate": 0.0002601463656660878,
+      "loss": 0.484,
+      "step": 1581
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.2738864421844482,
+      "learning_rate": 0.0002601091540560655,
+      "loss": 0.5697,
+      "step": 1582
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.813744306564331,
+      "learning_rate": 0.00026007194244604315,
+      "loss": 0.7979,
+      "step": 1583
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.8571112155914307,
+      "learning_rate": 0.0002600347308360208,
+      "loss": 0.6285,
+      "step": 1584
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.8317923545837402,
+      "learning_rate": 0.0002599975192259985,
+      "loss": 0.4895,
+      "step": 1585
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.4439423084259033,
+      "learning_rate": 0.00025996030761597615,
+      "loss": 0.4935,
+      "step": 1586
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.2733328342437744,
+      "learning_rate": 0.00025992309600595385,
+      "loss": 0.8773,
+      "step": 1587
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.249488592147827,
+      "learning_rate": 0.0002598858843959315,
+      "loss": 0.4,
+      "step": 1588
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9486145973205566,
+      "learning_rate": 0.00025984867278590915,
+      "loss": 0.48,
+      "step": 1589
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.063397407531738,
+      "learning_rate": 0.00025981146117588685,
+      "loss": 0.7096,
+      "step": 1590
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.8297181129455566,
+      "learning_rate": 0.0002597742495658645,
+      "loss": 0.5382,
+      "step": 1591
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.13065505027771,
+      "learning_rate": 0.0002597370379558422,
+      "loss": 0.6512,
+      "step": 1592
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.807405710220337,
+      "learning_rate": 0.0002596998263458199,
+      "loss": 0.4922,
+      "step": 1593
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.6079254150390625,
+      "learning_rate": 0.00025966261473579755,
+      "loss": 0.5638,
+      "step": 1594
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.535735607147217,
+      "learning_rate": 0.0002596254031257752,
+      "loss": 0.7087,
+      "step": 1595
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.7222447395324707,
+      "learning_rate": 0.0002595881915157529,
+      "loss": 0.4831,
+      "step": 1596
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.0065388679504395,
+      "learning_rate": 0.00025955097990573055,
+      "loss": 0.7503,
+      "step": 1597
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.3113796710968018,
+      "learning_rate": 0.00025951376829570826,
+      "loss": 0.5052,
+      "step": 1598
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.114770412445068,
+      "learning_rate": 0.0002594765566856859,
+      "loss": 0.6825,
+      "step": 1599
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.18342924118042,
+      "learning_rate": 0.0002594393450756636,
+      "loss": 0.4084,
+      "step": 1600
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 0.7260645627975464,
+      "eval_runtime": 49.8256,
+      "eval_samples_per_second": 43.512,
+      "eval_steps_per_second": 10.878,
+      "eval_wer": 0.6434872859366891,
+      "step": 1600
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.540189504623413,
+      "learning_rate": 0.00025940213346564126,
+      "loss": 0.9917,
+      "step": 1601
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 5.168649673461914,
+      "learning_rate": 0.0002593649218556189,
+      "loss": 1.1896,
+      "step": 1602
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.469858407974243,
+      "learning_rate": 0.0002593277102455966,
+      "loss": 0.588,
+      "step": 1603
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.415588617324829,
+      "learning_rate": 0.00025929049863557426,
+      "loss": 1.2277,
+      "step": 1604
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8117973804473877,
+      "learning_rate": 0.00025925328702555196,
+      "loss": 0.5193,
+      "step": 1605
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.193356990814209,
+      "learning_rate": 0.0002592160754155296,
+      "loss": 0.5361,
+      "step": 1606
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.278368949890137,
+      "learning_rate": 0.0002591788638055073,
+      "loss": 0.7885,
+      "step": 1607
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.5760748386383057,
+      "learning_rate": 0.00025914165219548496,
+      "loss": 0.6032,
+      "step": 1608
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.544609308242798,
+      "learning_rate": 0.00025910444058546266,
+      "loss": 0.7552,
+      "step": 1609
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.166447639465332,
+      "learning_rate": 0.0002590672289754403,
+      "loss": 0.9139,
+      "step": 1610
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 5.605889797210693,
+      "learning_rate": 0.000259030017365418,
+      "loss": 1.2701,
+      "step": 1611
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.2346339225769043,
+      "learning_rate": 0.00025899280575539566,
+      "loss": 0.807,
+      "step": 1612
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.9761223793029785,
+      "learning_rate": 0.0002589555941453733,
+      "loss": 0.6815,
+      "step": 1613
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0151445865631104,
+      "learning_rate": 0.000258918382535351,
+      "loss": 0.4635,
+      "step": 1614
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.8519450426101685,
+      "learning_rate": 0.00025888117092532866,
+      "loss": 0.3119,
+      "step": 1615
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.429759502410889,
+      "learning_rate": 0.00025884395931530637,
+      "loss": 0.7836,
+      "step": 1616
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 17.506669998168945,
+      "learning_rate": 0.000258806747705284,
+      "loss": 0.8412,
+      "step": 1617
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 5.09642219543457,
+      "learning_rate": 0.0002587695360952617,
+      "loss": 0.8578,
+      "step": 1618
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 7.458622932434082,
+      "learning_rate": 0.00025873232448523937,
+      "loss": 0.9713,
+      "step": 1619
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.237120628356934,
+      "learning_rate": 0.000258695112875217,
+      "loss": 1.0024,
+      "step": 1620
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.813836097717285,
+      "learning_rate": 0.0002586579012651947,
+      "loss": 0.8012,
+      "step": 1621
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 6.002732753753662,
+      "learning_rate": 0.00025862068965517237,
+      "loss": 1.0697,
+      "step": 1622
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 5.764606475830078,
+      "learning_rate": 0.00025858347804515007,
+      "loss": 0.8174,
+      "step": 1623
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.286224842071533,
+      "learning_rate": 0.00025854626643512777,
+      "loss": 0.3334,
+      "step": 1624
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.0024983882904053,
+      "learning_rate": 0.0002585090548251054,
+      "loss": 0.3869,
+      "step": 1625
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.4190189838409424,
+      "learning_rate": 0.00025847184321508307,
+      "loss": 0.7463,
+      "step": 1626
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.419149875640869,
+      "learning_rate": 0.0002584346316050608,
+      "loss": 0.8414,
+      "step": 1627
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.5799694061279297,
+      "learning_rate": 0.0002583974199950384,
+      "loss": 0.8311,
+      "step": 1628
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.2012771368026733,
+      "learning_rate": 0.0002583602083850161,
+      "loss": 0.2703,
+      "step": 1629
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.410515785217285,
+      "learning_rate": 0.0002583229967749938,
+      "loss": 0.7188,
+      "step": 1630
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3258681297302246,
+      "learning_rate": 0.0002582857851649714,
+      "loss": 0.4769,
+      "step": 1631
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.23000431060791,
+      "learning_rate": 0.0002582485735549491,
+      "loss": 0.3131,
+      "step": 1632
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.1836373805999756,
+      "learning_rate": 0.0002582113619449268,
+      "loss": 0.91,
+      "step": 1633
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.7543880939483643,
+      "learning_rate": 0.0002581741503349045,
+      "loss": 1.0753,
+      "step": 1634
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.235100030899048,
+      "learning_rate": 0.0002581369387248821,
+      "loss": 0.5141,
+      "step": 1635
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.6288204193115234,
+      "learning_rate": 0.00025809972711485983,
+      "loss": 0.8646,
+      "step": 1636
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.652747869491577,
+      "learning_rate": 0.0002580625155048375,
+      "loss": 1.2427,
+      "step": 1637
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 5.494747161865234,
+      "learning_rate": 0.0002580253038948151,
+      "loss": 0.784,
+      "step": 1638
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1231942176818848,
+      "learning_rate": 0.00025798809228479283,
+      "loss": 0.462,
+      "step": 1639
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.513335704803467,
+      "learning_rate": 0.00025795088067477053,
+      "loss": 0.7612,
+      "step": 1640
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.353682279586792,
+      "learning_rate": 0.0002579136690647482,
+      "loss": 0.6168,
+      "step": 1641
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.6810176372528076,
+      "learning_rate": 0.0002578764574547259,
+      "loss": 0.6426,
+      "step": 1642
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 6.6810126304626465,
+      "learning_rate": 0.00025783924584470353,
+      "loss": 2.6482,
+      "step": 1643
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.389218330383301,
+      "learning_rate": 0.0002578020342346812,
+      "loss": 0.3808,
+      "step": 1644
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.735605239868164,
+      "learning_rate": 0.0002577648226246589,
+      "loss": 0.5647,
+      "step": 1645
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 40.25916290283203,
+      "learning_rate": 0.00025772761101463653,
+      "loss": 0.6441,
+      "step": 1646
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.7484211921691895,
+      "learning_rate": 0.00025769039940461423,
+      "loss": 0.7046,
+      "step": 1647
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.8884170055389404,
+      "learning_rate": 0.0002576531877945919,
+      "loss": 0.9336,
+      "step": 1648
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 5.420833110809326,
+      "learning_rate": 0.0002576159761845696,
+      "loss": 1.2645,
+      "step": 1649
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.9601593017578125,
+      "learning_rate": 0.00025757876457454723,
+      "loss": 0.7021,
+      "step": 1650
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.882598400115967,
+      "learning_rate": 0.0002575415529645249,
+      "loss": 1.571,
+      "step": 1651
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.408437490463257,
+      "learning_rate": 0.0002575043413545026,
+      "loss": 0.6583,
+      "step": 1652
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.7199838161468506,
+      "learning_rate": 0.0002574671297444803,
+      "loss": 1.0358,
+      "step": 1653
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.46736216545105,
+      "learning_rate": 0.00025742991813445794,
+      "loss": 0.7032,
+      "step": 1654
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8680689334869385,
+      "learning_rate": 0.0002573927065244356,
+      "loss": 0.3942,
+      "step": 1655
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.436976671218872,
+      "learning_rate": 0.0002573554949144133,
+      "loss": 0.928,
+      "step": 1656
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8157801628112793,
+      "learning_rate": 0.00025731828330439094,
+      "loss": 0.5851,
+      "step": 1657
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2245826721191406,
+      "learning_rate": 0.00025728107169436864,
+      "loss": 0.829,
+      "step": 1658
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.383336067199707,
+      "learning_rate": 0.0002572438600843463,
+      "loss": 0.5824,
+      "step": 1659
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.794093370437622,
+      "learning_rate": 0.000257206648474324,
+      "loss": 0.7381,
+      "step": 1660
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 6.125317573547363,
+      "learning_rate": 0.00025716943686430164,
+      "loss": 0.6719,
+      "step": 1661
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.031485080718994,
+      "learning_rate": 0.0002571322252542793,
+      "loss": 1.0915,
+      "step": 1662
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.9629569053649902,
+      "learning_rate": 0.000257095013644257,
+      "loss": 0.8506,
+      "step": 1663
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.803297281265259,
+      "learning_rate": 0.00025705780203423464,
+      "loss": 0.6614,
+      "step": 1664
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 7.903172016143799,
+      "learning_rate": 0.00025702059042421234,
+      "loss": 1.5172,
+      "step": 1665
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.132394313812256,
+      "learning_rate": 0.00025698337881419,
+      "loss": 0.6185,
+      "step": 1666
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9272037744522095,
+      "learning_rate": 0.0002569461672041677,
+      "loss": 0.2627,
+      "step": 1667
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.5171618461608887,
+      "learning_rate": 0.00025690895559414534,
+      "loss": 0.8559,
+      "step": 1668
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.361370325088501,
+      "learning_rate": 0.000256871743984123,
+      "loss": 0.6559,
+      "step": 1669
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.13026762008667,
+      "learning_rate": 0.0002568345323741007,
+      "loss": 1.6543,
+      "step": 1670
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.8648040294647217,
+      "learning_rate": 0.0002567973207640784,
+      "loss": 0.784,
+      "step": 1671
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.692941188812256,
+      "learning_rate": 0.00025676010915405605,
+      "loss": 0.5386,
+      "step": 1672
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.755861520767212,
+      "learning_rate": 0.0002567228975440337,
+      "loss": 0.2788,
+      "step": 1673
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 5.936482906341553,
+      "learning_rate": 0.0002566856859340114,
+      "loss": 1.7085,
+      "step": 1674
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": NaN,
+      "learning_rate": 0.0002566856859340114,
+      "loss": 0.1353,
+      "step": 1675
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.012645959854126,
+      "learning_rate": 0.00025664847432398905,
+      "loss": 0.9096,
+      "step": 1676
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.613264799118042,
+      "learning_rate": 0.00025661126271396675,
+      "loss": 0.5428,
+      "step": 1677
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.0009877681732178,
+      "learning_rate": 0.0002565740511039444,
+      "loss": 0.5408,
+      "step": 1678
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.448207378387451,
+      "learning_rate": 0.0002565368394939221,
+      "loss": 0.5128,
+      "step": 1679
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.7628707885742188,
+      "learning_rate": 0.00025649962788389975,
+      "loss": 0.2766,
+      "step": 1680
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 18.24505043029785,
+      "learning_rate": 0.0002564624162738774,
+      "loss": 4.3011,
+      "step": 1681
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.7175798416137695,
+      "learning_rate": 0.0002564252046638551,
+      "loss": 0.9797,
+      "step": 1682
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.9099786281585693,
+      "learning_rate": 0.00025638799305383275,
+      "loss": 0.6673,
+      "step": 1683
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.798135280609131,
+      "learning_rate": 0.00025635078144381045,
+      "loss": 0.6912,
+      "step": 1684
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.6960597038269043,
+      "learning_rate": 0.00025631356983378816,
+      "loss": 0.3386,
+      "step": 1685
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.622521162033081,
+      "learning_rate": 0.0002562763582237658,
+      "loss": 0.3305,
+      "step": 1686
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.7763274908065796,
+      "learning_rate": 0.00025623914661374346,
+      "loss": 0.3437,
+      "step": 1687
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.253156900405884,
+      "learning_rate": 0.0002562019350037211,
+      "loss": 0.374,
+      "step": 1688
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.9651002883911133,
+      "learning_rate": 0.0002561647233936988,
+      "loss": 0.5962,
+      "step": 1689
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.5626463890075684,
+      "learning_rate": 0.0002561275117836765,
+      "loss": 0.7751,
+      "step": 1690
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.0992138385772705,
+      "learning_rate": 0.00025609030017365416,
+      "loss": 0.7268,
+      "step": 1691
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.610893487930298,
+      "learning_rate": 0.00025605308856363186,
+      "loss": 0.3764,
+      "step": 1692
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 5.090748310089111,
+      "learning_rate": 0.0002560158769536095,
+      "loss": 1.1423,
+      "step": 1693
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.872439861297607,
+      "learning_rate": 0.00025597866534358716,
+      "loss": 0.7031,
+      "step": 1694
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.71915340423584,
+      "learning_rate": 0.00025594145373356486,
+      "loss": 1.1228,
+      "step": 1695
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.626314163208008,
+      "learning_rate": 0.0002559042421235425,
+      "loss": 0.5694,
+      "step": 1696
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 6.56594181060791,
+      "learning_rate": 0.0002558670305135202,
+      "loss": 1.0496,
+      "step": 1697
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 6.181338310241699,
+      "learning_rate": 0.00025582981890349786,
+      "loss": 2.4125,
+      "step": 1698
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.708691358566284,
+      "learning_rate": 0.00025579260729347556,
+      "loss": 0.3454,
+      "step": 1699
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.9663429260253906,
+      "learning_rate": 0.0002557553956834532,
+      "loss": 0.5035,
+      "step": 1700
+    },
+    {
+      "epoch": 0.6,
+      "eval_loss": 0.7400824427604675,
+      "eval_runtime": 50.2926,
+      "eval_samples_per_second": 43.108,
+      "eval_steps_per_second": 10.777,
+      "eval_wer": 0.6105345096004151,
+      "step": 1700
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.6513137817382812,
+      "learning_rate": 0.00025571818407343086,
+      "loss": 1.472,
+      "step": 1701
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8909034729003906,
+      "learning_rate": 0.00025568097246340856,
+      "loss": 0.7735,
+      "step": 1702
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8732975721359253,
+      "learning_rate": 0.00025564376085338627,
+      "loss": 0.6441,
+      "step": 1703
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.949060082435608,
+      "learning_rate": 0.0002556065492433639,
+      "loss": 0.6141,
+      "step": 1704
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.6796951293945312,
+      "learning_rate": 0.00025556933763334157,
+      "loss": 0.3948,
+      "step": 1705
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.200995922088623,
+      "learning_rate": 0.0002555321260233192,
+      "loss": 0.4608,
+      "step": 1706
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.7621562480926514,
+      "learning_rate": 0.0002554949144132969,
+      "loss": 0.6803,
+      "step": 1707
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.656009197235107,
+      "learning_rate": 0.0002554577028032746,
+      "loss": 0.7062,
+      "step": 1708
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.225820302963257,
+      "learning_rate": 0.00025542049119325227,
+      "loss": 1.0558,
+      "step": 1709
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.8542261123657227,
+      "learning_rate": 0.00025538327958322997,
+      "loss": 1.0637,
+      "step": 1710
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.9913618564605713,
+      "learning_rate": 0.0002553460679732076,
+      "loss": 0.4731,
+      "step": 1711
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.990241289138794,
+      "learning_rate": 0.00025530885636318527,
+      "loss": 0.6727,
+      "step": 1712
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.4291019439697266,
+      "learning_rate": 0.00025527164475316297,
+      "loss": 0.4963,
+      "step": 1713
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.358236312866211,
+      "learning_rate": 0.0002552344331431406,
+      "loss": 0.3102,
+      "step": 1714
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.8951985836029053,
+      "learning_rate": 0.0002551972215331183,
+      "loss": 0.6374,
+      "step": 1715
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.7932424545288086,
+      "learning_rate": 0.00025516000992309597,
+      "loss": 0.3472,
+      "step": 1716
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.9364991188049316,
+      "learning_rate": 0.0002551227983130737,
+      "loss": 1.3303,
+      "step": 1717
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.639146566390991,
+      "learning_rate": 0.0002550855867030513,
+      "loss": 0.2635,
+      "step": 1718
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.6595664024353027,
+      "learning_rate": 0.00025504837509302897,
+      "loss": 0.4851,
+      "step": 1719
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.7507216930389404,
+      "learning_rate": 0.0002550111634830067,
+      "loss": 0.7927,
+      "step": 1720
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 6.605137348175049,
+      "learning_rate": 0.0002549739518729844,
+      "loss": 1.9443,
+      "step": 1721
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.5037612915039062,
+      "learning_rate": 0.000254936740262962,
+      "loss": 0.4611,
+      "step": 1722
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.2894933223724365,
+      "learning_rate": 0.0002548995286529397,
+      "loss": 0.4025,
+      "step": 1723
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.477785587310791,
+      "learning_rate": 0.0002548623170429174,
+      "loss": 1.0904,
+      "step": 1724
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9483599662780762,
+      "learning_rate": 0.000254825105432895,
+      "loss": 0.276,
+      "step": 1725
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0058553218841553,
+      "learning_rate": 0.00025478789382287273,
+      "loss": 0.5288,
+      "step": 1726
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.122645616531372,
+      "learning_rate": 0.0002547506822128504,
+      "loss": 0.5928,
+      "step": 1727
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.18483829498291,
+      "learning_rate": 0.0002547134706028281,
+      "loss": 0.579,
+      "step": 1728
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4548044204711914,
+      "learning_rate": 0.00025467625899280573,
+      "loss": 0.5435,
+      "step": 1729
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.351948261260986,
+      "learning_rate": 0.0002546390473827834,
+      "loss": 0.7439,
+      "step": 1730
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.7342119216918945,
+      "learning_rate": 0.0002546018357727611,
+      "loss": 0.5848,
+      "step": 1731
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.438290596008301,
+      "learning_rate": 0.00025456462416273873,
+      "loss": 0.4699,
+      "step": 1732
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.6222329139709473,
+      "learning_rate": 0.00025452741255271643,
+      "loss": 0.267,
+      "step": 1733
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.6478686332702637,
+      "learning_rate": 0.00025449020094269414,
+      "loss": 0.5306,
+      "step": 1734
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.7666268348693848,
+      "learning_rate": 0.0002544529893326718,
+      "loss": 0.676,
+      "step": 1735
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.8781347274780273,
+      "learning_rate": 0.00025441577772264943,
+      "loss": 0.7332,
+      "step": 1736
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.474586486816406,
+      "learning_rate": 0.0002543785661126271,
+      "loss": 1.1677,
+      "step": 1737
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.475315809249878,
+      "learning_rate": 0.0002543413545026048,
+      "loss": 0.4186,
+      "step": 1738
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4638478755950928,
+      "learning_rate": 0.0002543041428925825,
+      "loss": 0.5006,
+      "step": 1739
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.56996750831604,
+      "learning_rate": 0.00025426693128256014,
+      "loss": 0.3888,
+      "step": 1740
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 6.963738918304443,
+      "learning_rate": 0.00025422971967253784,
+      "loss": 0.6737,
+      "step": 1741
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.476889133453369,
+      "learning_rate": 0.0002541925080625155,
+      "loss": 0.2271,
+      "step": 1742
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.8654210567474365,
+      "learning_rate": 0.00025415529645249314,
+      "loss": 1.5383,
+      "step": 1743
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 6.272684097290039,
+      "learning_rate": 0.00025411808484247084,
+      "loss": 1.1683,
+      "step": 1744
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0367603302001953,
+      "learning_rate": 0.0002540808732324485,
+      "loss": 0.1529,
+      "step": 1745
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.270781993865967,
+      "learning_rate": 0.0002540436616224262,
+      "loss": 0.3932,
+      "step": 1746
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.1253650188446045,
+      "learning_rate": 0.00025400645001240384,
+      "loss": 0.2344,
+      "step": 1747
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.384814739227295,
+      "learning_rate": 0.0002539692384023815,
+      "loss": 0.6099,
+      "step": 1748
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.7127060890197754,
+      "learning_rate": 0.0002539320267923592,
+      "loss": 0.7978,
+      "step": 1749
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 8.662912368774414,
+      "learning_rate": 0.00025389481518233684,
+      "loss": 1.5128,
+      "step": 1750
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.237752914428711,
+      "learning_rate": 0.00025385760357231454,
+      "loss": 1.3841,
+      "step": 1751
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 6.472276210784912,
+      "learning_rate": 0.00025382039196229225,
+      "loss": 1.1862,
+      "step": 1752
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.888750076293945,
+      "learning_rate": 0.0002537831803522699,
+      "loss": 1.139,
+      "step": 1753
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.32738733291626,
+      "learning_rate": 0.00025374596874224754,
+      "loss": 0.6998,
+      "step": 1754
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.718627452850342,
+      "learning_rate": 0.0002537087571322252,
+      "loss": 1.035,
+      "step": 1755
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6345198154449463,
+      "learning_rate": 0.0002536715455222029,
+      "loss": 0.8893,
+      "step": 1756
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.600464105606079,
+      "learning_rate": 0.0002536343339121806,
+      "loss": 0.7878,
+      "step": 1757
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.2108471393585205,
+      "learning_rate": 0.00025359712230215825,
+      "loss": 0.8279,
+      "step": 1758
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.6531155109405518,
+      "learning_rate": 0.00025355991069213595,
+      "loss": 1.1393,
+      "step": 1759
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.240154504776001,
+      "learning_rate": 0.0002535226990821136,
+      "loss": 0.5395,
+      "step": 1760
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.4271785020828247,
+      "learning_rate": 0.00025348548747209125,
+      "loss": 0.2189,
+      "step": 1761
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.7750396728515625,
+      "learning_rate": 0.00025344827586206895,
+      "loss": 0.7323,
+      "step": 1762
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.3077313899993896,
+      "learning_rate": 0.0002534110642520466,
+      "loss": 0.4554,
+      "step": 1763
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.405827760696411,
+      "learning_rate": 0.0002533738526420243,
+      "loss": 0.6688,
+      "step": 1764
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.642482042312622,
+      "learning_rate": 0.00025333664103200195,
+      "loss": 0.2194,
+      "step": 1765
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6313064098358154,
+      "learning_rate": 0.00025329942942197965,
+      "loss": 0.6633,
+      "step": 1766
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 10.008116722106934,
+      "learning_rate": 0.0002532622178119573,
+      "loss": 2.5217,
+      "step": 1767
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.850469589233398,
+      "learning_rate": 0.00025322500620193495,
+      "loss": 1.1002,
+      "step": 1768
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.958583354949951,
+      "learning_rate": 0.00025318779459191265,
+      "loss": 0.5638,
+      "step": 1769
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5077438354492188,
+      "learning_rate": 0.00025315058298189036,
+      "loss": 0.4724,
+      "step": 1770
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 5.4337687492370605,
+      "learning_rate": 0.000253113371371868,
+      "loss": 1.4577,
+      "step": 1771
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.459052801132202,
+      "learning_rate": 0.00025307615976184565,
+      "loss": 0.6951,
+      "step": 1772
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 8.31182861328125,
+      "learning_rate": 0.00025303894815182336,
+      "loss": 1.0552,
+      "step": 1773
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.019193172454834,
+      "learning_rate": 0.000253001736541801,
+      "loss": 0.2639,
+      "step": 1774
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": NaN,
+      "learning_rate": 0.000253001736541801,
+      "loss": 0.1885,
+      "step": 1775
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.729272842407227,
+      "learning_rate": 0.0002529645249317787,
+      "loss": 1.1689,
+      "step": 1776
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.806692361831665,
+      "learning_rate": 0.00025292731332175636,
+      "loss": 0.9207,
+      "step": 1777
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.110698699951172,
+      "learning_rate": 0.00025289010171173406,
+      "loss": 0.6481,
+      "step": 1778
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.653905153274536,
+      "learning_rate": 0.0002528528901017117,
+      "loss": 0.7404,
+      "step": 1779
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.4081614017486572,
+      "learning_rate": 0.00025281567849168936,
+      "loss": 0.7589,
+      "step": 1780
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.3491172790527344,
+      "learning_rate": 0.00025277846688166706,
+      "loss": 0.5285,
+      "step": 1781
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.7228968143463135,
+      "learning_rate": 0.0002527412552716447,
+      "loss": 0.947,
+      "step": 1782
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.284248113632202,
+      "learning_rate": 0.0002527040436616224,
+      "loss": 0.4871,
+      "step": 1783
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.298612117767334,
+      "learning_rate": 0.0002526668320516001,
+      "loss": 0.6996,
+      "step": 1784
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.53757643699646,
+      "learning_rate": 0.00025262962044157776,
+      "loss": 0.3541,
+      "step": 1785
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.7956089973449707,
+      "learning_rate": 0.0002525924088315554,
+      "loss": 0.3465,
+      "step": 1786
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.37371826171875,
+      "learning_rate": 0.0002525551972215331,
+      "loss": 0.5631,
+      "step": 1787
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.794459819793701,
+      "learning_rate": 0.00025251798561151076,
+      "loss": 0.8957,
+      "step": 1788
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2786507606506348,
+      "learning_rate": 0.00025248077400148847,
+      "loss": 0.5798,
+      "step": 1789
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.390256404876709,
+      "learning_rate": 0.0002524435623914661,
+      "loss": 0.4864,
+      "step": 1790
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.0629777908325195,
+      "learning_rate": 0.00025240635078144376,
+      "loss": 0.6592,
+      "step": 1791
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.212480068206787,
+      "learning_rate": 0.00025236913917142147,
+      "loss": 1.3613,
+      "step": 1792
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.64257287979126,
+      "learning_rate": 0.0002523319275613991,
+      "loss": 0.4347,
+      "step": 1793
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.668285369873047,
+      "learning_rate": 0.0002522947159513768,
+      "loss": 0.7722,
+      "step": 1794
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.8437860012054443,
+      "learning_rate": 0.00025225750434135447,
+      "loss": 0.3561,
+      "step": 1795
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.092412233352661,
+      "learning_rate": 0.00025222029273133217,
+      "loss": 0.4881,
+      "step": 1796
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.986132264137268,
+      "learning_rate": 0.0002521830811213098,
+      "loss": 0.1549,
+      "step": 1797
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.6645147800445557,
+      "learning_rate": 0.00025214586951128747,
+      "loss": 0.4356,
+      "step": 1798
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.3920583724975586,
+      "learning_rate": 0.00025210865790126517,
+      "loss": 0.3476,
+      "step": 1799
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.257504463195801,
+      "learning_rate": 0.0002520714462912429,
+      "loss": 0.6923,
+      "step": 1800
+    },
+    {
+      "epoch": 0.63,
+      "eval_loss": 0.6619295477867126,
+      "eval_runtime": 50.7331,
+      "eval_samples_per_second": 42.733,
+      "eval_steps_per_second": 10.683,
+      "eval_wer": 0.5835495588998443,
+      "step": 1800
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.3466756343841553,
+      "learning_rate": 0.0002520342346812205,
+      "loss": 0.6547,
+      "step": 1801
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.27302885055542,
+      "learning_rate": 0.0002519970230711982,
+      "loss": 1.0436,
+      "step": 1802
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.4561309814453125,
+      "learning_rate": 0.0002519598114611759,
+      "loss": 0.8965,
+      "step": 1803
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.3739259243011475,
+      "learning_rate": 0.0002519225998511535,
+      "loss": 0.705,
+      "step": 1804
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.3616445064544678,
+      "learning_rate": 0.0002518853882411312,
+      "loss": 0.8433,
+      "step": 1805
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.9913322925567627,
+      "learning_rate": 0.0002518481766311089,
+      "loss": 0.4734,
+      "step": 1806
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.3768415451049805,
+      "learning_rate": 0.0002518109650210866,
+      "loss": 0.4895,
+      "step": 1807
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2762320041656494,
+      "learning_rate": 0.0002517737534110642,
+      "loss": 0.621,
+      "step": 1808
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.1981852054595947,
+      "learning_rate": 0.00025173654180104193,
+      "loss": 0.5364,
+      "step": 1809
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.924687147140503,
+      "learning_rate": 0.0002516993301910196,
+      "loss": 0.2811,
+      "step": 1810
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.7158939838409424,
+      "learning_rate": 0.0002516621185809972,
+      "loss": 0.6797,
+      "step": 1811
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.226498603820801,
+      "learning_rate": 0.00025162490697097493,
+      "loss": 0.3998,
+      "step": 1812
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0350279808044434,
+      "learning_rate": 0.0002515876953609526,
+      "loss": 0.2877,
+      "step": 1813
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.871344089508057,
+      "learning_rate": 0.0002515504837509303,
+      "loss": 0.6181,
+      "step": 1814
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.3787949085235596,
+      "learning_rate": 0.00025151327214090793,
+      "loss": 0.7396,
+      "step": 1815
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.748871326446533,
+      "learning_rate": 0.0002514760605308856,
+      "loss": 0.2645,
+      "step": 1816
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.798830986022949,
+      "learning_rate": 0.0002514388489208633,
+      "loss": 0.5294,
+      "step": 1817
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.125420570373535,
+      "learning_rate": 0.000251401637310841,
+      "loss": 0.565,
+      "step": 1818
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.471548080444336,
+      "learning_rate": 0.00025136442570081863,
+      "loss": 0.5297,
+      "step": 1819
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.885577917098999,
+      "learning_rate": 0.00025132721409079633,
+      "loss": 0.3337,
+      "step": 1820
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.8695416450500488,
+      "learning_rate": 0.000251290002480774,
+      "loss": 0.2729,
+      "step": 1821
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.600594520568848,
+      "learning_rate": 0.00025125279087075163,
+      "loss": 1.8971,
+      "step": 1822
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 6.382452011108398,
+      "learning_rate": 0.00025121557926072933,
+      "loss": 2.4557,
+      "step": 1823
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 7.36460542678833,
+      "learning_rate": 0.000251178367650707,
+      "loss": 2.337,
+      "step": 1824
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.724518299102783,
+      "learning_rate": 0.0002511411560406847,
+      "loss": 1.9224,
+      "step": 1825
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.918628454208374,
+      "learning_rate": 0.00025110394443066234,
+      "loss": 1.1783,
+      "step": 1826
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.5602731704711914,
+      "learning_rate": 0.00025106673282064004,
+      "loss": 1.2087,
+      "step": 1827
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.9868627786636353,
+      "learning_rate": 0.0002510295212106177,
+      "loss": 0.5769,
+      "step": 1828
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.202514410018921,
+      "learning_rate": 0.00025099230960059534,
+      "loss": 1.2293,
+      "step": 1829
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.9510756731033325,
+      "learning_rate": 0.00025095509799057304,
+      "loss": 0.4025,
+      "step": 1830
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.144929885864258,
+      "learning_rate": 0.00025091788638055074,
+      "loss": 0.4471,
+      "step": 1831
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.4620763063430786,
+      "learning_rate": 0.0002508806747705284,
+      "loss": 0.2977,
+      "step": 1832
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.262373208999634,
+      "learning_rate": 0.00025084346316050604,
+      "loss": 0.6991,
+      "step": 1833
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.069397211074829,
+      "learning_rate": 0.00025080625155048374,
+      "loss": 0.4164,
+      "step": 1834
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 13.504817008972168,
+      "learning_rate": 0.0002507690399404614,
+      "loss": 3.6967,
+      "step": 1835
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.4897937774658203,
+      "learning_rate": 0.0002507318283304391,
+      "loss": 0.8833,
+      "step": 1836
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.5838449001312256,
+      "learning_rate": 0.00025069461672041674,
+      "loss": 0.6879,
+      "step": 1837
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.084447860717773,
+      "learning_rate": 0.00025065740511039444,
+      "loss": 0.9948,
+      "step": 1838
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.7266480922698975,
+      "learning_rate": 0.0002506201935003721,
+      "loss": 0.978,
+      "step": 1839
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.592857599258423,
+      "learning_rate": 0.00025058298189034974,
+      "loss": 0.4904,
+      "step": 1840
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.836439609527588,
+      "learning_rate": 0.00025054577028032744,
+      "loss": 0.4452,
+      "step": 1841
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.47084379196167,
+      "learning_rate": 0.0002505085586703051,
+      "loss": 0.7633,
+      "step": 1842
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.29998517036438,
+      "learning_rate": 0.0002504713470602828,
+      "loss": 1.2435,
+      "step": 1843
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.52744460105896,
+      "learning_rate": 0.0002504341354502605,
+      "loss": 0.3426,
+      "step": 1844
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3847482204437256,
+      "learning_rate": 0.00025039692384023815,
+      "loss": 0.2179,
+      "step": 1845
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.141510009765625,
+      "learning_rate": 0.0002503597122302158,
+      "loss": 0.6102,
+      "step": 1846
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.136857509613037,
+      "learning_rate": 0.00025032250062019345,
+      "loss": 1.0573,
+      "step": 1847
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.8954155445098877,
+      "learning_rate": 0.00025028528901017115,
+      "loss": 0.4377,
+      "step": 1848
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4569027423858643,
+      "learning_rate": 0.00025024807740014885,
+      "loss": 0.2517,
+      "step": 1849
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.620333671569824,
+      "learning_rate": 0.0002502108657901265,
+      "loss": 1.7501,
+      "step": 1850
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.291147470474243,
+      "learning_rate": 0.0002501736541801042,
+      "loss": 1.2419,
+      "step": 1851
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.5810606479644775,
+      "learning_rate": 0.00025013644257008185,
+      "loss": 1.0098,
+      "step": 1852
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.492764472961426,
+      "learning_rate": 0.0002500992309600595,
+      "loss": 0.8263,
+      "step": 1853
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.030451536178589,
+      "learning_rate": 0.0002500620193500372,
+      "loss": 0.7977,
+      "step": 1854
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.7013792991638184,
+      "learning_rate": 0.00025002480774001485,
+      "loss": 0.7732,
+      "step": 1855
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.162282943725586,
+      "learning_rate": 0.00024998759612999255,
+      "loss": 0.8146,
+      "step": 1856
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.0203332901000977,
+      "learning_rate": 0.0002499503845199702,
+      "loss": 0.8234,
+      "step": 1857
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6350369453430176,
+      "learning_rate": 0.00024991317290994785,
+      "loss": 0.3615,
+      "step": 1858
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.785946846008301,
+      "learning_rate": 0.00024987596129992556,
+      "loss": 0.6719,
+      "step": 1859
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.30195426940918,
+      "learning_rate": 0.0002498387496899032,
+      "loss": 0.4817,
+      "step": 1860
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.285341262817383,
+      "learning_rate": 0.0002498015380798809,
+      "loss": 0.4845,
+      "step": 1861
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.9334990978240967,
+      "learning_rate": 0.0002497643264698586,
+      "loss": 0.6951,
+      "step": 1862
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.6328988075256348,
+      "learning_rate": 0.00024972711485983626,
+      "loss": 0.3517,
+      "step": 1863
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.6665728092193604,
+      "learning_rate": 0.0002496899032498139,
+      "loss": 1.0003,
+      "step": 1864
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.054015636444092,
+      "learning_rate": 0.00024965269163979156,
+      "loss": 0.4366,
+      "step": 1865
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.34871506690979,
+      "learning_rate": 0.00024961548002976926,
+      "loss": 0.3632,
+      "step": 1866
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.7466020584106445,
+      "learning_rate": 0.00024957826841974696,
+      "loss": 0.2544,
+      "step": 1867
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3395614624023438,
+      "learning_rate": 0.0002495410568097246,
+      "loss": 0.2607,
+      "step": 1868
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.0486836433410645,
+      "learning_rate": 0.0002495038451997023,
+      "loss": 0.6904,
+      "step": 1869
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.611022472381592,
+      "learning_rate": 0.00024946663358967996,
+      "loss": 0.643,
+      "step": 1870
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.7561452388763428,
+      "learning_rate": 0.0002494294219796576,
+      "loss": 0.2442,
+      "step": 1871
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.131147384643555,
+      "learning_rate": 0.0002493922103696353,
+      "loss": 0.6293,
+      "step": 1872
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 6.987359523773193,
+      "learning_rate": 0.00024935499875961296,
+      "loss": 0.5611,
+      "step": 1873
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.005058765411377,
+      "learning_rate": 0.00024931778714959066,
+      "loss": 1.9181,
+      "step": 1874
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.872385501861572,
+      "learning_rate": 0.0002492805755395683,
+      "loss": 0.4414,
+      "step": 1875
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.087235927581787,
+      "learning_rate": 0.000249243363929546,
+      "loss": 1.1648,
+      "step": 1876
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.866982936859131,
+      "learning_rate": 0.00024920615231952367,
+      "loss": 0.8997,
+      "step": 1877
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.6549713611602783,
+      "learning_rate": 0.0002491689407095013,
+      "loss": 0.7768,
+      "step": 1878
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.923379421234131,
+      "learning_rate": 0.000249131729099479,
+      "loss": 0.7754,
+      "step": 1879
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.286487340927124,
+      "learning_rate": 0.0002490945174894567,
+      "loss": 0.5886,
+      "step": 1880
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.357010841369629,
+      "learning_rate": 0.00024905730587943437,
+      "loss": 0.5951,
+      "step": 1881
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.6229811906814575,
+      "learning_rate": 0.000249020094269412,
+      "loss": 0.4376,
+      "step": 1882
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4344239234924316,
+      "learning_rate": 0.0002489828826593897,
+      "loss": 0.7466,
+      "step": 1883
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 6.589191436767578,
+      "learning_rate": 0.00024894567104936737,
+      "loss": 0.6586,
+      "step": 1884
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.2534403800964355,
+      "learning_rate": 0.00024890845943934507,
+      "loss": 0.7062,
+      "step": 1885
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.400862455368042,
+      "learning_rate": 0.0002488712478293227,
+      "loss": 0.306,
+      "step": 1886
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.253848075866699,
+      "learning_rate": 0.0002488340362193004,
+      "loss": 0.6433,
+      "step": 1887
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.553795337677002,
+      "learning_rate": 0.00024879682460927807,
+      "loss": 0.4091,
+      "step": 1888
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4395694732666016,
+      "learning_rate": 0.0002487596129992557,
+      "loss": 0.5482,
+      "step": 1889
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.8321399688720703,
+      "learning_rate": 0.0002487224013892334,
+      "loss": 0.6352,
+      "step": 1890
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.5882153511047363,
+      "learning_rate": 0.00024868518977921107,
+      "loss": 0.4969,
+      "step": 1891
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.412918567657471,
+      "learning_rate": 0.0002486479781691888,
+      "loss": 1.0272,
+      "step": 1892
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.513781547546387,
+      "learning_rate": 0.0002486107665591665,
+      "loss": 1.2136,
+      "step": 1893
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.6732091903686523,
+      "learning_rate": 0.0002485735549491441,
+      "loss": 0.3256,
+      "step": 1894
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.6457021236419678,
+      "learning_rate": 0.0002485363433391218,
+      "loss": 0.4514,
+      "step": 1895
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.561769485473633,
+      "learning_rate": 0.0002484991317290994,
+      "loss": 0.6833,
+      "step": 1896
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.9853644371032715,
+      "learning_rate": 0.0002484619201190771,
+      "loss": 1.2547,
+      "step": 1897
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9570162296295166,
+      "learning_rate": 0.00024842470850905483,
+      "loss": 0.2446,
+      "step": 1898
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.1816675662994385,
+      "learning_rate": 0.0002483874968990325,
+      "loss": 0.4237,
+      "step": 1899
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 8.543338775634766,
+      "learning_rate": 0.0002483502852890101,
+      "loss": 1.1266,
+      "step": 1900
+    },
+    {
+      "epoch": 0.67,
+      "eval_loss": 0.6919850707054138,
+      "eval_runtime": 51.2516,
+      "eval_samples_per_second": 42.301,
+      "eval_steps_per_second": 10.575,
+      "eval_wer": 0.6012800553537451,
+      "step": 1900
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.464935302734375,
+      "learning_rate": 0.00024831307367898783,
+      "loss": 0.9893,
+      "step": 1901
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.831908941268921,
+      "learning_rate": 0.0002482758620689655,
+      "loss": 0.9191,
+      "step": 1902
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.71111536026001,
+      "learning_rate": 0.0002482386504589432,
+      "loss": 1.5041,
+      "step": 1903
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.486187219619751,
+      "learning_rate": 0.00024820143884892083,
+      "loss": 0.4669,
+      "step": 1904
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.497227191925049,
+      "learning_rate": 0.00024816422723889853,
+      "loss": 0.8797,
+      "step": 1905
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.937802791595459,
+      "learning_rate": 0.0002481270156288762,
+      "loss": 0.7039,
+      "step": 1906
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5400185585021973,
+      "learning_rate": 0.00024808980401885383,
+      "loss": 0.7996,
+      "step": 1907
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.81622576713562,
+      "learning_rate": 0.00024805259240883153,
+      "loss": 0.5797,
+      "step": 1908
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.533505916595459,
+      "learning_rate": 0.0002480153807988092,
+      "loss": 0.7682,
+      "step": 1909
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5406887531280518,
+      "learning_rate": 0.0002479781691887869,
+      "loss": 0.8529,
+      "step": 1910
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.635380268096924,
+      "learning_rate": 0.0002479409575787646,
+      "loss": 0.5315,
+      "step": 1911
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.293522834777832,
+      "learning_rate": 0.00024790374596874224,
+      "loss": 0.762,
+      "step": 1912
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.7569010257720947,
+      "learning_rate": 0.0002478665343587199,
+      "loss": 0.8445,
+      "step": 1913
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.208041667938232,
+      "learning_rate": 0.00024782932274869753,
+      "loss": 0.7308,
+      "step": 1914
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.8499300479888916,
+      "learning_rate": 0.00024779211113867524,
+      "loss": 0.7733,
+      "step": 1915
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9192874431610107,
+      "learning_rate": 0.00024775489952865294,
+      "loss": 0.3927,
+      "step": 1916
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.4525372982025146,
+      "learning_rate": 0.0002477176879186306,
+      "loss": 0.2387,
+      "step": 1917
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.631014347076416,
+      "learning_rate": 0.0002476804763086083,
+      "loss": 0.7168,
+      "step": 1918
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9727121591567993,
+      "learning_rate": 0.00024764326469858594,
+      "loss": 0.5055,
+      "step": 1919
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 8.828970909118652,
+      "learning_rate": 0.0002476060530885636,
+      "loss": 1.4732,
+      "step": 1920
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.060351848602295,
+      "learning_rate": 0.0002475688414785413,
+      "loss": 1.418,
+      "step": 1921
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0158917903900146,
+      "learning_rate": 0.00024753162986851894,
+      "loss": 0.2002,
+      "step": 1922
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.4808578491210938,
+      "learning_rate": 0.00024749441825849664,
+      "loss": 0.5887,
+      "step": 1923
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 5.383602142333984,
+      "learning_rate": 0.0002474572066484743,
+      "loss": 1.5568,
+      "step": 1924
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.389026165008545,
+      "learning_rate": 0.000247419995038452,
+      "loss": 0.3765,
+      "step": 1925
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.9787375926971436,
+      "learning_rate": 0.00024738278342842964,
+      "loss": 0.9215,
+      "step": 1926
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.7015265226364136,
+      "learning_rate": 0.0002473455718184073,
+      "loss": 0.5214,
+      "step": 1927
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.7144050598144531,
+      "learning_rate": 0.000247308360208385,
+      "loss": 0.6626,
+      "step": 1928
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.8831634521484375,
+      "learning_rate": 0.0002472711485983627,
+      "loss": 0.7753,
+      "step": 1929
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.180262565612793,
+      "learning_rate": 0.00024723393698834035,
+      "loss": 0.5357,
+      "step": 1930
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.9402575492858887,
+      "learning_rate": 0.000247196725378318,
+      "loss": 0.4598,
+      "step": 1931
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.6572482585906982,
+      "learning_rate": 0.0002471595137682957,
+      "loss": 0.5624,
+      "step": 1932
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.5817763805389404,
+      "learning_rate": 0.00024712230215827335,
+      "loss": 0.6215,
+      "step": 1933
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.858384132385254,
+      "learning_rate": 0.00024708509054825105,
+      "loss": 0.879,
+      "step": 1934
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.5925073623657227,
+      "learning_rate": 0.0002470478789382287,
+      "loss": 0.7597,
+      "step": 1935
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.7075647115707397,
+      "learning_rate": 0.0002470106673282064,
+      "loss": 0.3585,
+      "step": 1936
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2654693126678467,
+      "learning_rate": 0.00024697345571818405,
+      "loss": 0.4981,
+      "step": 1937
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.9442179203033447,
+      "learning_rate": 0.0002469362441081617,
+      "loss": 0.5214,
+      "step": 1938
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.89056396484375,
+      "learning_rate": 0.0002468990324981394,
+      "loss": 0.4043,
+      "step": 1939
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2005202770233154,
+      "learning_rate": 0.00024686182088811705,
+      "loss": 0.2301,
+      "step": 1940
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.534944534301758,
+      "learning_rate": 0.00024682460927809475,
+      "loss": 0.8182,
+      "step": 1941
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.317547082901001,
+      "learning_rate": 0.0002467873976680724,
+      "loss": 0.8466,
+      "step": 1942
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.431687593460083,
+      "learning_rate": 0.0002467501860580501,
+      "loss": 0.4799,
+      "step": 1943
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.7903997898101807,
+      "learning_rate": 0.00024671297444802775,
+      "loss": 0.4264,
+      "step": 1944
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.921651840209961,
+      "learning_rate": 0.0002466757628380054,
+      "loss": 0.3343,
+      "step": 1945
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.756716728210449,
+      "learning_rate": 0.0002466385512279831,
+      "loss": 1.4059,
+      "step": 1946
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.8157451152801514,
+      "learning_rate": 0.0002466013396179608,
+      "loss": 0.2149,
+      "step": 1947
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.292349338531494,
+      "learning_rate": 0.00024656412800793846,
+      "loss": 0.6947,
+      "step": 1948
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 5.523711204528809,
+      "learning_rate": 0.0002465269163979161,
+      "loss": 1.8716,
+      "step": 1949
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 5.184019088745117,
+      "learning_rate": 0.0002464897047878938,
+      "loss": 1.5156,
+      "step": 1950
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 9.27737808227539,
+      "learning_rate": 0.00024645249317787146,
+      "loss": 1.8832,
+      "step": 1951
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 11.288307189941406,
+      "learning_rate": 0.00024641528156784916,
+      "loss": 1.2195,
+      "step": 1952
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 6.941704273223877,
+      "learning_rate": 0.0002463780699578268,
+      "loss": 1.0644,
+      "step": 1953
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 5.17105770111084,
+      "learning_rate": 0.0002463408583478045,
+      "loss": 1.0977,
+      "step": 1954
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.853844165802002,
+      "learning_rate": 0.00024630364673778216,
+      "loss": 0.734,
+      "step": 1955
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.7086217403411865,
+      "learning_rate": 0.0002462664351277598,
+      "loss": 0.9751,
+      "step": 1956
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.8849620819091797,
+      "learning_rate": 0.0002462292235177375,
+      "loss": 0.6382,
+      "step": 1957
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1819937229156494,
+      "learning_rate": 0.00024619201190771516,
+      "loss": 0.5407,
+      "step": 1958
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.5230162143707275,
+      "learning_rate": 0.00024615480029769286,
+      "loss": 1.6646,
+      "step": 1959
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.293956995010376,
+      "learning_rate": 0.00024611758868767057,
+      "loss": 0.5963,
+      "step": 1960
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.532461643218994,
+      "learning_rate": 0.0002460803770776482,
+      "loss": 0.7947,
+      "step": 1961
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1214241981506348,
+      "learning_rate": 0.00024604316546762586,
+      "loss": 0.6076,
+      "step": 1962
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.7289674282073975,
+      "learning_rate": 0.00024600595385760357,
+      "loss": 0.9689,
+      "step": 1963
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5347867012023926,
+      "learning_rate": 0.0002459687422475812,
+      "loss": 0.3691,
+      "step": 1964
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.878653645515442,
+      "learning_rate": 0.0002459315306375589,
+      "loss": 0.3384,
+      "step": 1965
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.7828047275543213,
+      "learning_rate": 0.00024589431902753657,
+      "loss": 0.6173,
+      "step": 1966
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.254936456680298,
+      "learning_rate": 0.00024585710741751427,
+      "loss": 0.8577,
+      "step": 1967
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9848906993865967,
+      "learning_rate": 0.0002458198958074919,
+      "loss": 0.2193,
+      "step": 1968
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.952481985092163,
+      "learning_rate": 0.00024578268419746957,
+      "loss": 1.0369,
+      "step": 1969
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.548969030380249,
+      "learning_rate": 0.00024574547258744727,
+      "loss": 0.749,
+      "step": 1970
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2937052249908447,
+      "learning_rate": 0.0002457082609774249,
+      "loss": 0.4634,
+      "step": 1971
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3762097358703613,
+      "learning_rate": 0.0002456710493674026,
+      "loss": 0.2366,
+      "step": 1972
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 5.370176792144775,
+      "learning_rate": 0.00024563383775738027,
+      "loss": 1.8996,
+      "step": 1973
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.1844329833984375,
+      "learning_rate": 0.0002455966261473579,
+      "loss": 0.3431,
+      "step": 1974
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5123300552368164,
+      "learning_rate": 0.0002455594145373356,
+      "loss": 0.2331,
+      "step": 1975
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.754368782043457,
+      "learning_rate": 0.0002455222029273133,
+      "loss": 1.245,
+      "step": 1976
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.0486223697662354,
+      "learning_rate": 0.000245484991317291,
+      "loss": 0.6925,
+      "step": 1977
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.9800477027893066,
+      "learning_rate": 0.0002454477797072687,
+      "loss": 1.1662,
+      "step": 1978
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9747953414916992,
+      "learning_rate": 0.0002454105680972463,
+      "loss": 0.3735,
+      "step": 1979
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3765416145324707,
+      "learning_rate": 0.000245373356487224,
+      "loss": 0.7723,
+      "step": 1980
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.1413557529449463,
+      "learning_rate": 0.0002453361448772017,
+      "loss": 0.666,
+      "step": 1981
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1634583473205566,
+      "learning_rate": 0.0002452989332671793,
+      "loss": 0.3816,
+      "step": 1982
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.8471031188964844,
+      "learning_rate": 0.00024526172165715703,
+      "loss": 0.6517,
+      "step": 1983
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.5720808506011963,
+      "learning_rate": 0.0002452245100471347,
+      "loss": 0.4428,
+      "step": 1984
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.9162683486938477,
+      "learning_rate": 0.0002451872984371124,
+      "loss": 0.4646,
+      "step": 1985
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.0276424884796143,
+      "learning_rate": 0.00024515008682709003,
+      "loss": 0.4527,
+      "step": 1986
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.1987595558166504,
+      "learning_rate": 0.0002451128752170677,
+      "loss": 0.631,
+      "step": 1987
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.891221284866333,
+      "learning_rate": 0.0002450756636070454,
+      "loss": 0.4301,
+      "step": 1988
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.999260425567627,
+      "learning_rate": 0.00024503845199702303,
+      "loss": 0.7106,
+      "step": 1989
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.418884754180908,
+      "learning_rate": 0.00024500124038700073,
+      "loss": 0.8794,
+      "step": 1990
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.5227293968200684,
+      "learning_rate": 0.0002449640287769784,
+      "loss": 0.6786,
+      "step": 1991
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.8759443759918213,
+      "learning_rate": 0.0002449268171669561,
+      "loss": 0.9838,
+      "step": 1992
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 5.6880059242248535,
+      "learning_rate": 0.00024488960555693373,
+      "loss": 1.1084,
+      "step": 1993
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 5.6343183517456055,
+      "learning_rate": 0.00024485239394691143,
+      "loss": 0.8567,
+      "step": 1994
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4194846153259277,
+      "learning_rate": 0.0002448151823368891,
+      "loss": 0.2741,
+      "step": 1995
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.982990264892578,
+      "learning_rate": 0.0002447779707268668,
+      "loss": 0.3463,
+      "step": 1996
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.392058849334717,
+      "learning_rate": 0.00024474075911684444,
+      "loss": 0.6745,
+      "step": 1997
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.701278209686279,
+      "learning_rate": 0.0002447035475068221,
+      "loss": 1.1925,
+      "step": 1998
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.1057417392730713,
+      "learning_rate": 0.0002446663358967998,
+      "loss": 0.225,
+      "step": 1999
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4700098037719727,
+      "learning_rate": 0.00024462912428677744,
+      "loss": 0.2833,
+      "step": 2000
+    },
+    {
+      "epoch": 0.7,
+      "eval_loss": 0.8290925025939941,
+      "eval_runtime": 50.9678,
+      "eval_samples_per_second": 42.537,
+      "eval_steps_per_second": 10.634,
+      "eval_wer": 0.6420169520844144,
+      "step": 2000
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 7.508627414703369,
+      "learning_rate": 0.00024459191267675514,
+      "loss": 1.6429,
+      "step": 2001
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.5678176879882812,
+      "learning_rate": 0.0002445547010667328,
+      "loss": 1.0799,
+      "step": 2002
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7853446006774902,
+      "learning_rate": 0.0002445174894567105,
+      "loss": 0.7929,
+      "step": 2003
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.158714532852173,
+      "learning_rate": 0.00024448027784668814,
+      "loss": 0.6522,
+      "step": 2004
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.3752233982086182,
+      "learning_rate": 0.0002444430662366658,
+      "loss": 0.407,
+      "step": 2005
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.856546401977539,
+      "learning_rate": 0.0002444058546266435,
+      "loss": 0.649,
+      "step": 2006
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.812596321105957,
+      "learning_rate": 0.0002443686430166212,
+      "loss": 0.4625,
+      "step": 2007
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.5984840393066406,
+      "learning_rate": 0.00024433143140659884,
+      "loss": 0.3318,
+      "step": 2008
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.0432894229888916,
+      "learning_rate": 0.00024429421979657654,
+      "loss": 0.3098,
+      "step": 2009
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.646289587020874,
+      "learning_rate": 0.0002442570081865542,
+      "loss": 0.4007,
+      "step": 2010
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.5235435962677,
+      "learning_rate": 0.00024421979657653184,
+      "loss": 0.3968,
+      "step": 2011
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.276270866394043,
+      "learning_rate": 0.00024418258496650954,
+      "loss": 1.1839,
+      "step": 2012
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.5912859439849854,
+      "learning_rate": 0.0002441453733564872,
+      "loss": 0.5202,
+      "step": 2013
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.616440296173096,
+      "learning_rate": 0.00024410816174646487,
+      "loss": 1.8386,
+      "step": 2014
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.167287349700928,
+      "learning_rate": 0.00024407095013644255,
+      "loss": 0.7833,
+      "step": 2015
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.9030117988586426,
+      "learning_rate": 0.00024403373852642022,
+      "loss": 0.9212,
+      "step": 2016
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 5.886682987213135,
+      "learning_rate": 0.0002439965269163979,
+      "loss": 0.3937,
+      "step": 2017
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.818249464035034,
+      "learning_rate": 0.00024395931530637555,
+      "loss": 0.305,
+      "step": 2018
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.023711204528809,
+      "learning_rate": 0.00024392210369635325,
+      "loss": 0.5857,
+      "step": 2019
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.1700754165649414,
+      "learning_rate": 0.00024388489208633092,
+      "loss": 0.5954,
+      "step": 2020
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.295510768890381,
+      "learning_rate": 0.00024384768047630857,
+      "loss": 0.742,
+      "step": 2021
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.545529365539551,
+      "learning_rate": 0.00024381046886628628,
+      "loss": 0.9577,
+      "step": 2022
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.260254383087158,
+      "learning_rate": 0.00024377325725626392,
+      "loss": 0.2636,
+      "step": 2023
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.4573659896850586,
+      "learning_rate": 0.0002437360456462416,
+      "loss": 0.313,
+      "step": 2024
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.888119220733643,
+      "learning_rate": 0.0002436988340362193,
+      "loss": 0.3761,
+      "step": 2025
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.143555164337158,
+      "learning_rate": 0.00024366162242619695,
+      "loss": 1.1434,
+      "step": 2026
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.872159719467163,
+      "learning_rate": 0.00024362441081617463,
+      "loss": 0.8784,
+      "step": 2027
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 15.59583568572998,
+      "learning_rate": 0.00024358719920615228,
+      "loss": 6.3413,
+      "step": 2028
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.5535459518432617,
+      "learning_rate": 0.00024354998759612998,
+      "loss": 0.6744,
+      "step": 2029
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.3724849224090576,
+      "learning_rate": 0.00024351277598610766,
+      "loss": 0.6494,
+      "step": 2030
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 12.914846420288086,
+      "learning_rate": 0.0002434755643760853,
+      "loss": 4.2932,
+      "step": 2031
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.6765973567962646,
+      "learning_rate": 0.000243438352766063,
+      "loss": 0.3652,
+      "step": 2032
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.295680284500122,
+      "learning_rate": 0.00024340114115604068,
+      "loss": 0.6002,
+      "step": 2033
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.5459048748016357,
+      "learning_rate": 0.00024336392954601833,
+      "loss": 0.4525,
+      "step": 2034
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.9346370697021484,
+      "learning_rate": 0.000243326717935996,
+      "loss": 0.4579,
+      "step": 2035
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.241410732269287,
+      "learning_rate": 0.00024328950632597368,
+      "loss": 0.7072,
+      "step": 2036
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.559287667274475,
+      "learning_rate": 0.00024325229471595136,
+      "loss": 0.2728,
+      "step": 2037
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.5284101963043213,
+      "learning_rate": 0.00024321508310592903,
+      "loss": 0.5484,
+      "step": 2038
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.7311956882476807,
+      "learning_rate": 0.00024317787149590668,
+      "loss": 0.8191,
+      "step": 2039
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 12.553634643554688,
+      "learning_rate": 0.00024314065988588439,
+      "loss": 1.2444,
+      "step": 2040
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 6.129791259765625,
+      "learning_rate": 0.00024310344827586203,
+      "loss": 0.8698,
+      "step": 2041
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.7819488048553467,
+      "learning_rate": 0.0002430662366658397,
+      "loss": 0.4961,
+      "step": 2042
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.355811357498169,
+      "learning_rate": 0.0002430290250558174,
+      "loss": 0.5451,
+      "step": 2043
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.024493932723999,
+      "learning_rate": 0.00024299181344579506,
+      "loss": 0.7597,
+      "step": 2044
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.637416124343872,
+      "learning_rate": 0.00024295460183577274,
+      "loss": 0.7338,
+      "step": 2045
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.891045570373535,
+      "learning_rate": 0.0002429173902257504,
+      "loss": 0.3468,
+      "step": 2046
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 5.023557186126709,
+      "learning_rate": 0.0002428801786157281,
+      "loss": 0.424,
+      "step": 2047
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.320523262023926,
+      "learning_rate": 0.00024284296700570577,
+      "loss": 0.3529,
+      "step": 2048
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 7.243979454040527,
+      "learning_rate": 0.00024280575539568341,
+      "loss": 2.7054,
+      "step": 2049
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.9604123830795288,
+      "learning_rate": 0.00024276854378566112,
+      "loss": 0.2114,
+      "step": 2050
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.573541164398193,
+      "learning_rate": 0.0002427313321756388,
+      "loss": 1.3774,
+      "step": 2051
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.40055251121521,
+      "learning_rate": 0.00024269412056561644,
+      "loss": 0.8643,
+      "step": 2052
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.0245659351348877,
+      "learning_rate": 0.00024265690895559414,
+      "loss": 0.4089,
+      "step": 2053
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.8446860313415527,
+      "learning_rate": 0.0002426196973455718,
+      "loss": 0.8581,
+      "step": 2054
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.76045823097229,
+      "learning_rate": 0.00024258248573554947,
+      "loss": 0.4217,
+      "step": 2055
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.384861707687378,
+      "learning_rate": 0.00024254527412552714,
+      "loss": 0.5413,
+      "step": 2056
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.9222930669784546,
+      "learning_rate": 0.00024250806251550482,
+      "loss": 0.3125,
+      "step": 2057
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.540152072906494,
+      "learning_rate": 0.0002424708509054825,
+      "loss": 1.8312,
+      "step": 2058
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.780682325363159,
+      "learning_rate": 0.00024243363929546014,
+      "loss": 0.6677,
+      "step": 2059
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.9040141105651855,
+      "learning_rate": 0.00024239642768543782,
+      "loss": 0.7344,
+      "step": 2060
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.254075050354004,
+      "learning_rate": 0.00024235921607541552,
+      "loss": 0.9292,
+      "step": 2061
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.1678977012634277,
+      "learning_rate": 0.00024232200446539317,
+      "loss": 1.1215,
+      "step": 2062
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.6753976345062256,
+      "learning_rate": 0.00024228479285537085,
+      "loss": 0.4991,
+      "step": 2063
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.377896308898926,
+      "learning_rate": 0.00024224758124534855,
+      "loss": 0.5978,
+      "step": 2064
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.7869486808776855,
+      "learning_rate": 0.0002422103696353262,
+      "loss": 0.8578,
+      "step": 2065
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.29856014251709,
+      "learning_rate": 0.00024217315802530388,
+      "loss": 1.6217,
+      "step": 2066
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.67818021774292,
+      "learning_rate": 0.00024213594641528152,
+      "loss": 0.5943,
+      "step": 2067
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.469092845916748,
+      "learning_rate": 0.00024209873480525923,
+      "loss": 1.1634,
+      "step": 2068
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.7627439498901367,
+      "learning_rate": 0.0002420615231952369,
+      "loss": 0.591,
+      "step": 2069
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3488175868988037,
+      "learning_rate": 0.00024202431158521455,
+      "loss": 0.2379,
+      "step": 2070
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.005652904510498,
+      "learning_rate": 0.00024198709997519225,
+      "loss": 0.2445,
+      "step": 2071
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.345755100250244,
+      "learning_rate": 0.0002419498883651699,
+      "loss": 0.7675,
+      "step": 2072
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 7.79234504699707,
+      "learning_rate": 0.00024191267675514758,
+      "loss": 0.6093,
+      "step": 2073
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 6.379326820373535,
+      "learning_rate": 0.00024187546514512528,
+      "loss": 1.4354,
+      "step": 2074
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.6521170139312744,
+      "learning_rate": 0.00024183825353510293,
+      "loss": 0.7191,
+      "step": 2075
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.690326452255249,
+      "learning_rate": 0.0002418010419250806,
+      "loss": 1.3118,
+      "step": 2076
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.035961627960205,
+      "learning_rate": 0.00024176383031505828,
+      "loss": 1.0016,
+      "step": 2077
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.05783748626709,
+      "learning_rate": 0.00024172661870503596,
+      "loss": 0.6541,
+      "step": 2078
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.9695038795471191,
+      "learning_rate": 0.00024168940709501363,
+      "loss": 0.4606,
+      "step": 2079
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.9987822771072388,
+      "learning_rate": 0.00024165219548499128,
+      "loss": 0.6701,
+      "step": 2080
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.004821538925171,
+      "learning_rate": 0.00024161498387496896,
+      "loss": 0.4184,
+      "step": 2081
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6122078895568848,
+      "learning_rate": 0.00024157777226494666,
+      "loss": 0.3568,
+      "step": 2082
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.2118942737579346,
+      "learning_rate": 0.0002415405606549243,
+      "loss": 0.4723,
+      "step": 2083
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.2981133460998535,
+      "learning_rate": 0.00024150334904490199,
+      "loss": 0.5278,
+      "step": 2084
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.0347321033477783,
+      "learning_rate": 0.00024146613743487963,
+      "loss": 0.8326,
+      "step": 2085
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.122243881225586,
+      "learning_rate": 0.00024142892582485734,
+      "loss": 0.6792,
+      "step": 2086
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.2204928398132324,
+      "learning_rate": 0.000241391714214835,
+      "loss": 0.6001,
+      "step": 2087
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.347808837890625,
+      "learning_rate": 0.00024135450260481266,
+      "loss": 1.9925,
+      "step": 2088
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3485236167907715,
+      "learning_rate": 0.00024131729099479036,
+      "loss": 0.3445,
+      "step": 2089
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.8277785778045654,
+      "learning_rate": 0.000241280079384768,
+      "loss": 0.5412,
+      "step": 2090
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.600690484046936,
+      "learning_rate": 0.0002412428677747457,
+      "loss": 0.1389,
+      "step": 2091
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.351870059967041,
+      "learning_rate": 0.0002412056561647234,
+      "loss": 0.1752,
+      "step": 2092
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.944504499435425,
+      "learning_rate": 0.00024116844455470104,
+      "loss": 0.5421,
+      "step": 2093
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.7904212474823,
+      "learning_rate": 0.00024113123294467872,
+      "loss": 0.481,
+      "step": 2094
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.337879180908203,
+      "learning_rate": 0.00024109402133465642,
+      "loss": 1.1637,
+      "step": 2095
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.739457607269287,
+      "learning_rate": 0.00024105680972463407,
+      "loss": 0.6533,
+      "step": 2096
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.9053199291229248,
+      "learning_rate": 0.00024101959811461174,
+      "loss": 0.326,
+      "step": 2097
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 10.310988426208496,
+      "learning_rate": 0.0002409823865045894,
+      "loss": 1.6222,
+      "step": 2098
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2947075366973877,
+      "learning_rate": 0.0002409451748945671,
+      "loss": 0.2548,
+      "step": 2099
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.7225122451782227,
+      "learning_rate": 0.00024090796328454477,
+      "loss": 0.2533,
+      "step": 2100
+    },
+    {
+      "epoch": 0.74,
+      "eval_loss": 0.7072895169258118,
+      "eval_runtime": 50.5869,
+      "eval_samples_per_second": 42.857,
+      "eval_steps_per_second": 10.714,
+      "eval_wer": 0.5960906417574814,
+      "step": 2100
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.087221622467041,
+      "learning_rate": 0.00024087075167452242,
+      "loss": 0.9829,
+      "step": 2101
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 5.295423984527588,
+      "learning_rate": 0.0002408335400645001,
+      "loss": 1.442,
+      "step": 2102
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.7258617877960205,
+      "learning_rate": 0.00024079632845447777,
+      "loss": 0.4635,
+      "step": 2103
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.04913067817688,
+      "learning_rate": 0.00024075911684445545,
+      "loss": 0.462,
+      "step": 2104
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.6161487102508545,
+      "learning_rate": 0.00024072190523443312,
+      "loss": 0.6827,
+      "step": 2105
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.9730608463287354,
+      "learning_rate": 0.00024068469362441077,
+      "loss": 0.9988,
+      "step": 2106
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.8048250675201416,
+      "learning_rate": 0.00024064748201438847,
+      "loss": 1.0969,
+      "step": 2107
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.332301616668701,
+      "learning_rate": 0.00024061027040436615,
+      "loss": 0.4455,
+      "step": 2108
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.4177818298339844,
+      "learning_rate": 0.0002405730587943438,
+      "loss": 2.1572,
+      "step": 2109
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 5.298447132110596,
+      "learning_rate": 0.0002405358471843215,
+      "loss": 0.3914,
+      "step": 2110
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.914675235748291,
+      "learning_rate": 0.00024049863557429915,
+      "loss": 1.0013,
+      "step": 2111
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.703636646270752,
+      "learning_rate": 0.00024046142396427683,
+      "loss": 0.4293,
+      "step": 2112
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.7955198287963867,
+      "learning_rate": 0.00024042421235425453,
+      "loss": 0.5958,
+      "step": 2113
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.913386583328247,
+      "learning_rate": 0.00024038700074423218,
+      "loss": 0.7706,
+      "step": 2114
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.9934210777282715,
+      "learning_rate": 0.00024034978913420985,
+      "loss": 0.4863,
+      "step": 2115
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.5534229278564453,
+      "learning_rate": 0.0002403125775241875,
+      "loss": 0.2289,
+      "step": 2116
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.206246852874756,
+      "learning_rate": 0.0002402753659141652,
+      "loss": 1.0357,
+      "step": 2117
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.4681451320648193,
+      "learning_rate": 0.00024023815430414288,
+      "loss": 0.9145,
+      "step": 2118
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.2907421588897705,
+      "learning_rate": 0.00024020094269412053,
+      "loss": 0.6565,
+      "step": 2119
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.354326009750366,
+      "learning_rate": 0.00024016373108409823,
+      "loss": 0.5887,
+      "step": 2120
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.702387571334839,
+      "learning_rate": 0.0002401265194740759,
+      "loss": 0.4698,
+      "step": 2121
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.9041430950164795,
+      "learning_rate": 0.00024008930786405356,
+      "loss": 0.5025,
+      "step": 2122
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.8005261421203613,
+      "learning_rate": 0.00024005209625403123,
+      "loss": 0.4798,
+      "step": 2123
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.7690351009368896,
+      "learning_rate": 0.0002400148846440089,
+      "loss": 0.2449,
+      "step": 2124
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 5.324347496032715,
+      "learning_rate": 0.00023997767303398658,
+      "loss": 1.9174,
+      "step": 2125
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.6067378520965576,
+      "learning_rate": 0.00023994046142396426,
+      "loss": 0.6974,
+      "step": 2126
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2820823192596436,
+      "learning_rate": 0.0002399032498139419,
+      "loss": 0.3203,
+      "step": 2127
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.06552255153656,
+      "learning_rate": 0.0002398660382039196,
+      "loss": 0.2081,
+      "step": 2128
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3411970138549805,
+      "learning_rate": 0.00023982882659389726,
+      "loss": 0.6395,
+      "step": 2129
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.0925540924072266,
+      "learning_rate": 0.00023979161498387494,
+      "loss": 0.6678,
+      "step": 2130
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3923542499542236,
+      "learning_rate": 0.00023975440337385264,
+      "loss": 0.5511,
+      "step": 2131
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.139909267425537,
+      "learning_rate": 0.0002397171917638303,
+      "loss": 0.4857,
+      "step": 2132
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 5.286214828491211,
+      "learning_rate": 0.00023967998015380796,
+      "loss": 0.7586,
+      "step": 2133
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 6.681207656860352,
+      "learning_rate": 0.0002396427685437856,
+      "loss": 0.7079,
+      "step": 2134
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.236323595046997,
+      "learning_rate": 0.00023960555693376332,
+      "loss": 0.6949,
+      "step": 2135
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.150146722793579,
+      "learning_rate": 0.000239568345323741,
+      "loss": 0.7396,
+      "step": 2136
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.668715715408325,
+      "learning_rate": 0.00023953113371371864,
+      "loss": 0.7919,
+      "step": 2137
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.448146104812622,
+      "learning_rate": 0.00023949392210369634,
+      "loss": 0.4682,
+      "step": 2138
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.377995014190674,
+      "learning_rate": 0.00023945671049367402,
+      "loss": 0.6805,
+      "step": 2139
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.0972490310668945,
+      "learning_rate": 0.00023941949888365167,
+      "loss": 0.388,
+      "step": 2140
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.356851100921631,
+      "learning_rate": 0.00023938228727362937,
+      "loss": 0.9222,
+      "step": 2141
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.6163222789764404,
+      "learning_rate": 0.00023934507566360702,
+      "loss": 0.3178,
+      "step": 2142
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.244340181350708,
+      "learning_rate": 0.0002393078640535847,
+      "loss": 0.414,
+      "step": 2143
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.176497220993042,
+      "learning_rate": 0.00023927065244356237,
+      "loss": 0.3438,
+      "step": 2144
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.1383144855499268,
+      "learning_rate": 0.00023923344083354005,
+      "loss": 0.3254,
+      "step": 2145
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.8384389877319336,
+      "learning_rate": 0.00023919622922351772,
+      "loss": 0.3931,
+      "step": 2146
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.945198059082031,
+      "learning_rate": 0.00023915901761349537,
+      "loss": 0.6616,
+      "step": 2147
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2696187496185303,
+      "learning_rate": 0.00023912180600347305,
+      "loss": 0.1269,
+      "step": 2148
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.868678092956543,
+      "learning_rate": 0.00023908459439345075,
+      "loss": 1.2393,
+      "step": 2149
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.27547025680542,
+      "learning_rate": 0.0002390473827834284,
+      "loss": 0.3182,
+      "step": 2150
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.301406383514404,
+      "learning_rate": 0.00023901017117340607,
+      "loss": 1.3776,
+      "step": 2151
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.5143606662750244,
+      "learning_rate": 0.00023897295956338378,
+      "loss": 0.5273,
+      "step": 2152
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.4446358680725098,
+      "learning_rate": 0.00023893574795336143,
+      "loss": 0.7657,
+      "step": 2153
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.8754398822784424,
+      "learning_rate": 0.0002388985363433391,
+      "loss": 0.8753,
+      "step": 2154
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.9459950923919678,
+      "learning_rate": 0.00023886132473331675,
+      "loss": 1.1413,
+      "step": 2155
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.867654800415039,
+      "learning_rate": 0.00023882411312329445,
+      "loss": 0.612,
+      "step": 2156
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5319912433624268,
+      "learning_rate": 0.00023878690151327213,
+      "loss": 0.633,
+      "step": 2157
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.040459632873535,
+      "learning_rate": 0.00023874968990324978,
+      "loss": 0.3897,
+      "step": 2158
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.5026168823242188,
+      "learning_rate": 0.00023871247829322748,
+      "loss": 0.6734,
+      "step": 2159
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.2834792137145996,
+      "learning_rate": 0.00023867526668320513,
+      "loss": 0.4584,
+      "step": 2160
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.716819167137146,
+      "learning_rate": 0.0002386380550731828,
+      "loss": 0.4489,
+      "step": 2161
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.5152437686920166,
+      "learning_rate": 0.0002386008434631605,
+      "loss": 1.6475,
+      "step": 2162
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.483433961868286,
+      "learning_rate": 0.00023856363185313816,
+      "loss": 0.2928,
+      "step": 2163
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.2262301445007324,
+      "learning_rate": 0.00023852642024311583,
+      "loss": 0.5461,
+      "step": 2164
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.866180896759033,
+      "learning_rate": 0.0002384892086330935,
+      "loss": 0.4271,
+      "step": 2165
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.7532100677490234,
+      "learning_rate": 0.00023845199702307118,
+      "loss": 0.3409,
+      "step": 2166
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.9978840351104736,
+      "learning_rate": 0.00023841478541304886,
+      "loss": 1.1862,
+      "step": 2167
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.0184621810913086,
+      "learning_rate": 0.0002383775738030265,
+      "loss": 0.1015,
+      "step": 2168
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.473518133163452,
+      "learning_rate": 0.00023834036219300418,
+      "loss": 0.4473,
+      "step": 2169
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.6468117237091064,
+      "learning_rate": 0.0002383031505829819,
+      "loss": 0.49,
+      "step": 2170
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.993438720703125,
+      "learning_rate": 0.00023826593897295954,
+      "loss": 0.6108,
+      "step": 2171
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.0172436237335205,
+      "learning_rate": 0.0002382287273629372,
+      "loss": 0.4763,
+      "step": 2172
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 5.393608570098877,
+      "learning_rate": 0.0002381915157529149,
+      "loss": 0.4268,
+      "step": 2173
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.117353439331055,
+      "learning_rate": 0.00023815430414289256,
+      "loss": 0.7324,
+      "step": 2174
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.40974235534668,
+      "learning_rate": 0.00023811709253287024,
+      "loss": 0.3379,
+      "step": 2175
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 5.692419052124023,
+      "learning_rate": 0.0002380798809228479,
+      "loss": 1.1812,
+      "step": 2176
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 5.502355098724365,
+      "learning_rate": 0.0002380426693128256,
+      "loss": 1.0161,
+      "step": 2177
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.216310501098633,
+      "learning_rate": 0.00023800545770280324,
+      "loss": 0.488,
+      "step": 2178
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.1364212036132812,
+      "learning_rate": 0.00023796824609278091,
+      "loss": 0.528,
+      "step": 2179
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4948229789733887,
+      "learning_rate": 0.00023793103448275862,
+      "loss": 0.6491,
+      "step": 2180
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.0776448249816895,
+      "learning_rate": 0.00023789382287273627,
+      "loss": 0.1683,
+      "step": 2181
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.0587196350097656,
+      "learning_rate": 0.00023785661126271394,
+      "loss": 0.4055,
+      "step": 2182
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.165605306625366,
+      "learning_rate": 0.00023781939965269164,
+      "loss": 0.7523,
+      "step": 2183
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3306822776794434,
+      "learning_rate": 0.0002377821880426693,
+      "loss": 0.2969,
+      "step": 2184
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.706977128982544,
+      "learning_rate": 0.00023774497643264697,
+      "loss": 0.7947,
+      "step": 2185
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.0041239261627197,
+      "learning_rate": 0.00023770776482262462,
+      "loss": 0.6185,
+      "step": 2186
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.9066877365112305,
+      "learning_rate": 0.00023767055321260232,
+      "loss": 0.3012,
+      "step": 2187
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.505563974380493,
+      "learning_rate": 0.00023763334160258,
+      "loss": 0.321,
+      "step": 2188
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.046771049499512,
+      "learning_rate": 0.00023759612999255765,
+      "loss": 1.1728,
+      "step": 2189
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.1281588077545166,
+      "learning_rate": 0.00023755891838253532,
+      "loss": 0.4046,
+      "step": 2190
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.7986767292022705,
+      "learning_rate": 0.000237521706772513,
+      "loss": 0.8822,
+      "step": 2191
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.6561527252197266,
+      "learning_rate": 0.00023748449516249067,
+      "loss": 0.6199,
+      "step": 2192
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.620680332183838,
+      "learning_rate": 0.00023744728355246835,
+      "loss": 0.9583,
+      "step": 2193
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.1020591259002686,
+      "learning_rate": 0.00023741007194244602,
+      "loss": 0.464,
+      "step": 2194
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.0864760875701904,
+      "learning_rate": 0.0002373728603324237,
+      "loss": 0.5002,
+      "step": 2195
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.774919033050537,
+      "learning_rate": 0.00023733564872240138,
+      "loss": 0.7486,
+      "step": 2196
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.6889686584472656,
+      "learning_rate": 0.00023729843711237902,
+      "loss": 0.5443,
+      "step": 2197
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.7951676845550537,
+      "learning_rate": 0.00023726122550235673,
+      "loss": 1.2228,
+      "step": 2198
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.316519260406494,
+      "learning_rate": 0.00023722401389233438,
+      "loss": 1.6501,
+      "step": 2199
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2464311122894287,
+      "learning_rate": 0.00023718680228231205,
+      "loss": 0.3804,
+      "step": 2200
+    },
+    {
+      "epoch": 0.77,
+      "eval_loss": 0.6850249171257019,
+      "eval_runtime": 50.5746,
+      "eval_samples_per_second": 42.867,
+      "eval_steps_per_second": 10.717,
+      "eval_wer": 0.5974744853831517,
+      "step": 2200
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 8.690844535827637,
+      "learning_rate": 0.00023714959067228976,
+      "loss": 2.5763,
+      "step": 2201
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.9838602542877197,
+      "learning_rate": 0.0002371123790622674,
+      "loss": 0.7831,
+      "step": 2202
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.0435171127319336,
+      "learning_rate": 0.00023707516745224508,
+      "loss": 0.5686,
+      "step": 2203
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.6351277828216553,
+      "learning_rate": 0.00023703795584222273,
+      "loss": 0.9036,
+      "step": 2204
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.1336121559143066,
+      "learning_rate": 0.00023700074423220043,
+      "loss": 0.6445,
+      "step": 2205
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 9.804746627807617,
+      "learning_rate": 0.0002369635326221781,
+      "loss": 2.4538,
+      "step": 2206
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.9941658973693848,
+      "learning_rate": 0.00023692632101215576,
+      "loss": 0.5441,
+      "step": 2207
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.662944793701172,
+      "learning_rate": 0.00023688910940213346,
+      "loss": 0.2834,
+      "step": 2208
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.271035671234131,
+      "learning_rate": 0.00023685189779211113,
+      "loss": 0.373,
+      "step": 2209
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.727461099624634,
+      "learning_rate": 0.00023681468618208878,
+      "loss": 0.4508,
+      "step": 2210
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.807447910308838,
+      "learning_rate": 0.00023677747457206646,
+      "loss": 0.5131,
+      "step": 2211
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.0257365703582764,
+      "learning_rate": 0.00023674026296204413,
+      "loss": 0.4543,
+      "step": 2212
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.6574392318725586,
+      "learning_rate": 0.0002367030513520218,
+      "loss": 0.8255,
+      "step": 2213
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.633670330047607,
+      "learning_rate": 0.00023666583974199949,
+      "loss": 0.7957,
+      "step": 2214
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.175776720046997,
+      "learning_rate": 0.00023662862813197716,
+      "loss": 0.7773,
+      "step": 2215
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.383112907409668,
+      "learning_rate": 0.00023659141652195484,
+      "loss": 0.4066,
+      "step": 2216
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.62146258354187,
+      "learning_rate": 0.0002365542049119325,
+      "loss": 0.8173,
+      "step": 2217
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 48.06704330444336,
+      "learning_rate": 0.00023651699330191016,
+      "loss": 0.7836,
+      "step": 2218
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.3420729637145996,
+      "learning_rate": 0.00023647978169188787,
+      "loss": 0.2653,
+      "step": 2219
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.6878349781036377,
+      "learning_rate": 0.00023644257008186551,
+      "loss": 1.2167,
+      "step": 2220
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.163689613342285,
+      "learning_rate": 0.0002364053584718432,
+      "loss": 0.2967,
+      "step": 2221
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.2857308387756348,
+      "learning_rate": 0.0002363681468618209,
+      "loss": 0.2476,
+      "step": 2222
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.968120574951172,
+      "learning_rate": 0.00023633093525179854,
+      "loss": 0.4314,
+      "step": 2223
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.3224382400512695,
+      "learning_rate": 0.00023629372364177622,
+      "loss": 0.6707,
+      "step": 2224
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.102696895599365,
+      "learning_rate": 0.00023625651203175387,
+      "loss": 0.7693,
+      "step": 2225
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.199782371520996,
+      "learning_rate": 0.00023621930042173157,
+      "loss": 1.5092,
+      "step": 2226
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.694085121154785,
+      "learning_rate": 0.00023618208881170924,
+      "loss": 1.4029,
+      "step": 2227
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.0905003547668457,
+      "learning_rate": 0.0002361448772016869,
+      "loss": 0.5829,
+      "step": 2228
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.0144598484039307,
+      "learning_rate": 0.0002361076655916646,
+      "loss": 0.6812,
+      "step": 2229
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.288403034210205,
+      "learning_rate": 0.00023607045398164224,
+      "loss": 0.5396,
+      "step": 2230
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.9742906093597412,
+      "learning_rate": 0.00023603324237161992,
+      "loss": 0.3993,
+      "step": 2231
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.8184242248535156,
+      "learning_rate": 0.0002359960307615976,
+      "loss": 0.6374,
+      "step": 2232
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.8115568161010742,
+      "learning_rate": 0.00023595881915157527,
+      "loss": 0.6224,
+      "step": 2233
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.4084148406982422,
+      "learning_rate": 0.00023592160754155295,
+      "loss": 0.3404,
+      "step": 2234
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.8344879150390625,
+      "learning_rate": 0.0002358843959315306,
+      "loss": 0.7027,
+      "step": 2235
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.2461719512939453,
+      "learning_rate": 0.0002358471843215083,
+      "loss": 0.4054,
+      "step": 2236
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.190657377243042,
+      "learning_rate": 0.00023580997271148598,
+      "loss": 0.7986,
+      "step": 2237
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.7157618999481201,
+      "learning_rate": 0.00023577276110146362,
+      "loss": 0.3132,
+      "step": 2238
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.2127954959869385,
+      "learning_rate": 0.0002357355494914413,
+      "loss": 0.3935,
+      "step": 2239
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.468662738800049,
+      "learning_rate": 0.000235698337881419,
+      "loss": 0.6404,
+      "step": 2240
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6316206455230713,
+      "learning_rate": 0.00023566112627139665,
+      "loss": 0.5468,
+      "step": 2241
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.6631693840026855,
+      "learning_rate": 0.00023562391466137433,
+      "loss": 0.7712,
+      "step": 2242
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.116514205932617,
+      "learning_rate": 0.00023558670305135198,
+      "loss": 0.7397,
+      "step": 2243
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6557669639587402,
+      "learning_rate": 0.00023554949144132968,
+      "loss": 0.5441,
+      "step": 2244
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.5287699699401855,
+      "learning_rate": 0.00023551227983130735,
+      "loss": 0.4418,
+      "step": 2245
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.1134016513824463,
+      "learning_rate": 0.000235475068221285,
+      "loss": 0.4333,
+      "step": 2246
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.637173652648926,
+      "learning_rate": 0.0002354378566112627,
+      "loss": 0.5619,
+      "step": 2247
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.621034145355225,
+      "learning_rate": 0.00023540064500124035,
+      "loss": 0.8087,
+      "step": 2248
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.706472635269165,
+      "learning_rate": 0.00023536343339121803,
+      "loss": 0.3778,
+      "step": 2249
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 6.068023204803467,
+      "learning_rate": 0.00023532622178119573,
+      "loss": 0.5066,
+      "step": 2250
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.4167532920837402,
+      "learning_rate": 0.00023528901017117338,
+      "loss": 1.2843,
+      "step": 2251
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.9952954053878784,
+      "learning_rate": 0.00023525179856115106,
+      "loss": 0.7945,
+      "step": 2252
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.9978554248809814,
+      "learning_rate": 0.00023521458695112876,
+      "loss": 0.5449,
+      "step": 2253
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.489971876144409,
+      "learning_rate": 0.0002351773753411064,
+      "loss": 0.6356,
+      "step": 2254
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.167726993560791,
+      "learning_rate": 0.00023514016373108409,
+      "loss": 0.4688,
+      "step": 2255
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.427248477935791,
+      "learning_rate": 0.00023510295212106173,
+      "loss": 0.7861,
+      "step": 2256
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.5070106983184814,
+      "learning_rate": 0.00023506574051103944,
+      "loss": 0.9434,
+      "step": 2257
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.904707908630371,
+      "learning_rate": 0.0002350285289010171,
+      "loss": 0.7917,
+      "step": 2258
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.7793803215026855,
+      "learning_rate": 0.00023499131729099476,
+      "loss": 1.5236,
+      "step": 2259
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.193175792694092,
+      "learning_rate": 0.00023495410568097244,
+      "loss": 0.9601,
+      "step": 2260
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.6871867179870605,
+      "learning_rate": 0.0002349168940709501,
+      "loss": 0.3522,
+      "step": 2261
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.900092124938965,
+      "learning_rate": 0.0002348796824609278,
+      "loss": 0.9403,
+      "step": 2262
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.3442814350128174,
+      "learning_rate": 0.00023484247085090546,
+      "loss": 0.807,
+      "step": 2263
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.9604439735412598,
+      "learning_rate": 0.0002348052592408831,
+      "loss": 0.6192,
+      "step": 2264
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 5.023985862731934,
+      "learning_rate": 0.00023476804763086082,
+      "loss": 1.084,
+      "step": 2265
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.1374659538269043,
+      "learning_rate": 0.0002347308360208385,
+      "loss": 0.2996,
+      "step": 2266
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.652350425720215,
+      "learning_rate": 0.00023469362441081614,
+      "loss": 0.5887,
+      "step": 2267
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.5104193687438965,
+      "learning_rate": 0.00023465641280079384,
+      "loss": 0.6121,
+      "step": 2268
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.1145853996276855,
+      "learning_rate": 0.0002346192011907715,
+      "loss": 0.5084,
+      "step": 2269
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.850748538970947,
+      "learning_rate": 0.00023458198958074917,
+      "loss": 0.6602,
+      "step": 2270
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.4793652296066284,
+      "learning_rate": 0.00023454477797072687,
+      "loss": 0.1197,
+      "step": 2271
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.932457447052002,
+      "learning_rate": 0.00023450756636070452,
+      "loss": 0.1678,
+      "step": 2272
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.8412857055664062,
+      "learning_rate": 0.0002344703547506822,
+      "loss": 0.1298,
+      "step": 2273
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 5.570624828338623,
+      "learning_rate": 0.00023443314314065984,
+      "loss": 0.5715,
+      "step": 2274
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.606353759765625,
+      "learning_rate": 0.00023439593153063755,
+      "loss": 0.3448,
+      "step": 2275
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.9131336212158203,
+      "learning_rate": 0.00023435871992061522,
+      "loss": 1.2812,
+      "step": 2276
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.4033138751983643,
+      "learning_rate": 0.00023432150831059287,
+      "loss": 0.5682,
+      "step": 2277
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5214476585388184,
+      "learning_rate": 0.00023428429670057057,
+      "loss": 0.734,
+      "step": 2278
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.3294756412506104,
+      "learning_rate": 0.00023424708509054822,
+      "loss": 0.685,
+      "step": 2279
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.0864174365997314,
+      "learning_rate": 0.0002342098734805259,
+      "loss": 0.6529,
+      "step": 2280
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.4727028608322144,
+      "learning_rate": 0.00023417266187050357,
+      "loss": 0.3351,
+      "step": 2281
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.7328686714172363,
+      "learning_rate": 0.00023413545026048125,
+      "loss": 0.4686,
+      "step": 2282
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6917636394500732,
+      "learning_rate": 0.00023409823865045893,
+      "loss": 0.4922,
+      "step": 2283
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5129282474517822,
+      "learning_rate": 0.0002340610270404366,
+      "loss": 0.4395,
+      "step": 2284
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.758345603942871,
+      "learning_rate": 0.00023402381543041425,
+      "loss": 0.5023,
+      "step": 2285
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.338339328765869,
+      "learning_rate": 0.00023398660382039195,
+      "loss": 0.6605,
+      "step": 2286
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.811253786087036,
+      "learning_rate": 0.0002339493922103696,
+      "loss": 0.5628,
+      "step": 2287
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3005897998809814,
+      "learning_rate": 0.00023391218060034728,
+      "loss": 0.644,
+      "step": 2288
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.50140118598938,
+      "learning_rate": 0.00023387496899032498,
+      "loss": 1.2557,
+      "step": 2289
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.0913150310516357,
+      "learning_rate": 0.00023383775738030263,
+      "loss": 0.4355,
+      "step": 2290
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.851500988006592,
+      "learning_rate": 0.0002338005457702803,
+      "loss": 0.4956,
+      "step": 2291
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.441885471343994,
+      "learning_rate": 0.00023376333416025795,
+      "loss": 0.6377,
+      "step": 2292
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.17043399810791,
+      "learning_rate": 0.00023372612255023566,
+      "loss": 0.5092,
+      "step": 2293
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.0366196632385254,
+      "learning_rate": 0.00023368891094021333,
+      "loss": 0.1851,
+      "step": 2294
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.8057641983032227,
+      "learning_rate": 0.00023365169933019098,
+      "loss": 0.733,
+      "step": 2295
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 6.428284168243408,
+      "learning_rate": 0.00023361448772016868,
+      "loss": 0.7048,
+      "step": 2296
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.9241631031036377,
+      "learning_rate": 0.00023357727611014636,
+      "loss": 1.9665,
+      "step": 2297
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.794024467468262,
+      "learning_rate": 0.000233540064500124,
+      "loss": 0.953,
+      "step": 2298
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.998034954071045,
+      "learning_rate": 0.0002335028528901017,
+      "loss": 1.7438,
+      "step": 2299
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.720144271850586,
+      "learning_rate": 0.00023346564128007936,
+      "loss": 0.7473,
+      "step": 2300
+    },
+    {
+      "epoch": 0.81,
+      "eval_loss": 0.8185052275657654,
+      "eval_runtime": 50.5567,
+      "eval_samples_per_second": 42.883,
+      "eval_steps_per_second": 10.721,
+      "eval_wer": 0.7170039785504237,
+      "step": 2300
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.0130574703216553,
+      "learning_rate": 0.00023342842967005704,
+      "loss": 0.6867,
+      "step": 2301
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.1833200454711914,
+      "learning_rate": 0.0002333912180600347,
+      "loss": 0.9255,
+      "step": 2302
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.382704019546509,
+      "learning_rate": 0.0002333540064500124,
+      "loss": 0.7009,
+      "step": 2303
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 18.74046516418457,
+      "learning_rate": 0.00023331679483999006,
+      "loss": 5.78,
+      "step": 2304
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.770467758178711,
+      "learning_rate": 0.0002332795832299677,
+      "loss": 0.7705,
+      "step": 2305
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4621565341949463,
+      "learning_rate": 0.0002332423716199454,
+      "loss": 0.7438,
+      "step": 2306
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 8.334583282470703,
+      "learning_rate": 0.0002332051600099231,
+      "loss": 0.7081,
+      "step": 2307
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.128434658050537,
+      "learning_rate": 0.00023316794839990074,
+      "loss": 0.4139,
+      "step": 2308
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 5.296910285949707,
+      "learning_rate": 0.00023313073678987842,
+      "loss": 0.8335,
+      "step": 2309
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.945010185241699,
+      "learning_rate": 0.00023309352517985612,
+      "loss": 0.6668,
+      "step": 2310
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.273423433303833,
+      "learning_rate": 0.00023305631356983377,
+      "loss": 0.7138,
+      "step": 2311
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.95408034324646,
+      "learning_rate": 0.00023301910195981144,
+      "loss": 0.4044,
+      "step": 2312
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 7.172848224639893,
+      "learning_rate": 0.0002329818903497891,
+      "loss": 2.0813,
+      "step": 2313
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.5199854373931885,
+      "learning_rate": 0.0002329446787397668,
+      "loss": 0.4251,
+      "step": 2314
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.866971969604492,
+      "learning_rate": 0.00023290746712974447,
+      "loss": 0.5057,
+      "step": 2315
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.345081090927124,
+      "learning_rate": 0.00023287025551972212,
+      "loss": 0.4534,
+      "step": 2316
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.3792688846588135,
+      "learning_rate": 0.00023283304390969982,
+      "loss": 0.6358,
+      "step": 2317
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.9677326679229736,
+      "learning_rate": 0.00023279583229967747,
+      "loss": 1.3059,
+      "step": 2318
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.3454091548919678,
+      "learning_rate": 0.00023275862068965515,
+      "loss": 0.4036,
+      "step": 2319
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.9923884868621826,
+      "learning_rate": 0.00023272140907963285,
+      "loss": 1.2093,
+      "step": 2320
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.1582930088043213,
+      "learning_rate": 0.0002326841974696105,
+      "loss": 0.3846,
+      "step": 2321
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.0936267375946045,
+      "learning_rate": 0.00023264698585958817,
+      "loss": 0.5771,
+      "step": 2322
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.7140424251556396,
+      "learning_rate": 0.00023260977424956582,
+      "loss": 0.3977,
+      "step": 2323
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.4949116706848145,
+      "learning_rate": 0.00023257256263954353,
+      "loss": 1.5159,
+      "step": 2324
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.300839424133301,
+      "learning_rate": 0.0002325353510295212,
+      "loss": 0.3565,
+      "step": 2325
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.1757450103759766,
+      "learning_rate": 0.00023249813941949885,
+      "loss": 0.9017,
+      "step": 2326
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.34287166595459,
+      "learning_rate": 0.00023246092780947653,
+      "loss": 0.5174,
+      "step": 2327
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.0251080989837646,
+      "learning_rate": 0.00023242371619945423,
+      "loss": 0.678,
+      "step": 2328
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.359485149383545,
+      "learning_rate": 0.00023238650458943188,
+      "loss": 0.6204,
+      "step": 2329
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.325961112976074,
+      "learning_rate": 0.00023234929297940955,
+      "loss": 0.6806,
+      "step": 2330
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.272260904312134,
+      "learning_rate": 0.0002323120813693872,
+      "loss": 1.1038,
+      "step": 2331
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.266369581222534,
+      "learning_rate": 0.0002322748697593649,
+      "loss": 0.5986,
+      "step": 2332
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7468202114105225,
+      "learning_rate": 0.00023223765814934258,
+      "loss": 0.6712,
+      "step": 2333
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.0581917762756348,
+      "learning_rate": 0.00023220044653932023,
+      "loss": 0.4981,
+      "step": 2334
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.185770273208618,
+      "learning_rate": 0.00023216323492929793,
+      "loss": 0.8548,
+      "step": 2335
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.0419859886169434,
+      "learning_rate": 0.00023212602331927558,
+      "loss": 0.2562,
+      "step": 2336
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.8037047386169434,
+      "learning_rate": 0.00023208881170925326,
+      "loss": 0.625,
+      "step": 2337
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 6.072167873382568,
+      "learning_rate": 0.00023205160009923096,
+      "loss": 1.8957,
+      "step": 2338
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.633638858795166,
+      "learning_rate": 0.0002320143884892086,
+      "loss": 0.7044,
+      "step": 2339
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.252713680267334,
+      "learning_rate": 0.00023197717687918628,
+      "loss": 0.481,
+      "step": 2340
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7188823223114014,
+      "learning_rate": 0.000231939965269164,
+      "loss": 0.6524,
+      "step": 2341
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.263317108154297,
+      "learning_rate": 0.00023190275365914164,
+      "loss": 0.3887,
+      "step": 2342
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.679230213165283,
+      "learning_rate": 0.0002318655420491193,
+      "loss": 0.4308,
+      "step": 2343
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.2064208984375,
+      "learning_rate": 0.00023182833043909696,
+      "loss": 1.0281,
+      "step": 2344
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.935098171234131,
+      "learning_rate": 0.00023179111882907466,
+      "loss": 0.9202,
+      "step": 2345
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.840257167816162,
+      "learning_rate": 0.00023175390721905234,
+      "loss": 1.914,
+      "step": 2346
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.3795762062072754,
+      "learning_rate": 0.00023171669560903,
+      "loss": 0.6141,
+      "step": 2347
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.437031269073486,
+      "learning_rate": 0.00023167948399900766,
+      "loss": 0.9272,
+      "step": 2348
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.7707891464233398,
+      "learning_rate": 0.00023164227238898534,
+      "loss": 0.2347,
+      "step": 2349
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.9971076250076294,
+      "learning_rate": 0.00023160506077896301,
+      "loss": 0.1552,
+      "step": 2350
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.799489974975586,
+      "learning_rate": 0.0002315678491689407,
+      "loss": 1.1959,
+      "step": 2351
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.57328462600708,
+      "learning_rate": 0.00023153063755891834,
+      "loss": 0.448,
+      "step": 2352
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.8030247688293457,
+      "learning_rate": 0.00023149342594889604,
+      "loss": 1.108,
+      "step": 2353
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7969067096710205,
+      "learning_rate": 0.00023145621433887372,
+      "loss": 0.9344,
+      "step": 2354
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.8098201751708984,
+      "learning_rate": 0.00023141900272885137,
+      "loss": 0.2884,
+      "step": 2355
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.249875783920288,
+      "learning_rate": 0.00023138179111882907,
+      "loss": 0.6573,
+      "step": 2356
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.7881550788879395,
+      "learning_rate": 0.00023134457950880672,
+      "loss": 1.0587,
+      "step": 2357
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.2492504119873047,
+      "learning_rate": 0.0002313073678987844,
+      "loss": 0.8628,
+      "step": 2358
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8697025775909424,
+      "learning_rate": 0.0002312701562887621,
+      "loss": 0.622,
+      "step": 2359
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.8673083782196045,
+      "learning_rate": 0.00023123294467873975,
+      "loss": 0.7776,
+      "step": 2360
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.7203593254089355,
+      "learning_rate": 0.00023119573306871742,
+      "loss": 0.8226,
+      "step": 2361
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 10.66352653503418,
+      "learning_rate": 0.00023115852145869507,
+      "loss": 2.7217,
+      "step": 2362
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4907355308532715,
+      "learning_rate": 0.00023112130984867277,
+      "loss": 0.3725,
+      "step": 2363
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 6.29020357131958,
+      "learning_rate": 0.00023108409823865045,
+      "loss": 2.0326,
+      "step": 2364
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.2957611083984375,
+      "learning_rate": 0.0002310468866286281,
+      "loss": 0.6962,
+      "step": 2365
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.59990930557251,
+      "learning_rate": 0.0002310096750186058,
+      "loss": 0.9934,
+      "step": 2366
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.0413990020751953,
+      "learning_rate": 0.00023097246340858345,
+      "loss": 0.4285,
+      "step": 2367
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.2277538776397705,
+      "learning_rate": 0.00023093525179856112,
+      "loss": 0.4112,
+      "step": 2368
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.3201451301574707,
+      "learning_rate": 0.0002308980401885388,
+      "loss": 0.3224,
+      "step": 2369
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.672980785369873,
+      "learning_rate": 0.00023086082857851648,
+      "loss": 0.4006,
+      "step": 2370
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6918740272521973,
+      "learning_rate": 0.00023082361696849415,
+      "loss": 0.4209,
+      "step": 2371
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.196498394012451,
+      "learning_rate": 0.00023078640535847183,
+      "loss": 0.6705,
+      "step": 2372
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.965975046157837,
+      "learning_rate": 0.00023074919374844948,
+      "loss": 0.2825,
+      "step": 2373
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.151905059814453,
+      "learning_rate": 0.00023071198213842718,
+      "loss": 0.5709,
+      "step": 2374
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 9.685364723205566,
+      "learning_rate": 0.00023067477052840483,
+      "loss": 2.5129,
+      "step": 2375
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.8640772104263306,
+      "learning_rate": 0.0002306375589183825,
+      "loss": 0.7282,
+      "step": 2376
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.197077989578247,
+      "learning_rate": 0.0002306003473083602,
+      "loss": 0.6786,
+      "step": 2377
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.610832691192627,
+      "learning_rate": 0.00023056313569833786,
+      "loss": 0.4671,
+      "step": 2378
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.7640953063964844,
+      "learning_rate": 0.00023052592408831553,
+      "loss": 0.4904,
+      "step": 2379
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.4002923965454102,
+      "learning_rate": 0.00023048871247829318,
+      "loss": 0.3408,
+      "step": 2380
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.018538475036621,
+      "learning_rate": 0.00023045150086827088,
+      "loss": 0.5375,
+      "step": 2381
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6300442218780518,
+      "learning_rate": 0.00023041428925824856,
+      "loss": 0.4204,
+      "step": 2382
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 11.127724647521973,
+      "learning_rate": 0.0002303770776482262,
+      "loss": 2.8526,
+      "step": 2383
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.2783316373825073,
+      "learning_rate": 0.0002303398660382039,
+      "loss": 0.1801,
+      "step": 2384
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.074375867843628,
+      "learning_rate": 0.00023030265442818159,
+      "loss": 0.5052,
+      "step": 2385
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.3270208835601807,
+      "learning_rate": 0.00023026544281815923,
+      "loss": 0.306,
+      "step": 2386
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.8103556632995605,
+      "learning_rate": 0.00023022823120813694,
+      "loss": 1.2487,
+      "step": 2387
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.1849074363708496,
+      "learning_rate": 0.0002301910195981146,
+      "loss": 0.5772,
+      "step": 2388
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.9481422901153564,
+      "learning_rate": 0.00023015380798809226,
+      "loss": 1.5224,
+      "step": 2389
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.1063621044158936,
+      "learning_rate": 0.00023011659637806994,
+      "loss": 0.3233,
+      "step": 2390
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.34477162361145,
+      "learning_rate": 0.00023007938476804761,
+      "loss": 0.4639,
+      "step": 2391
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.563861608505249,
+      "learning_rate": 0.0002300421731580253,
+      "loss": 0.203,
+      "step": 2392
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.3610353469848633,
+      "learning_rate": 0.00023000496154800294,
+      "loss": 0.5348,
+      "step": 2393
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.9858109951019287,
+      "learning_rate": 0.00022996774993798061,
+      "loss": 0.2315,
+      "step": 2394
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 6.193589210510254,
+      "learning_rate": 0.00022993053832795832,
+      "loss": 3.4532,
+      "step": 2395
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 6.4159698486328125,
+      "learning_rate": 0.00022989332671793597,
+      "loss": 1.3103,
+      "step": 2396
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.0241963863372803,
+      "learning_rate": 0.00022985611510791364,
+      "loss": 0.1741,
+      "step": 2397
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.740751266479492,
+      "learning_rate": 0.00022981890349789134,
+      "loss": 0.7566,
+      "step": 2398
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.657395362854004,
+      "learning_rate": 0.000229781691887869,
+      "loss": 0.4493,
+      "step": 2399
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.8431265354156494,
+      "learning_rate": 0.00022974448027784667,
+      "loss": 0.6928,
+      "step": 2400
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 0.6465885639190674,
+      "eval_runtime": 50.1998,
+      "eval_samples_per_second": 43.187,
+      "eval_steps_per_second": 10.797,
+      "eval_wer": 0.5759384189586577,
+      "step": 2400
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.204583168029785,
+      "learning_rate": 0.00022970726866782432,
+      "loss": 0.8461,
+      "step": 2401
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.2085318565368652,
+      "learning_rate": 0.00022967005705780202,
+      "loss": 0.7323,
+      "step": 2402
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.1746833324432373,
+      "learning_rate": 0.0002296328454477797,
+      "loss": 0.5866,
+      "step": 2403
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.1107888221740723,
+      "learning_rate": 0.00022959563383775734,
+      "loss": 0.4081,
+      "step": 2404
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.436554193496704,
+      "learning_rate": 0.00022955842222773505,
+      "loss": 0.5447,
+      "step": 2405
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.6409112215042114,
+      "learning_rate": 0.0002295212106177127,
+      "loss": 0.4951,
+      "step": 2406
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.4227728843688965,
+      "learning_rate": 0.00022948399900769037,
+      "loss": 1.4331,
+      "step": 2407
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.892978549003601,
+      "learning_rate": 0.00022944678739766808,
+      "loss": 0.3022,
+      "step": 2408
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.203372001647949,
+      "learning_rate": 0.00022940957578764572,
+      "loss": 0.5053,
+      "step": 2409
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.97529411315918,
+      "learning_rate": 0.0002293723641776234,
+      "loss": 2.5665,
+      "step": 2410
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.133910894393921,
+      "learning_rate": 0.00022933515256760105,
+      "loss": 0.5585,
+      "step": 2411
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.481238842010498,
+      "learning_rate": 0.00022929794095757875,
+      "loss": 0.7493,
+      "step": 2412
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.110828399658203,
+      "learning_rate": 0.00022926072934755643,
+      "loss": 0.6066,
+      "step": 2413
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.420081377029419,
+      "learning_rate": 0.00022922351773753408,
+      "loss": 0.4951,
+      "step": 2414
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.824462890625,
+      "learning_rate": 0.00022918630612751175,
+      "loss": 0.6661,
+      "step": 2415
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.048468828201294,
+      "learning_rate": 0.00022914909451748945,
+      "loss": 0.6657,
+      "step": 2416
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.503145217895508,
+      "learning_rate": 0.0002291118829074671,
+      "loss": 0.4398,
+      "step": 2417
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5302720069885254,
+      "learning_rate": 0.00022907467129744478,
+      "loss": 0.4753,
+      "step": 2418
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.1612811088562012,
+      "learning_rate": 0.00022903745968742245,
+      "loss": 0.143,
+      "step": 2419
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 8.667667388916016,
+      "learning_rate": 0.00022900024807740013,
+      "loss": 0.6554,
+      "step": 2420
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.002796649932861,
+      "learning_rate": 0.0002289630364673778,
+      "loss": 0.5772,
+      "step": 2421
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 6.211899280548096,
+      "learning_rate": 0.00022892582485735546,
+      "loss": 1.3073,
+      "step": 2422
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.510040521621704,
+      "learning_rate": 0.00022888861324733316,
+      "loss": 0.6746,
+      "step": 2423
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 8.804295539855957,
+      "learning_rate": 0.0002288514016373108,
+      "loss": 2.8085,
+      "step": 2424
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.8787487745285034,
+      "learning_rate": 0.00022881419002728848,
+      "loss": 0.3381,
+      "step": 2425
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.092188596725464,
+      "learning_rate": 0.00022877697841726619,
+      "loss": 0.9032,
+      "step": 2426
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.2084264755249023,
+      "learning_rate": 0.00022873976680724383,
+      "loss": 0.6057,
+      "step": 2427
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.3689112663269043,
+      "learning_rate": 0.0002287025551972215,
+      "loss": 0.9147,
+      "step": 2428
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.9687527418136597,
+      "learning_rate": 0.0002286653435871992,
+      "loss": 0.5522,
+      "step": 2429
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.207888603210449,
+      "learning_rate": 0.00022862813197717686,
+      "loss": 0.6443,
+      "step": 2430
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4033074378967285,
+      "learning_rate": 0.00022859092036715454,
+      "loss": 0.5481,
+      "step": 2431
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.924738883972168,
+      "learning_rate": 0.00022855370875713219,
+      "loss": 0.2973,
+      "step": 2432
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.1252126693725586,
+      "learning_rate": 0.0002285164971471099,
+      "loss": 0.3515,
+      "step": 2433
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.03769850730896,
+      "learning_rate": 0.00022847928553708756,
+      "loss": 0.5084,
+      "step": 2434
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.743135929107666,
+      "learning_rate": 0.0002284420739270652,
+      "loss": 0.7894,
+      "step": 2435
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.572986602783203,
+      "learning_rate": 0.00022840486231704292,
+      "loss": 0.6504,
+      "step": 2436
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5167198181152344,
+      "learning_rate": 0.00022836765070702056,
+      "loss": 0.479,
+      "step": 2437
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.302143096923828,
+      "learning_rate": 0.00022833043909699824,
+      "loss": 0.6252,
+      "step": 2438
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.8964296579360962,
+      "learning_rate": 0.00022829322748697592,
+      "loss": 0.2638,
+      "step": 2439
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.6442536115646362,
+      "learning_rate": 0.0002282560158769536,
+      "loss": 0.1767,
+      "step": 2440
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.2055866718292236,
+      "learning_rate": 0.00022821880426693127,
+      "loss": 0.2613,
+      "step": 2441
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.640995979309082,
+      "learning_rate": 0.00022818159265690894,
+      "loss": 0.5495,
+      "step": 2442
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.59956955909729,
+      "learning_rate": 0.0002281443810468866,
+      "loss": 0.5332,
+      "step": 2443
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.761136531829834,
+      "learning_rate": 0.0002281071694368643,
+      "loss": 1.1939,
+      "step": 2444
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.0465445518493652,
+      "learning_rate": 0.00022806995782684194,
+      "loss": 0.3178,
+      "step": 2445
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.101271390914917,
+      "learning_rate": 0.00022803274621681962,
+      "loss": 0.5765,
+      "step": 2446
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.0883326530456543,
+      "learning_rate": 0.00022799553460679732,
+      "loss": 0.2466,
+      "step": 2447
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.459491491317749,
+      "learning_rate": 0.00022795832299677497,
+      "loss": 0.3144,
+      "step": 2448
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.362196445465088,
+      "learning_rate": 0.00022792111138675265,
+      "loss": 0.6192,
+      "step": 2449
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.9791548252105713,
+      "learning_rate": 0.0002278838997767303,
+      "loss": 0.548,
+      "step": 2450
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.309065103530884,
+      "learning_rate": 0.000227846688166708,
+      "loss": 1.0592,
+      "step": 2451
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.0640029907226562,
+      "learning_rate": 0.00022780947655668567,
+      "loss": 1.2627,
+      "step": 2452
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.2788937091827393,
+      "learning_rate": 0.00022777226494666332,
+      "loss": 0.7759,
+      "step": 2453
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.9871941804885864,
+      "learning_rate": 0.00022773505333664103,
+      "loss": 0.7314,
+      "step": 2454
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.758849620819092,
+      "learning_rate": 0.0002276978417266187,
+      "loss": 0.5533,
+      "step": 2455
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.1479053497314453,
+      "learning_rate": 0.00022766063011659635,
+      "loss": 0.6265,
+      "step": 2456
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5701639652252197,
+      "learning_rate": 0.00022762341850657405,
+      "loss": 0.6011,
+      "step": 2457
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6615350246429443,
+      "learning_rate": 0.0002275862068965517,
+      "loss": 0.6043,
+      "step": 2458
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.9056451320648193,
+      "learning_rate": 0.00022754899528652938,
+      "loss": 0.4244,
+      "step": 2459
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.3481318950653076,
+      "learning_rate": 0.00022751178367650705,
+      "loss": 0.3759,
+      "step": 2460
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.501145601272583,
+      "learning_rate": 0.00022747457206648473,
+      "loss": 0.3211,
+      "step": 2461
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.2316216230392456,
+      "learning_rate": 0.0002274373604564624,
+      "loss": 0.213,
+      "step": 2462
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.9179749488830566,
+      "learning_rate": 0.00022740014884644005,
+      "loss": 0.5273,
+      "step": 2463
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6479461193084717,
+      "learning_rate": 0.00022736293723641773,
+      "loss": 0.3405,
+      "step": 2464
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.821375846862793,
+      "learning_rate": 0.00022732572562639543,
+      "loss": 0.4056,
+      "step": 2465
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.5695273876190186,
+      "learning_rate": 0.00022728851401637308,
+      "loss": 0.65,
+      "step": 2466
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.268481969833374,
+      "learning_rate": 0.00022725130240635076,
+      "loss": 0.4224,
+      "step": 2467
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.3104357719421387,
+      "learning_rate": 0.0002272140907963284,
+      "loss": 0.5315,
+      "step": 2468
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.089568853378296,
+      "learning_rate": 0.0002271768791863061,
+      "loss": 0.5053,
+      "step": 2469
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.867792129516602,
+      "learning_rate": 0.00022713966757628378,
+      "loss": 1.8816,
+      "step": 2470
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.6889870166778564,
+      "learning_rate": 0.00022710245596626143,
+      "loss": 0.3397,
+      "step": 2471
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 5.504836559295654,
+      "learning_rate": 0.00022706524435623914,
+      "loss": 0.8586,
+      "step": 2472
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.099041700363159,
+      "learning_rate": 0.0002270280327462168,
+      "loss": 0.3921,
+      "step": 2473
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.710519790649414,
+      "learning_rate": 0.00022699082113619446,
+      "loss": 0.3917,
+      "step": 2474
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.710888624191284,
+      "learning_rate": 0.00022695360952617216,
+      "loss": 0.1841,
+      "step": 2475
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.646345376968384,
+      "learning_rate": 0.0002269163979161498,
+      "loss": 0.9153,
+      "step": 2476
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.3178653717041016,
+      "learning_rate": 0.0002268791863061275,
+      "loss": 0.6073,
+      "step": 2477
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 32.63193130493164,
+      "learning_rate": 0.0002268419746961052,
+      "loss": 6.8604,
+      "step": 2478
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.098633289337158,
+      "learning_rate": 0.00022680476308608284,
+      "loss": 0.6981,
+      "step": 2479
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.7247930765151978,
+      "learning_rate": 0.00022676755147606052,
+      "loss": 0.3698,
+      "step": 2480
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.1738440990448,
+      "learning_rate": 0.00022673033986603816,
+      "loss": 0.5705,
+      "step": 2481
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.0334391593933105,
+      "learning_rate": 0.00022669312825601587,
+      "loss": 0.3283,
+      "step": 2482
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.9234660863876343,
+      "learning_rate": 0.00022665591664599354,
+      "loss": 0.633,
+      "step": 2483
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.026090621948242,
+      "learning_rate": 0.0002266187050359712,
+      "loss": 0.3883,
+      "step": 2484
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.060999631881714,
+      "learning_rate": 0.00022658149342594887,
+      "loss": 0.6174,
+      "step": 2485
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.9415786266326904,
+      "learning_rate": 0.00022654428181592657,
+      "loss": 0.7415,
+      "step": 2486
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.4952282905578613,
+      "learning_rate": 0.00022650707020590422,
+      "loss": 0.7448,
+      "step": 2487
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.7627339363098145,
+      "learning_rate": 0.0002264698585958819,
+      "loss": 0.86,
+      "step": 2488
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.7875306606292725,
+      "learning_rate": 0.00022643264698585954,
+      "loss": 0.6431,
+      "step": 2489
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.5709052085876465,
+      "learning_rate": 0.00022639543537583725,
+      "loss": 0.4823,
+      "step": 2490
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.550630569458008,
+      "learning_rate": 0.00022635822376581492,
+      "loss": 0.951,
+      "step": 2491
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.399900436401367,
+      "learning_rate": 0.00022632101215579257,
+      "loss": 0.3805,
+      "step": 2492
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.55071759223938,
+      "learning_rate": 0.00022628380054577027,
+      "loss": 0.3896,
+      "step": 2493
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.296781063079834,
+      "learning_rate": 0.00022624658893574792,
+      "loss": 0.3268,
+      "step": 2494
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.0526139736175537,
+      "learning_rate": 0.0002262093773257256,
+      "loss": 0.6264,
+      "step": 2495
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.58652400970459,
+      "learning_rate": 0.0002261721657157033,
+      "loss": 0.5402,
+      "step": 2496
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 6.031370162963867,
+      "learning_rate": 0.00022613495410568095,
+      "loss": 1.0957,
+      "step": 2497
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.1321797370910645,
+      "learning_rate": 0.00022609774249565863,
+      "loss": 0.4783,
+      "step": 2498
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.8745222091674805,
+      "learning_rate": 0.00022606053088563633,
+      "loss": 0.3831,
+      "step": 2499
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.6690893173217773,
+      "learning_rate": 0.00022602331927561398,
+      "loss": 0.6712,
+      "step": 2500
+    },
+    {
+      "epoch": 0.88,
+      "eval_loss": 0.5969449877738953,
+      "eval_runtime": 50.5023,
+      "eval_samples_per_second": 42.929,
+      "eval_steps_per_second": 10.732,
+      "eval_wer": 0.5697111226431413,
+      "step": 2500
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.2524619102478027,
+      "learning_rate": 0.00022598610766559165,
+      "loss": 1.1098,
+      "step": 2501
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.3563032150268555,
+      "learning_rate": 0.0002259488960555693,
+      "loss": 0.8505,
+      "step": 2502
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.617898941040039,
+      "learning_rate": 0.000225911684445547,
+      "loss": 0.5266,
+      "step": 2503
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.1156997680664062,
+      "learning_rate": 0.00022587447283552468,
+      "loss": 0.285,
+      "step": 2504
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.1009535789489746,
+      "learning_rate": 0.00022583726122550233,
+      "loss": 0.4942,
+      "step": 2505
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.473078966140747,
+      "learning_rate": 0.00022580004961548,
+      "loss": 0.5901,
+      "step": 2506
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.716780662536621,
+      "learning_rate": 0.00022576283800545768,
+      "loss": 0.834,
+      "step": 2507
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.1853911876678467,
+      "learning_rate": 0.00022572562639543536,
+      "loss": 0.7706,
+      "step": 2508
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.181541919708252,
+      "learning_rate": 0.00022568841478541303,
+      "loss": 0.9358,
+      "step": 2509
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.0345635414123535,
+      "learning_rate": 0.00022565120317539068,
+      "loss": 0.6332,
+      "step": 2510
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.495234966278076,
+      "learning_rate": 0.00022561399156536838,
+      "loss": 0.5035,
+      "step": 2511
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.819892168045044,
+      "learning_rate": 0.00022557677995534603,
+      "loss": 0.4648,
+      "step": 2512
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.7033729553222656,
+      "learning_rate": 0.0002255395683453237,
+      "loss": 0.5328,
+      "step": 2513
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.782414197921753,
+      "learning_rate": 0.0002255023567353014,
+      "loss": 0.5677,
+      "step": 2514
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.6469204425811768,
+      "learning_rate": 0.00022546514512527906,
+      "loss": 0.6008,
+      "step": 2515
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.3379626274108887,
+      "learning_rate": 0.00022542793351525674,
+      "loss": 0.621,
+      "step": 2516
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.2679710388183594,
+      "learning_rate": 0.00022539072190523444,
+      "loss": 0.5374,
+      "step": 2517
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.7875244617462158,
+      "learning_rate": 0.0002253535102952121,
+      "loss": 0.2722,
+      "step": 2518
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.986699104309082,
+      "learning_rate": 0.00022531629868518976,
+      "loss": 0.4861,
+      "step": 2519
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.055794715881348,
+      "learning_rate": 0.0002252790870751674,
+      "loss": 0.8421,
+      "step": 2520
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.854576587677002,
+      "learning_rate": 0.00022524187546514511,
+      "loss": 0.2723,
+      "step": 2521
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.405861854553223,
+      "learning_rate": 0.0002252046638551228,
+      "loss": 0.6845,
+      "step": 2522
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 7.5870161056518555,
+      "learning_rate": 0.00022516745224510044,
+      "loss": 2.0838,
+      "step": 2523
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9509167671203613,
+      "learning_rate": 0.00022513024063507814,
+      "loss": 0.2952,
+      "step": 2524
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.6641058921813965,
+      "learning_rate": 0.0002250930290250558,
+      "loss": 0.4337,
+      "step": 2525
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.685914993286133,
+      "learning_rate": 0.00022505581741503347,
+      "loss": 1.4865,
+      "step": 2526
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.8381335735321045,
+      "learning_rate": 0.00022501860580501114,
+      "loss": 0.8048,
+      "step": 2527
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.5004918575286865,
+      "learning_rate": 0.00022498139419498882,
+      "loss": 0.8519,
+      "step": 2528
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.1345882415771484,
+      "learning_rate": 0.0002249441825849665,
+      "loss": 0.6214,
+      "step": 2529
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.871791362762451,
+      "learning_rate": 0.00022490697097494417,
+      "loss": 1.7872,
+      "step": 2530
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.4108870029449463,
+      "learning_rate": 0.00022486975936492182,
+      "loss": 0.4127,
+      "step": 2531
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.154738426208496,
+      "learning_rate": 0.00022483254775489952,
+      "loss": 0.3094,
+      "step": 2532
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.8952206373214722,
+      "learning_rate": 0.00022479533614487717,
+      "loss": 0.4161,
+      "step": 2533
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7930335998535156,
+      "learning_rate": 0.00022475812453485485,
+      "loss": 0.7349,
+      "step": 2534
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.5914433002471924,
+      "learning_rate": 0.00022472091292483255,
+      "loss": 0.6303,
+      "step": 2535
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.9063329696655273,
+      "learning_rate": 0.0002246837013148102,
+      "loss": 0.7971,
+      "step": 2536
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.814877986907959,
+      "learning_rate": 0.00022464648970478787,
+      "loss": 0.6404,
+      "step": 2537
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.2921807765960693,
+      "learning_rate": 0.00022460927809476552,
+      "loss": 0.6443,
+      "step": 2538
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.2125370502471924,
+      "learning_rate": 0.00022457206648474322,
+      "loss": 1.599,
+      "step": 2539
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.036359786987305,
+      "learning_rate": 0.0002245348548747209,
+      "loss": 0.8502,
+      "step": 2540
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.4251275062561035,
+      "learning_rate": 0.00022449764326469855,
+      "loss": 0.3993,
+      "step": 2541
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.6843395233154297,
+      "learning_rate": 0.00022446043165467625,
+      "loss": 1.2595,
+      "step": 2542
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.9409282207489014,
+      "learning_rate": 0.00022442322004465393,
+      "loss": 0.213,
+      "step": 2543
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.4829013347625732,
+      "learning_rate": 0.00022438600843463158,
+      "loss": 0.4155,
+      "step": 2544
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 12.528837203979492,
+      "learning_rate": 0.00022434879682460928,
+      "loss": 2.7458,
+      "step": 2545
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.5831103324890137,
+      "learning_rate": 0.00022431158521458693,
+      "loss": 0.9758,
+      "step": 2546
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.7440414428710938,
+      "learning_rate": 0.0002242743736045646,
+      "loss": 0.8341,
+      "step": 2547
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.1389410495758057,
+      "learning_rate": 0.00022423716199454228,
+      "loss": 0.3766,
+      "step": 2548
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.8604750633239746,
+      "learning_rate": 0.00022419995038451996,
+      "loss": 0.4142,
+      "step": 2549
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.646600246429443,
+      "learning_rate": 0.00022416273877449763,
+      "loss": 0.4988,
+      "step": 2550
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.2489912509918213,
+      "learning_rate": 0.00022412552716447528,
+      "loss": 1.1134,
+      "step": 2551
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.9654885530471802,
+      "learning_rate": 0.00022408831555445296,
+      "loss": 0.4167,
+      "step": 2552
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.2615177631378174,
+      "learning_rate": 0.00022405110394443066,
+      "loss": 0.6725,
+      "step": 2553
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.323215961456299,
+      "learning_rate": 0.0002240138923344083,
+      "loss": 0.9081,
+      "step": 2554
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.1853933334350586,
+      "learning_rate": 0.00022397668072438598,
+      "loss": 0.5613,
+      "step": 2555
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7656657695770264,
+      "learning_rate": 0.00022393946911436363,
+      "loss": 0.4834,
+      "step": 2556
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5364134311676025,
+      "learning_rate": 0.00022390225750434133,
+      "loss": 0.8214,
+      "step": 2557
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.2517011165618896,
+      "learning_rate": 0.000223865045894319,
+      "loss": 0.6773,
+      "step": 2558
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 6.1538166999816895,
+      "learning_rate": 0.00022382783428429666,
+      "loss": 2.0649,
+      "step": 2559
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.786186695098877,
+      "learning_rate": 0.00022379062267427436,
+      "loss": 0.6274,
+      "step": 2560
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.8489811420440674,
+      "learning_rate": 0.00022375341106425204,
+      "loss": 0.5626,
+      "step": 2561
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 14.115158081054688,
+      "learning_rate": 0.0002237161994542297,
+      "loss": 3.3753,
+      "step": 2562
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.628080129623413,
+      "learning_rate": 0.0002236789878442074,
+      "loss": 0.333,
+      "step": 2563
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.309471607208252,
+      "learning_rate": 0.00022364177623418504,
+      "loss": 0.3252,
+      "step": 2564
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.064052104949951,
+      "learning_rate": 0.00022360456462416271,
+      "loss": 0.7157,
+      "step": 2565
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.8947696685791016,
+      "learning_rate": 0.00022356735301414042,
+      "loss": 0.3062,
+      "step": 2566
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.9529271125793457,
+      "learning_rate": 0.00022353014140411807,
+      "loss": 0.3285,
+      "step": 2567
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.573422431945801,
+      "learning_rate": 0.00022349292979409574,
+      "loss": 0.462,
+      "step": 2568
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.323622703552246,
+      "learning_rate": 0.0002234557181840734,
+      "loss": 0.4158,
+      "step": 2569
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.2690248489379883,
+      "learning_rate": 0.0002234185065740511,
+      "loss": 1.531,
+      "step": 2570
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.335678219795227,
+      "learning_rate": 0.00022338129496402877,
+      "loss": 0.1792,
+      "step": 2571
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.555920362472534,
+      "learning_rate": 0.00022334408335400642,
+      "loss": 0.3438,
+      "step": 2572
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.97938072681427,
+      "learning_rate": 0.0002233068717439841,
+      "loss": 0.1992,
+      "step": 2573
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.451127529144287,
+      "learning_rate": 0.0002232696601339618,
+      "loss": 0.9753,
+      "step": 2574
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.8145601749420166,
+      "learning_rate": 0.00022323244852393944,
+      "loss": 0.5944,
+      "step": 2575
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.062714099884033,
+      "learning_rate": 0.00022319523691391712,
+      "loss": 1.197,
+      "step": 2576
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.360086441040039,
+      "learning_rate": 0.00022315802530389477,
+      "loss": 0.4267,
+      "step": 2577
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.727750301361084,
+      "learning_rate": 0.00022312081369387247,
+      "loss": 1.1279,
+      "step": 2578
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.5544394254684448,
+      "learning_rate": 0.00022308360208385015,
+      "loss": 0.5312,
+      "step": 2579
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.51163387298584,
+      "learning_rate": 0.0002230463904738278,
+      "loss": 0.3286,
+      "step": 2580
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.9520326852798462,
+      "learning_rate": 0.0002230091788638055,
+      "loss": 0.6118,
+      "step": 2581
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.6972302198410034,
+      "learning_rate": 0.00022297196725378315,
+      "loss": 0.289,
+      "step": 2582
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.5757126808166504,
+      "learning_rate": 0.00022293475564376082,
+      "loss": 0.4856,
+      "step": 2583
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.543616533279419,
+      "learning_rate": 0.00022289754403373853,
+      "loss": 0.5646,
+      "step": 2584
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.380565643310547,
+      "learning_rate": 0.00022286033242371618,
+      "loss": 0.5098,
+      "step": 2585
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.8600701093673706,
+      "learning_rate": 0.00022282312081369385,
+      "loss": 0.4146,
+      "step": 2586
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.9469465017318726,
+      "learning_rate": 0.00022278590920367155,
+      "loss": 0.3405,
+      "step": 2587
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 6.387368679046631,
+      "learning_rate": 0.0002227486975936492,
+      "loss": 2.7503,
+      "step": 2588
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.1771650314331055,
+      "learning_rate": 0.00022271148598362688,
+      "loss": 0.2495,
+      "step": 2589
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.2582592964172363,
+      "learning_rate": 0.00022267427437360453,
+      "loss": 0.7721,
+      "step": 2590
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.26287317276001,
+      "learning_rate": 0.00022263706276358223,
+      "loss": 0.3901,
+      "step": 2591
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.030311584472656,
+      "learning_rate": 0.0002225998511535599,
+      "loss": 1.1871,
+      "step": 2592
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.2739689350128174,
+      "learning_rate": 0.00022256263954353756,
+      "loss": 0.6562,
+      "step": 2593
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.621596574783325,
+      "learning_rate": 0.00022252542793351523,
+      "loss": 0.3165,
+      "step": 2594
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 9.08333683013916,
+      "learning_rate": 0.0002224882163234929,
+      "loss": 2.6412,
+      "step": 2595
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8135602474212646,
+      "learning_rate": 0.00022245100471347058,
+      "loss": 0.5285,
+      "step": 2596
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.7928903102874756,
+      "learning_rate": 0.00022241379310344826,
+      "loss": 0.3371,
+      "step": 2597
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.8602352142333984,
+      "learning_rate": 0.0002223765814934259,
+      "loss": 0.2245,
+      "step": 2598
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.1992337703704834,
+      "learning_rate": 0.0002223393698834036,
+      "loss": 0.5035,
+      "step": 2599
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 7.390182971954346,
+      "learning_rate": 0.00022230215827338126,
+      "loss": 0.8551,
+      "step": 2600
+    },
+    {
+      "epoch": 0.91,
+      "eval_loss": 0.6041048169136047,
+      "eval_runtime": 51.0519,
+      "eval_samples_per_second": 42.467,
+      "eval_steps_per_second": 10.617,
+      "eval_wer": 0.5422072305829442,
+      "step": 2600
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8501031398773193,
+      "learning_rate": 0.00022226494666335893,
+      "loss": 0.9814,
+      "step": 2601
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.3961238861083984,
+      "learning_rate": 0.00022222773505333664,
+      "loss": 1.0017,
+      "step": 2602
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.359489917755127,
+      "learning_rate": 0.00022219052344331429,
+      "loss": 0.5428,
+      "step": 2603
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.420226573944092,
+      "learning_rate": 0.00022215331183329196,
+      "loss": 0.854,
+      "step": 2604
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.7213860750198364,
+      "learning_rate": 0.00022211610022326966,
+      "loss": 0.2745,
+      "step": 2605
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.067789077758789,
+      "learning_rate": 0.0002220788886132473,
+      "loss": 0.5725,
+      "step": 2606
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9434151649475098,
+      "learning_rate": 0.000222041677003225,
+      "loss": 0.4392,
+      "step": 2607
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.986076593399048,
+      "learning_rate": 0.00022200446539320264,
+      "loss": 0.5028,
+      "step": 2608
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.050105094909668,
+      "learning_rate": 0.00022196725378318034,
+      "loss": 0.506,
+      "step": 2609
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 8.282485961914062,
+      "learning_rate": 0.00022193004217315802,
+      "loss": 2.1749,
+      "step": 2610
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.559239864349365,
+      "learning_rate": 0.00022189283056313567,
+      "loss": 0.4975,
+      "step": 2611
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.908665895462036,
+      "learning_rate": 0.00022185561895311337,
+      "loss": 0.3902,
+      "step": 2612
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.392807722091675,
+      "learning_rate": 0.00022181840734309102,
+      "loss": 0.2691,
+      "step": 2613
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.977257251739502,
+      "learning_rate": 0.0002217811957330687,
+      "loss": 0.5115,
+      "step": 2614
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.195502758026123,
+      "learning_rate": 0.00022174398412304637,
+      "loss": 0.3884,
+      "step": 2615
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.966074228286743,
+      "learning_rate": 0.00022170677251302404,
+      "loss": 0.7567,
+      "step": 2616
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.6738837957382202,
+      "learning_rate": 0.00022166956090300172,
+      "loss": 0.238,
+      "step": 2617
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.954295039176941,
+      "learning_rate": 0.0002216323492929794,
+      "loss": 0.4697,
+      "step": 2618
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.3093600273132324,
+      "learning_rate": 0.00022159513768295704,
+      "loss": 0.2654,
+      "step": 2619
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 9.452103614807129,
+      "learning_rate": 0.00022155792607293475,
+      "loss": 0.5994,
+      "step": 2620
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.7643561363220215,
+      "learning_rate": 0.0002215207144629124,
+      "loss": 0.2038,
+      "step": 2621
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.625439167022705,
+      "learning_rate": 0.00022148350285289007,
+      "loss": 0.6295,
+      "step": 2622
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 6.1190996170043945,
+      "learning_rate": 0.00022144629124286777,
+      "loss": 0.4905,
+      "step": 2623
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.208706378936768,
+      "learning_rate": 0.00022140907963284542,
+      "loss": 1.0413,
+      "step": 2624
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 6.129053592681885,
+      "learning_rate": 0.0002213718680228231,
+      "loss": 0.58,
+      "step": 2625
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.2219321727752686,
+      "learning_rate": 0.00022133465641280075,
+      "loss": 1.1687,
+      "step": 2626
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.6231391429901123,
+      "learning_rate": 0.00022129744480277845,
+      "loss": 0.6352,
+      "step": 2627
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.8433243036270142,
+      "learning_rate": 0.00022126023319275613,
+      "loss": 0.7305,
+      "step": 2628
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.111185073852539,
+      "learning_rate": 0.00022122302158273378,
+      "loss": 0.4803,
+      "step": 2629
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.3844072818756104,
+      "learning_rate": 0.00022118580997271148,
+      "loss": 0.5074,
+      "step": 2630
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.9091522693634033,
+      "learning_rate": 0.00022114859836268915,
+      "loss": 0.6794,
+      "step": 2631
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.5186028480529785,
+      "learning_rate": 0.0002211113867526668,
+      "loss": 0.6303,
+      "step": 2632
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.9880521297454834,
+      "learning_rate": 0.0002210741751426445,
+      "loss": 0.3124,
+      "step": 2633
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.2950042486190796,
+      "learning_rate": 0.00022103696353262215,
+      "loss": 0.1439,
+      "step": 2634
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.8072837591171265,
+      "learning_rate": 0.00022099975192259983,
+      "loss": 0.1895,
+      "step": 2635
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.449474334716797,
+      "learning_rate": 0.0002209625403125775,
+      "loss": 0.5064,
+      "step": 2636
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.273451805114746,
+      "learning_rate": 0.00022092532870255518,
+      "loss": 0.7899,
+      "step": 2637
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.689117193222046,
+      "learning_rate": 0.00022088811709253286,
+      "loss": 0.4124,
+      "step": 2638
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.5043230056762695,
+      "learning_rate": 0.0002208509054825105,
+      "loss": 0.6174,
+      "step": 2639
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7596065998077393,
+      "learning_rate": 0.0002208136938724882,
+      "loss": 0.0876,
+      "step": 2640
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.0988588333129883,
+      "learning_rate": 0.00022077648226246588,
+      "loss": 0.486,
+      "step": 2641
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.2585132122039795,
+      "learning_rate": 0.00022073927065244353,
+      "loss": 0.7462,
+      "step": 2642
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.4397363662719727,
+      "learning_rate": 0.0002207020590424212,
+      "loss": 0.2855,
+      "step": 2643
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.3539910316467285,
+      "learning_rate": 0.0002206648474323989,
+      "loss": 0.6796,
+      "step": 2644
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 6.321644306182861,
+      "learning_rate": 0.00022062763582237656,
+      "loss": 1.5811,
+      "step": 2645
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.399683952331543,
+      "learning_rate": 0.00022059042421235424,
+      "loss": 0.7615,
+      "step": 2646
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.508324146270752,
+      "learning_rate": 0.00022055321260233189,
+      "loss": 1.0115,
+      "step": 2647
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.360750198364258,
+      "learning_rate": 0.0002205160009923096,
+      "loss": 0.6244,
+      "step": 2648
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8480136394500732,
+      "learning_rate": 0.00022047878938228726,
+      "loss": 0.444,
+      "step": 2649
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": NaN,
+      "learning_rate": 0.00022047878938228726,
+      "loss": 0.3347,
+      "step": 2650
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.590242385864258,
+      "learning_rate": 0.0002204415777722649,
+      "loss": 0.506,
+      "step": 2651
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.5283608436584473,
+      "learning_rate": 0.00022040436616224262,
+      "loss": 0.5284,
+      "step": 2652
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.57216477394104,
+      "learning_rate": 0.00022036715455222026,
+      "loss": 1.126,
+      "step": 2653
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9254002571105957,
+      "learning_rate": 0.00022032994294219794,
+      "loss": 0.7635,
+      "step": 2654
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.162715911865234,
+      "learning_rate": 0.00022029273133217564,
+      "loss": 0.7601,
+      "step": 2655
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.1532511711120605,
+      "learning_rate": 0.0002202555197221533,
+      "loss": 0.941,
+      "step": 2656
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.203185796737671,
+      "learning_rate": 0.00022021830811213097,
+      "loss": 0.4708,
+      "step": 2657
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.62638521194458,
+      "learning_rate": 0.00022018109650210862,
+      "loss": 0.3295,
+      "step": 2658
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.998661994934082,
+      "learning_rate": 0.00022014388489208632,
+      "loss": 1.4867,
+      "step": 2659
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.389922857284546,
+      "learning_rate": 0.000220106673282064,
+      "loss": 1.2759,
+      "step": 2660
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.583156108856201,
+      "learning_rate": 0.00022006946167204164,
+      "loss": 0.4192,
+      "step": 2661
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.610195159912109,
+      "learning_rate": 0.00022003225006201935,
+      "loss": 0.5941,
+      "step": 2662
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.2611424922943115,
+      "learning_rate": 0.00021999503845199702,
+      "loss": 0.7521,
+      "step": 2663
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.2116403579711914,
+      "learning_rate": 0.00021995782684197467,
+      "loss": 0.4965,
+      "step": 2664
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.487800359725952,
+      "learning_rate": 0.00021992061523195235,
+      "loss": 0.7573,
+      "step": 2665
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.6182491779327393,
+      "learning_rate": 0.00021988340362193002,
+      "loss": 0.4018,
+      "step": 2666
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.2056775093078613,
+      "learning_rate": 0.0002198461920119077,
+      "loss": 0.3728,
+      "step": 2667
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.8955894708633423,
+      "learning_rate": 0.00021980898040188537,
+      "loss": 0.4454,
+      "step": 2668
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.290935754776001,
+      "learning_rate": 0.00021977176879186302,
+      "loss": 0.8278,
+      "step": 2669
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.875994682312012,
+      "learning_rate": 0.00021973455718184073,
+      "loss": 0.3937,
+      "step": 2670
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 7.943798065185547,
+      "learning_rate": 0.00021969734557181837,
+      "loss": 2.0068,
+      "step": 2671
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.384636878967285,
+      "learning_rate": 0.00021966013396179605,
+      "loss": 0.6497,
+      "step": 2672
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.006543159484863,
+      "learning_rate": 0.00021962292235177375,
+      "loss": 0.4773,
+      "step": 2673
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.17258358001709,
+      "learning_rate": 0.0002195857107417514,
+      "loss": 0.3188,
+      "step": 2674
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.57729434967041,
+      "learning_rate": 0.00021954849913172908,
+      "loss": 0.857,
+      "step": 2675
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.9307923316955566,
+      "learning_rate": 0.00021951128752170678,
+      "loss": 0.6661,
+      "step": 2676
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.026794910430908,
+      "learning_rate": 0.00021947407591168443,
+      "loss": 0.4217,
+      "step": 2677
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6623916625976562,
+      "learning_rate": 0.0002194368643016621,
+      "loss": 0.5877,
+      "step": 2678
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.1093175411224365,
+      "learning_rate": 0.00021939965269163975,
+      "loss": 0.3837,
+      "step": 2679
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.958167552947998,
+      "learning_rate": 0.00021936244108161746,
+      "loss": 0.438,
+      "step": 2680
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.3107874393463135,
+      "learning_rate": 0.00021932522947159513,
+      "loss": 0.2563,
+      "step": 2681
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.5440564155578613,
+      "learning_rate": 0.00021928801786157278,
+      "loss": 0.5432,
+      "step": 2682
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.4444539546966553,
+      "learning_rate": 0.00021925080625155048,
+      "loss": 0.8262,
+      "step": 2683
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.0069375038146973,
+      "learning_rate": 0.00021921359464152813,
+      "loss": 0.8865,
+      "step": 2684
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6746938228607178,
+      "learning_rate": 0.0002191763830315058,
+      "loss": 0.425,
+      "step": 2685
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.956552028656006,
+      "learning_rate": 0.00021913917142148348,
+      "loss": 0.6153,
+      "step": 2686
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7592060565948486,
+      "learning_rate": 0.00021910195981146116,
+      "loss": 0.6633,
+      "step": 2687
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.972530841827393,
+      "learning_rate": 0.00021906474820143884,
+      "loss": 0.5743,
+      "step": 2688
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.044546365737915,
+      "learning_rate": 0.0002190275365914165,
+      "loss": 0.2586,
+      "step": 2689
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6088757514953613,
+      "learning_rate": 0.00021899032498139416,
+      "loss": 0.2942,
+      "step": 2690
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.929062843322754,
+      "learning_rate": 0.00021895311337137186,
+      "loss": 0.4355,
+      "step": 2691
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.158072471618652,
+      "learning_rate": 0.0002189159017613495,
+      "loss": 0.597,
+      "step": 2692
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.822199821472168,
+      "learning_rate": 0.0002188786901513272,
+      "loss": 0.5923,
+      "step": 2693
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6804251670837402,
+      "learning_rate": 0.0002188414785413049,
+      "loss": 0.3537,
+      "step": 2694
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.3298931121826172,
+      "learning_rate": 0.00021880426693128254,
+      "loss": 0.1007,
+      "step": 2695
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.8608430624008179,
+      "learning_rate": 0.00021876705532126022,
+      "loss": 0.1824,
+      "step": 2696
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.879000663757324,
+      "learning_rate": 0.00021872984371123786,
+      "loss": 0.3099,
+      "step": 2697
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.222035884857178,
+      "learning_rate": 0.00021869263210121557,
+      "loss": 0.6009,
+      "step": 2698
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.855086326599121,
+      "learning_rate": 0.00021865542049119324,
+      "loss": 0.6408,
+      "step": 2699
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.642244577407837,
+      "learning_rate": 0.0002186182088811709,
+      "loss": 0.2426,
+      "step": 2700
+    },
+    {
+      "epoch": 0.95,
+      "eval_loss": 0.5336953401565552,
+      "eval_runtime": 51.4996,
+      "eval_samples_per_second": 42.097,
+      "eval_steps_per_second": 10.524,
+      "eval_wer": 0.48719944646254976,
+      "step": 2700
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.4252710342407227,
+      "learning_rate": 0.0002185809972711486,
+      "loss": 0.9991,
+      "step": 2701
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.518662452697754,
+      "learning_rate": 0.00021854378566112624,
+      "loss": 0.5021,
+      "step": 2702
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.2588770389556885,
+      "learning_rate": 0.00021850657405110392,
+      "loss": 0.6613,
+      "step": 2703
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.61147141456604,
+      "learning_rate": 0.00021846936244108162,
+      "loss": 0.6095,
+      "step": 2704
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.213031053543091,
+      "learning_rate": 0.00021843215083105927,
+      "loss": 0.3952,
+      "step": 2705
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.7817362546920776,
+      "learning_rate": 0.00021839493922103695,
+      "loss": 0.467,
+      "step": 2706
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.5945087671279907,
+      "learning_rate": 0.00021835772761101462,
+      "loss": 0.2962,
+      "step": 2707
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.3003673553466797,
+      "learning_rate": 0.0002183205160009923,
+      "loss": 0.5996,
+      "step": 2708
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.145582675933838,
+      "learning_rate": 0.00021828330439096997,
+      "loss": 0.4746,
+      "step": 2709
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.858013153076172,
+      "learning_rate": 0.00021824609278094762,
+      "loss": 0.7782,
+      "step": 2710
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.5435516834259033,
+      "learning_rate": 0.0002182088811709253,
+      "loss": 0.5275,
+      "step": 2711
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.836350202560425,
+      "learning_rate": 0.000218171669560903,
+      "loss": 0.4082,
+      "step": 2712
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.8937970399856567,
+      "learning_rate": 0.00021813445795088065,
+      "loss": 0.3822,
+      "step": 2713
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.3945305347442627,
+      "learning_rate": 0.00021809724634085833,
+      "loss": 0.371,
+      "step": 2714
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.8893818855285645,
+      "learning_rate": 0.00021806003473083597,
+      "loss": 0.2202,
+      "step": 2715
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.0246646404266357,
+      "learning_rate": 0.00021802282312081368,
+      "loss": 0.1932,
+      "step": 2716
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.8880715370178223,
+      "learning_rate": 0.00021798561151079135,
+      "loss": 0.2989,
+      "step": 2717
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.3225202560424805,
+      "learning_rate": 0.000217948399900769,
+      "loss": 1.991,
+      "step": 2718
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.143655776977539,
+      "learning_rate": 0.0002179111882907467,
+      "loss": 0.3331,
+      "step": 2719
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.7824084758758545,
+      "learning_rate": 0.00021787397668072438,
+      "loss": 0.3482,
+      "step": 2720
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.4806097745895386,
+      "learning_rate": 0.00021783676507070203,
+      "loss": 0.1659,
+      "step": 2721
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.375702381134033,
+      "learning_rate": 0.00021779955346067973,
+      "loss": 0.7355,
+      "step": 2722
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.235441207885742,
+      "learning_rate": 0.00021776234185065738,
+      "loss": 0.5138,
+      "step": 2723
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.325510501861572,
+      "learning_rate": 0.00021772513024063506,
+      "loss": 0.2818,
+      "step": 2724
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.0752792358398438,
+      "learning_rate": 0.00021768791863061276,
+      "loss": 0.1656,
+      "step": 2725
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6207728385925293,
+      "learning_rate": 0.0002176507070205904,
+      "loss": 1.094,
+      "step": 2726
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.5188002586364746,
+      "learning_rate": 0.00021761349541056808,
+      "loss": 0.2545,
+      "step": 2727
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.4537734985351562,
+      "learning_rate": 0.00021757628380054573,
+      "loss": 0.476,
+      "step": 2728
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.6389610767364502,
+      "learning_rate": 0.00021753907219052343,
+      "loss": 0.3701,
+      "step": 2729
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6082892417907715,
+      "learning_rate": 0.0002175018605805011,
+      "loss": 0.5414,
+      "step": 2730
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.334270477294922,
+      "learning_rate": 0.00021746464897047876,
+      "loss": 0.5533,
+      "step": 2731
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.05968713760376,
+      "learning_rate": 0.00021742743736045644,
+      "loss": 0.7159,
+      "step": 2732
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.399036407470703,
+      "learning_rate": 0.00021739022575043414,
+      "loss": 1.2499,
+      "step": 2733
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.3443745374679565,
+      "learning_rate": 0.0002173530141404118,
+      "loss": 0.176,
+      "step": 2734
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.077892303466797,
+      "learning_rate": 0.00021731580253038946,
+      "loss": 0.2954,
+      "step": 2735
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6642885208129883,
+      "learning_rate": 0.0002172785909203671,
+      "loss": 0.2387,
+      "step": 2736
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.107677459716797,
+      "learning_rate": 0.00021724137931034481,
+      "loss": 0.6326,
+      "step": 2737
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.31645131111145,
+      "learning_rate": 0.0002172041677003225,
+      "loss": 0.4842,
+      "step": 2738
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.741530418395996,
+      "learning_rate": 0.00021716695609030014,
+      "loss": 0.3405,
+      "step": 2739
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.6625313758850098,
+      "learning_rate": 0.00021712974448027784,
+      "loss": 0.1492,
+      "step": 2740
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.23145866394043,
+      "learning_rate": 0.0002170925328702555,
+      "loss": 0.4991,
+      "step": 2741
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.0887067317962646,
+      "learning_rate": 0.00021705532126023317,
+      "loss": 0.3526,
+      "step": 2742
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.2543463706970215,
+      "learning_rate": 0.00021701810965021087,
+      "loss": 0.6271,
+      "step": 2743
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.850689172744751,
+      "learning_rate": 0.00021698089804018852,
+      "loss": 0.5781,
+      "step": 2744
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.8406736850738525,
+      "learning_rate": 0.0002169436864301662,
+      "loss": 0.5398,
+      "step": 2745
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.770142078399658,
+      "learning_rate": 0.00021690647482014384,
+      "loss": 0.1511,
+      "step": 2746
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.917293071746826,
+      "learning_rate": 0.00021686926321012154,
+      "loss": 1.3509,
+      "step": 2747
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.7973790168762207,
+      "learning_rate": 0.00021683205160009922,
+      "loss": 0.1045,
+      "step": 2748
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.61274790763855,
+      "learning_rate": 0.00021679483999007687,
+      "loss": 0.3338,
+      "step": 2749
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.060636043548584,
+      "learning_rate": 0.00021675762838005457,
+      "loss": 0.2876,
+      "step": 2750
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.3534083366394043,
+      "learning_rate": 0.00021672041677003225,
+      "loss": 1.3637,
+      "step": 2751
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.5700647830963135,
+      "learning_rate": 0.0002166832051600099,
+      "loss": 0.9309,
+      "step": 2752
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.867060661315918,
+      "learning_rate": 0.00021664599354998757,
+      "loss": 0.4095,
+      "step": 2753
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.7170629501342773,
+      "learning_rate": 0.00021660878193996525,
+      "loss": 0.7493,
+      "step": 2754
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.105668783187866,
+      "learning_rate": 0.00021657157032994292,
+      "loss": 0.6006,
+      "step": 2755
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.1052336692810059,
+      "learning_rate": 0.0002165343587199206,
+      "loss": 0.097,
+      "step": 2756
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.781369924545288,
+      "learning_rate": 0.00021649714710989825,
+      "loss": 0.5647,
+      "step": 2757
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.3035714626312256,
+      "learning_rate": 0.00021645993549987595,
+      "loss": 0.3785,
+      "step": 2758
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.3320902585983276,
+      "learning_rate": 0.0002164227238898536,
+      "loss": 0.2026,
+      "step": 2759
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.859253168106079,
+      "learning_rate": 0.00021638551227983128,
+      "loss": 0.2957,
+      "step": 2760
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.7369508743286133,
+      "learning_rate": 0.00021634830066980898,
+      "loss": 0.6835,
+      "step": 2761
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.1216232776641846,
+      "learning_rate": 0.00021631108905978663,
+      "loss": 0.2972,
+      "step": 2762
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.9394500255584717,
+      "learning_rate": 0.0002162738774497643,
+      "loss": 0.5818,
+      "step": 2763
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.9596755504608154,
+      "learning_rate": 0.000216236665839742,
+      "loss": 0.4004,
+      "step": 2764
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.52447509765625,
+      "learning_rate": 0.00021619945422971966,
+      "loss": 0.4622,
+      "step": 2765
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.721889019012451,
+      "learning_rate": 0.00021616224261969733,
+      "loss": 0.2771,
+      "step": 2766
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.8889129161834717,
+      "learning_rate": 0.00021612503100967498,
+      "loss": 0.6328,
+      "step": 2767
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.644382953643799,
+      "learning_rate": 0.00021608781939965268,
+      "loss": 0.5304,
+      "step": 2768
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.635915279388428,
+      "learning_rate": 0.00021605060778963036,
+      "loss": 0.6429,
+      "step": 2769
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.2131359577178955,
+      "learning_rate": 0.000216013396179608,
+      "loss": 1.1534,
+      "step": 2770
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.65970516204834,
+      "learning_rate": 0.0002159761845695857,
+      "loss": 0.3379,
+      "step": 2771
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.9941884279251099,
+      "learning_rate": 0.00021593897295956336,
+      "loss": 0.2206,
+      "step": 2772
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.2422287464141846,
+      "learning_rate": 0.00021590176134954103,
+      "loss": 0.2301,
+      "step": 2773
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5298879146575928,
+      "learning_rate": 0.0002158645497395187,
+      "loss": 0.2831,
+      "step": 2774
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.899098873138428,
+      "learning_rate": 0.00021582733812949639,
+      "loss": 2.1931,
+      "step": 2775
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.338663101196289,
+      "learning_rate": 0.00021579012651947406,
+      "loss": 0.8356,
+      "step": 2776
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9031519889831543,
+      "learning_rate": 0.00021575291490945174,
+      "loss": 1.0787,
+      "step": 2777
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.205125093460083,
+      "learning_rate": 0.00021571570329942939,
+      "loss": 0.9548,
+      "step": 2778
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.117971181869507,
+      "learning_rate": 0.0002156784916894071,
+      "loss": 0.4246,
+      "step": 2779
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.159374713897705,
+      "learning_rate": 0.00021564128007938474,
+      "loss": 0.3248,
+      "step": 2780
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.537348508834839,
+      "learning_rate": 0.00021560406846936241,
+      "loss": 0.7855,
+      "step": 2781
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.984494924545288,
+      "learning_rate": 0.00021556685685934012,
+      "loss": 0.6787,
+      "step": 2782
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.4059112071990967,
+      "learning_rate": 0.00021552964524931777,
+      "loss": 0.5901,
+      "step": 2783
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.266899824142456,
+      "learning_rate": 0.00021549243363929544,
+      "loss": 0.4769,
+      "step": 2784
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.490513563156128,
+      "learning_rate": 0.0002154552220292731,
+      "loss": 0.5872,
+      "step": 2785
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.9180526733398438,
+      "learning_rate": 0.0002154180104192508,
+      "loss": 0.3963,
+      "step": 2786
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.8597452640533447,
+      "learning_rate": 0.00021538079880922847,
+      "loss": 0.5766,
+      "step": 2787
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.8061572313308716,
+      "learning_rate": 0.00021534358719920612,
+      "loss": 0.4378,
+      "step": 2788
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.9399218559265137,
+      "learning_rate": 0.00021530637558918382,
+      "loss": 0.5846,
+      "step": 2789
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.21833872795105,
+      "learning_rate": 0.00021526916397916147,
+      "loss": 0.424,
+      "step": 2790
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.508014678955078,
+      "learning_rate": 0.00021523195236913914,
+      "loss": 0.3813,
+      "step": 2791
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.099500060081482,
+      "learning_rate": 0.00021519474075911685,
+      "loss": 0.2054,
+      "step": 2792
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.6345198154449463,
+      "learning_rate": 0.0002151575291490945,
+      "loss": 0.537,
+      "step": 2793
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.041452646255493,
+      "learning_rate": 0.00021512031753907217,
+      "loss": 0.5931,
+      "step": 2794
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.204749822616577,
+      "learning_rate": 0.00021508310592904985,
+      "loss": 0.418,
+      "step": 2795
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.116248369216919,
+      "learning_rate": 0.00021504589431902752,
+      "loss": 0.4851,
+      "step": 2796
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.110528469085693,
+      "learning_rate": 0.0002150086827090052,
+      "loss": 0.4382,
+      "step": 2797
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.8598759174346924,
+      "learning_rate": 0.00021497147109898285,
+      "loss": 0.542,
+      "step": 2798
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.3355400562286377,
+      "learning_rate": 0.00021493425948896052,
+      "loss": 0.3713,
+      "step": 2799
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 5.150740623474121,
+      "learning_rate": 0.00021489704787893823,
+      "loss": 2.0462,
+      "step": 2800
+    },
+    {
+      "epoch": 0.98,
+      "eval_loss": 0.5526766180992126,
+      "eval_runtime": 51.4541,
+      "eval_samples_per_second": 42.135,
+      "eval_steps_per_second": 10.534,
+      "eval_wer": 0.49014011416709913,
+      "step": 2800
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.123936414718628,
+      "learning_rate": 0.00021485983626891588,
+      "loss": 0.9954,
+      "step": 2801
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 25.905872344970703,
+      "learning_rate": 0.00021482262465889355,
+      "loss": 5.7803,
+      "step": 2802
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5377426147460938,
+      "learning_rate": 0.0002147854130488712,
+      "loss": 0.5039,
+      "step": 2803
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.8195780515670776,
+      "learning_rate": 0.0002147482014388489,
+      "loss": 0.8649,
+      "step": 2804
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.3947935104370117,
+      "learning_rate": 0.00021471098982882658,
+      "loss": 0.7395,
+      "step": 2805
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.462125062942505,
+      "learning_rate": 0.00021467377821880423,
+      "loss": 0.3764,
+      "step": 2806
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.7861380577087402,
+      "learning_rate": 0.00021463656660878193,
+      "loss": 1.1175,
+      "step": 2807
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.9747540950775146,
+      "learning_rate": 0.0002145993549987596,
+      "loss": 0.8438,
+      "step": 2808
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.355682611465454,
+      "learning_rate": 0.00021456214338873725,
+      "loss": 0.6275,
+      "step": 2809
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.986989140510559,
+      "learning_rate": 0.00021452493177871496,
+      "loss": 0.4578,
+      "step": 2810
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.2945938110351562,
+      "learning_rate": 0.0002144877201686926,
+      "loss": 0.4254,
+      "step": 2811
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7034192085266113,
+      "learning_rate": 0.00021445050855867028,
+      "loss": 0.8129,
+      "step": 2812
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.6484246253967285,
+      "learning_rate": 0.00021441329694864798,
+      "loss": 0.5471,
+      "step": 2813
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 9.947596549987793,
+      "learning_rate": 0.00021437608533862563,
+      "loss": 2.901,
+      "step": 2814
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.3516814708709717,
+      "learning_rate": 0.0002143388737286033,
+      "loss": 0.3611,
+      "step": 2815
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.094783067703247,
+      "learning_rate": 0.00021430166211858096,
+      "loss": 0.4288,
+      "step": 2816
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.745293617248535,
+      "learning_rate": 0.00021426445050855866,
+      "loss": 0.4119,
+      "step": 2817
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.0576019287109375,
+      "learning_rate": 0.00021422723889853634,
+      "loss": 0.3339,
+      "step": 2818
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.2697802782058716,
+      "learning_rate": 0.00021419002728851399,
+      "loss": 0.1731,
+      "step": 2819
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.58626127243042,
+      "learning_rate": 0.00021415281567849166,
+      "loss": 0.4412,
+      "step": 2820
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.652714490890503,
+      "learning_rate": 0.00021411560406846936,
+      "loss": 0.2896,
+      "step": 2821
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.500732660293579,
+      "learning_rate": 0.000214078392458447,
+      "loss": 0.2454,
+      "step": 2822
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.2118359804153442,
+      "learning_rate": 0.0002140411808484247,
+      "loss": 0.0464,
+      "step": 2823
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.111560106277466,
+      "learning_rate": 0.00021400396923840236,
+      "loss": 0.3128,
+      "step": 2824
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.749283790588379,
+      "learning_rate": 0.00021396675762838004,
+      "loss": 0.9602,
+      "step": 2825
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.080014228820801,
+      "learning_rate": 0.00021392954601835772,
+      "loss": 0.9871,
+      "step": 2826
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.2783267498016357,
+      "learning_rate": 0.00021389233440833536,
+      "loss": 0.6355,
+      "step": 2827
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.5478532314300537,
+      "learning_rate": 0.00021385512279831307,
+      "loss": 0.3404,
+      "step": 2828
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.262725353240967,
+      "learning_rate": 0.00021381791118829072,
+      "loss": 0.3556,
+      "step": 2829
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.2624759674072266,
+      "learning_rate": 0.0002137806995782684,
+      "loss": 0.4647,
+      "step": 2830
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.1995623111724854,
+      "learning_rate": 0.0002137434879682461,
+      "loss": 0.4685,
+      "step": 2831
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.1879045963287354,
+      "learning_rate": 0.00021370627635822374,
+      "loss": 0.5635,
+      "step": 2832
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.52578604221344,
+      "learning_rate": 0.00021366906474820142,
+      "loss": 0.3173,
+      "step": 2833
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.537405490875244,
+      "learning_rate": 0.00021363185313817907,
+      "loss": 0.7481,
+      "step": 2834
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.3475868701934814,
+      "learning_rate": 0.00021359464152815677,
+      "loss": 0.2276,
+      "step": 2835
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.945054769515991,
+      "learning_rate": 0.00021355742991813445,
+      "loss": 0.7077,
+      "step": 2836
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.155667304992676,
+      "learning_rate": 0.0002135202183081121,
+      "loss": 0.4307,
+      "step": 2837
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.234096527099609,
+      "learning_rate": 0.0002134830066980898,
+      "loss": 0.3034,
+      "step": 2838
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.2338759899139404,
+      "learning_rate": 0.00021344579508806747,
+      "loss": 0.4335,
+      "step": 2839
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.541741371154785,
+      "learning_rate": 0.00021340858347804512,
+      "loss": 1.6635,
+      "step": 2840
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 5.528841495513916,
+      "learning_rate": 0.0002133713718680228,
+      "loss": 2.2904,
+      "step": 2841
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.8137770891189575,
+      "learning_rate": 0.00021333416025800047,
+      "loss": 0.2749,
+      "step": 2842
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.021759033203125,
+      "learning_rate": 0.00021329694864797815,
+      "loss": 0.2995,
+      "step": 2843
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.046627044677734,
+      "learning_rate": 0.00021325973703795583,
+      "loss": 0.5484,
+      "step": 2844
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.797898292541504,
+      "learning_rate": 0.0002132225254279335,
+      "loss": 0.5586,
+      "step": 2845
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.8933186531066895,
+      "learning_rate": 0.00021318531381791118,
+      "loss": 0.2478,
+      "step": 2846
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.601649045944214,
+      "learning_rate": 0.00021314810220788883,
+      "loss": 0.2398,
+      "step": 2847
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.6861727237701416,
+      "learning_rate": 0.0002131108905978665,
+      "loss": 0.3503,
+      "step": 2848
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.729224681854248,
+      "learning_rate": 0.0002130736789878442,
+      "loss": 0.5147,
+      "step": 2849
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.694298267364502,
+      "learning_rate": 0.00021303646737782185,
+      "loss": 1.8265,
+      "step": 2850
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.0727405548095703,
+      "learning_rate": 0.00021299925576779953,
+      "loss": 0.7055,
+      "step": 2851
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.8349599838256836,
+      "learning_rate": 0.00021296204415777723,
+      "loss": 0.373,
+      "step": 2852
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.7524197101593018,
+      "learning_rate": 0.00021292483254775488,
+      "loss": 0.3642,
+      "step": 2853
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.7273423671722412,
+      "learning_rate": 0.00021288762093773256,
+      "loss": 0.4562,
+      "step": 2854
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.9599108695983887,
+      "learning_rate": 0.0002128504093277102,
+      "loss": 0.8411,
+      "step": 2855
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.2834584712982178,
+      "learning_rate": 0.0002128131977176879,
+      "loss": 0.6458,
+      "step": 2856
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.6620326042175293,
+      "learning_rate": 0.00021277598610766558,
+      "loss": 0.7874,
+      "step": 2857
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.7331115007400513,
+      "learning_rate": 0.00021273877449764323,
+      "loss": 0.4673,
+      "step": 2858
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.6492310762405396,
+      "learning_rate": 0.00021270156288762094,
+      "loss": 0.4048,
+      "step": 2859
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.8001649379730225,
+      "learning_rate": 0.00021266435127759858,
+      "loss": 0.5527,
+      "step": 2860
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.7418673038482666,
+      "learning_rate": 0.00021262713966757626,
+      "loss": 0.8918,
+      "step": 2861
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.890509843826294,
+      "learning_rate": 0.00021258992805755394,
+      "loss": 0.4264,
+      "step": 2862
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.927704334259033,
+      "learning_rate": 0.0002125527164475316,
+      "loss": 0.7042,
+      "step": 2863
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.7126057147979736,
+      "learning_rate": 0.0002125155048375093,
+      "loss": 0.4692,
+      "step": 2864
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.3469655513763428,
+      "learning_rate": 0.00021247829322748696,
+      "loss": 0.3412,
+      "step": 2865
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.2996063232421875,
+      "learning_rate": 0.00021244108161746464,
+      "loss": 0.2926,
+      "step": 2866
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.5315076112747192,
+      "learning_rate": 0.00021240387000744232,
+      "loss": 0.1615,
+      "step": 2867
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.6261684894561768,
+      "learning_rate": 0.00021236665839741996,
+      "loss": 0.7396,
+      "step": 2868
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.5153400897979736,
+      "learning_rate": 0.00021232944678739764,
+      "loss": 0.2164,
+      "step": 2869
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 3.3273916244506836,
+      "learning_rate": 0.00021229223517737534,
+      "loss": 0.6213,
+      "step": 2870
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 4.1875104904174805,
+      "learning_rate": 0.000212255023567353,
+      "loss": 0.7009,
+      "step": 2871
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.51561439037323,
+      "learning_rate": 0.00021221781195733067,
+      "loss": 0.1144,
+      "step": 2872
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 2.509672164916992,
+      "learning_rate": 0.00021218060034730832,
+      "loss": 0.3553,
+      "step": 2873
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.6920043230056763,
+      "learning_rate": 0.00021214338873728602,
+      "loss": 0.1711,
+      "step": 2874
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 3.063408136367798,
+      "learning_rate": 0.0002121061771272637,
+      "loss": 0.195,
+      "step": 2875
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 4.922022819519043,
+      "learning_rate": 0.00021206896551724134,
+      "loss": 0.6325,
+      "step": 2876
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 3.254242181777954,
+      "learning_rate": 0.00021203175390721905,
+      "loss": 0.4139,
+      "step": 2877
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.8609694242477417,
+      "learning_rate": 0.00021199454229719672,
+      "loss": 0.3271,
+      "step": 2878
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 4.9110283851623535,
+      "learning_rate": 0.00021195733068717437,
+      "loss": 0.4979,
+      "step": 2879
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 2.849888324737549,
+      "learning_rate": 0.00021192011907715207,
+      "loss": 0.9667,
+      "step": 2880
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 3.0285425186157227,
+      "learning_rate": 0.00021188290746712972,
+      "loss": 0.6187,
+      "step": 2881
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 2.866227388381958,
+      "learning_rate": 0.0002118456958571074,
+      "loss": 0.5702,
+      "step": 2882
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 2.6997718811035156,
+      "learning_rate": 0.00021180848424708507,
+      "loss": 0.5016,
+      "step": 2883
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.718056559562683,
+      "learning_rate": 0.00021177127263706275,
+      "loss": 0.289,
+      "step": 2884
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 3.1275880336761475,
+      "learning_rate": 0.00021173406102704043,
+      "loss": 0.376,
+      "step": 2885
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 4.316718101501465,
+      "learning_rate": 0.00021169684941701807,
+      "loss": 0.8418,
+      "step": 2886
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 2.493377685546875,
+      "learning_rate": 0.00021165963780699578,
+      "loss": 0.2839,
+      "step": 2887
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 2.5852391719818115,
+      "learning_rate": 0.00021162242619697345,
+      "loss": 0.4074,
+      "step": 2888
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 2.4773120880126953,
+      "learning_rate": 0.0002115852145869511,
+      "loss": 0.4167,
+      "step": 2889
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.7587331533432007,
+      "learning_rate": 0.00021154800297692878,
+      "loss": 0.2211,
+      "step": 2890
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 2.402329444885254,
+      "learning_rate": 0.00021151079136690645,
+      "loss": 0.2386,
+      "step": 2891
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 2.4426639080047607,
+      "learning_rate": 0.00021147357975688413,
+      "loss": 0.5694,
+      "step": 2892
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.6130284070968628,
+      "learning_rate": 0.0002114363681468618,
+      "loss": 0.3316,
+      "step": 2893
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 2.9390599727630615,
+      "learning_rate": 0.00021139915653683945,
+      "loss": 0.2903,
+      "step": 2894
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 3.764599561691284,
+      "learning_rate": 0.00021136194492681716,
+      "loss": 0.5749,
+      "step": 2895
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.7966651916503906,
+      "learning_rate": 0.00021132473331679483,
+      "loss": 0.1905,
+      "step": 2896
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 2.91357684135437,
+      "learning_rate": 0.00021128752170677248,
+      "loss": 0.9233,
+      "step": 2897
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 2.6174521446228027,
+      "learning_rate": 0.00021125031009675018,
+      "loss": 0.3764,
+      "step": 2898
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 5.59943151473999,
+      "learning_rate": 0.00021121309848672783,
+      "loss": 0.5162,
+      "step": 2899
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 6.940462589263916,
+      "learning_rate": 0.0002111758868767055,
+      "loss": 0.2316,
+      "step": 2900
+    },
+    {
+      "epoch": 1.02,
+      "eval_loss": 0.6327888369560242,
+      "eval_runtime": 51.2836,
+      "eval_samples_per_second": 42.275,
+      "eval_steps_per_second": 10.569,
+      "eval_wer": 0.5484345268984605,
+      "step": 2900
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 2.0894784927368164,
+      "learning_rate": 0.0002111386752666832,
+      "loss": 0.1064,
+      "step": 2901
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 3.177110195159912,
+      "learning_rate": 0.00021110146365666086,
+      "loss": 0.3558,
+      "step": 2902
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 4.781301021575928,
+      "learning_rate": 0.00021106425204663854,
+      "loss": 0.9566,
+      "step": 2903
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.9870142936706543,
+      "learning_rate": 0.00021102704043661618,
+      "loss": 0.3168,
+      "step": 2904
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 3.021113395690918,
+      "learning_rate": 0.0002109898288265939,
+      "loss": 1.3912,
+      "step": 2905
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.378859519958496,
+      "learning_rate": 0.00021095261721657156,
+      "loss": 0.6019,
+      "step": 2906
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 2.121788740158081,
+      "learning_rate": 0.0002109154056065492,
+      "loss": 0.6099,
+      "step": 2907
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.3554753065109253,
+      "learning_rate": 0.00021087819399652691,
+      "loss": 0.2695,
+      "step": 2908
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 2.836920738220215,
+      "learning_rate": 0.0002108409823865046,
+      "loss": 0.4332,
+      "step": 2909
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 2.007983922958374,
+      "learning_rate": 0.00021080377077648224,
+      "loss": 0.3614,
+      "step": 2910
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 2.3177947998046875,
+      "learning_rate": 0.00021076655916645991,
+      "loss": 0.4861,
+      "step": 2911
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 2.913270950317383,
+      "learning_rate": 0.0002107293475564376,
+      "loss": 0.5023,
+      "step": 2912
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 5.127606391906738,
+      "learning_rate": 0.00021069213594641527,
+      "loss": 0.8786,
+      "step": 2913
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 2.110982894897461,
+      "learning_rate": 0.00021065492433639294,
+      "loss": 0.4902,
+      "step": 2914
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 4.175901889801025,
+      "learning_rate": 0.0002106177127263706,
+      "loss": 0.4782,
+      "step": 2915
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 2.3658077716827393,
+      "learning_rate": 0.0002105805011163483,
+      "loss": 0.3401,
+      "step": 2916
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 3.592463970184326,
+      "learning_rate": 0.00021054328950632594,
+      "loss": 0.882,
+      "step": 2917
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 2.3057944774627686,
+      "learning_rate": 0.00021050607789630362,
+      "loss": 0.5945,
+      "step": 2918
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 3.1992056369781494,
+      "learning_rate": 0.00021046886628628132,
+      "loss": 0.7865,
+      "step": 2919
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 5.556008815765381,
+      "learning_rate": 0.00021043165467625897,
+      "loss": 0.8176,
+      "step": 2920
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 4.94827127456665,
+      "learning_rate": 0.00021039444306623665,
+      "loss": 0.3772,
+      "step": 2921
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 2.1931419372558594,
+      "learning_rate": 0.00021035723145621435,
+      "loss": 0.1395,
+      "step": 2922
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 2.6224441528320312,
+      "learning_rate": 0.000210320019846192,
+      "loss": 0.3006,
+      "step": 2923
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 14.6669282913208,
+      "learning_rate": 0.00021028280823616967,
+      "loss": 3.7282,
+      "step": 2924
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 2.3192381858825684,
+      "learning_rate": 0.00021024559662614732,
+      "loss": 0.4179,
+      "step": 2925
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.563372254371643,
+      "learning_rate": 0.00021020838501612502,
+      "loss": 0.1219,
+      "step": 2926
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.8706691265106201,
+      "learning_rate": 0.0002101711734061027,
+      "loss": 0.1685,
+      "step": 2927
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 2.756892204284668,
+      "learning_rate": 0.00021013396179608035,
+      "loss": 0.2088,
+      "step": 2928
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 4.2238030433654785,
+      "learning_rate": 0.00021009675018605805,
+      "loss": 1.3,
+      "step": 2929
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 4.2373576164245605,
+      "learning_rate": 0.0002100595385760357,
+      "loss": 0.8429,
+      "step": 2930
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.899154782295227,
+      "learning_rate": 0.00021002232696601338,
+      "loss": 0.524,
+      "step": 2931
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 2.224109411239624,
+      "learning_rate": 0.00020998511535599105,
+      "loss": 0.6571,
+      "step": 2932
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 3.5380167961120605,
+      "learning_rate": 0.00020994790374596873,
+      "loss": 0.998,
+      "step": 2933
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.9109605550765991,
+      "learning_rate": 0.0002099106921359464,
+      "loss": 0.2862,
+      "step": 2934
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 3.0752928256988525,
+      "learning_rate": 0.00020987348052592405,
+      "loss": 0.4658,
+      "step": 2935
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 2.45890212059021,
+      "learning_rate": 0.00020983626891590173,
+      "loss": 0.7092,
+      "step": 2936
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 3.076648235321045,
+      "learning_rate": 0.00020979905730587943,
+      "loss": 0.6868,
+      "step": 2937
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 2.497008800506592,
+      "learning_rate": 0.00020976184569585708,
+      "loss": 0.2683,
+      "step": 2938
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.9469985961914062,
+      "learning_rate": 0.00020972463408583476,
+      "loss": 0.2959,
+      "step": 2939
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.4684535264968872,
+      "learning_rate": 0.00020968742247581246,
+      "loss": 0.1444,
+      "step": 2940
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 2.822787046432495,
+      "learning_rate": 0.0002096502108657901,
+      "loss": 0.4749,
+      "step": 2941
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 2.6502792835235596,
+      "learning_rate": 0.00020961299925576778,
+      "loss": 0.5122,
+      "step": 2942
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 9.322734832763672,
+      "learning_rate": 0.00020957578764574543,
+      "loss": 2.2989,
+      "step": 2943
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 2.103727340698242,
+      "learning_rate": 0.00020953857603572313,
+      "loss": 0.444,
+      "step": 2944
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 3.0473310947418213,
+      "learning_rate": 0.0002095013644257008,
+      "loss": 0.4347,
+      "step": 2945
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.7083921432495117,
+      "learning_rate": 0.00020946415281567846,
+      "loss": 0.3226,
+      "step": 2946
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 3.072146415710449,
+      "learning_rate": 0.00020942694120565616,
+      "loss": 0.5612,
+      "step": 2947
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 2.4064691066741943,
+      "learning_rate": 0.0002093897295956338,
+      "loss": 0.1788,
+      "step": 2948
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.3841482400894165,
+      "learning_rate": 0.00020935251798561149,
+      "loss": 0.1171,
+      "step": 2949
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 2.551501989364624,
+      "learning_rate": 0.0002093153063755892,
+      "loss": 0.616,
+      "step": 2950
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 3.9594438076019287,
+      "learning_rate": 0.00020927809476556684,
+      "loss": 0.7705,
+      "step": 2951
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 4.557310104370117,
+      "learning_rate": 0.00020924088315554451,
+      "loss": 1.3497,
+      "step": 2952
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 2.093480348587036,
+      "learning_rate": 0.0002092036715455222,
+      "loss": 0.2775,
+      "step": 2953
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 3.45406174659729,
+      "learning_rate": 0.00020916645993549987,
+      "loss": 0.619,
+      "step": 2954
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 2.039518356323242,
+      "learning_rate": 0.00020912924832547754,
+      "loss": 0.9185,
+      "step": 2955
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 2.3017618656158447,
+      "learning_rate": 0.0002090920367154552,
+      "loss": 0.5944,
+      "step": 2956
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.7939972877502441,
+      "learning_rate": 0.00020905482510543287,
+      "loss": 0.5319,
+      "step": 2957
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 29.27108383178711,
+      "learning_rate": 0.00020901761349541057,
+      "loss": 5.4024,
+      "step": 2958
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 3.3019933700561523,
+      "learning_rate": 0.00020898040188538822,
+      "loss": 0.8981,
+      "step": 2959
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 2.4643125534057617,
+      "learning_rate": 0.0002089431902753659,
+      "loss": 0.8616,
+      "step": 2960
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 2.9230358600616455,
+      "learning_rate": 0.00020890597866534354,
+      "loss": 0.5856,
+      "step": 2961
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.3859621286392212,
+      "learning_rate": 0.00020886876705532124,
+      "loss": 0.3128,
+      "step": 2962
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 3.490842342376709,
+      "learning_rate": 0.00020883155544529892,
+      "loss": 0.4453,
+      "step": 2963
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 2.791914701461792,
+      "learning_rate": 0.00020879434383527657,
+      "loss": 0.6412,
+      "step": 2964
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 2.955834150314331,
+      "learning_rate": 0.00020875713222525427,
+      "loss": 0.3226,
+      "step": 2965
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.8695831298828125,
+      "learning_rate": 0.00020871992061523195,
+      "loss": 0.3433,
+      "step": 2966
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 2.185141086578369,
+      "learning_rate": 0.0002086827090052096,
+      "loss": 0.2187,
+      "step": 2967
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 2.4184865951538086,
+      "learning_rate": 0.0002086454973951873,
+      "loss": 0.2672,
+      "step": 2968
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 7.882993221282959,
+      "learning_rate": 0.00020860828578516495,
+      "loss": 0.2307,
+      "step": 2969
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.4877923727035522,
+      "learning_rate": 0.00020857107417514262,
+      "loss": 0.1831,
+      "step": 2970
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.8400189876556396,
+      "learning_rate": 0.00020853386256512033,
+      "loss": 0.3421,
+      "step": 2971
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.835120439529419,
+      "learning_rate": 0.00020849665095509798,
+      "loss": 0.2359,
+      "step": 2972
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 2.832714080810547,
+      "learning_rate": 0.00020845943934507565,
+      "loss": 0.4349,
+      "step": 2973
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.8223691582679749,
+      "learning_rate": 0.0002084222277350533,
+      "loss": 0.039,
+      "step": 2974
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 3.099144697189331,
+      "learning_rate": 0.000208385016125031,
+      "loss": 0.5629,
+      "step": 2975
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 2.728029251098633,
+      "learning_rate": 0.00020834780451500868,
+      "loss": 0.3675,
+      "step": 2976
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 2.1751224994659424,
+      "learning_rate": 0.00020831059290498633,
+      "loss": 0.1833,
+      "step": 2977
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 8.328917503356934,
+      "learning_rate": 0.000208273381294964,
+      "loss": 0.5198,
+      "step": 2978
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 6.681622505187988,
+      "learning_rate": 0.00020823616968494168,
+      "loss": 2.3852,
+      "step": 2979
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 3.279292583465576,
+      "learning_rate": 0.00020819895807491935,
+      "loss": 1.1333,
+      "step": 2980
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.996398687362671,
+      "learning_rate": 0.00020816174646489703,
+      "loss": 0.5301,
+      "step": 2981
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.8816949129104614,
+      "learning_rate": 0.00020812453485487468,
+      "loss": 0.4437,
+      "step": 2982
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 2.7721152305603027,
+      "learning_rate": 0.00020808732324485238,
+      "loss": 0.7248,
+      "step": 2983
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.3053263425827026,
+      "learning_rate": 0.00020805011163483006,
+      "loss": 0.1844,
+      "step": 2984
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.6006126403808594,
+      "learning_rate": 0.0002080129000248077,
+      "loss": 0.3389,
+      "step": 2985
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 6.084453582763672,
+      "learning_rate": 0.0002079756884147854,
+      "loss": 2.0407,
+      "step": 2986
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.4119192361831665,
+      "learning_rate": 0.00020793847680476306,
+      "loss": 0.2708,
+      "step": 2987
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 2.911585807800293,
+      "learning_rate": 0.00020790126519474073,
+      "loss": 0.3626,
+      "step": 2988
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.2379357814788818,
+      "learning_rate": 0.00020786405358471844,
+      "loss": 0.1421,
+      "step": 2989
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.679255485534668,
+      "learning_rate": 0.00020782684197469609,
+      "loss": 0.2169,
+      "step": 2990
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.825289011001587,
+      "learning_rate": 0.00020778963036467376,
+      "loss": 0.1638,
+      "step": 2991
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 2.66688871383667,
+      "learning_rate": 0.0002077524187546514,
+      "loss": 0.6995,
+      "step": 2992
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 3.020879030227661,
+      "learning_rate": 0.0002077152071446291,
+      "loss": 0.4142,
+      "step": 2993
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 2.3098480701446533,
+      "learning_rate": 0.0002076779955346068,
+      "loss": 0.371,
+      "step": 2994
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 4.051997661590576,
+      "learning_rate": 0.00020764078392458444,
+      "loss": 0.4462,
+      "step": 2995
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 2.9029579162597656,
+      "learning_rate": 0.00020760357231456214,
+      "loss": 0.3817,
+      "step": 2996
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 5.890556812286377,
+      "learning_rate": 0.00020756636070453982,
+      "loss": 1.261,
+      "step": 2997
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 2.089731216430664,
+      "learning_rate": 0.00020752914909451746,
+      "loss": 0.3077,
+      "step": 2998
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 4.995520114898682,
+      "learning_rate": 0.00020749193748449514,
+      "loss": 1.0239,
+      "step": 2999
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 4.084024429321289,
+      "learning_rate": 0.00020745472587447282,
+      "loss": 0.6881,
+      "step": 3000
+    },
+    {
+      "epoch": 1.05,
+      "eval_loss": 0.6195849776268005,
+      "eval_runtime": 50.8747,
+      "eval_samples_per_second": 42.615,
+      "eval_steps_per_second": 10.654,
+      "eval_wer": 0.5003459609064176,
+      "step": 3000
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 7.645015239715576,
+      "learning_rate": 0.0002074175142644505,
+      "loss": 0.9345,
+      "step": 3001
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 3.707676887512207,
+      "learning_rate": 0.00020738030265442817,
+      "loss": 0.3963,
+      "step": 3002
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 2.314399003982544,
+      "learning_rate": 0.00020734309104440582,
+      "loss": 0.2629,
+      "step": 3003
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 5.601327896118164,
+      "learning_rate": 0.00020730587943438352,
+      "loss": 0.5629,
+      "step": 3004
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 3.5226032733917236,
+      "learning_rate": 0.00020726866782436117,
+      "loss": 1.5056,
+      "step": 3005
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.585764765739441,
+      "learning_rate": 0.00020723145621433884,
+      "loss": 0.4755,
+      "step": 3006
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.83793044090271,
+      "learning_rate": 0.00020719424460431655,
+      "loss": 0.5136,
+      "step": 3007
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 2.307407855987549,
+      "learning_rate": 0.0002071570329942942,
+      "loss": 0.5605,
+      "step": 3008
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.9383962154388428,
+      "learning_rate": 0.00020711982138427187,
+      "loss": 0.2642,
+      "step": 3009
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 5.103401184082031,
+      "learning_rate": 0.00020708260977424957,
+      "loss": 0.685,
+      "step": 3010
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 2.193871259689331,
+      "learning_rate": 0.00020704539816422722,
+      "loss": 0.363,
+      "step": 3011
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 2.350487232208252,
+      "learning_rate": 0.0002070081865542049,
+      "loss": 0.4608,
+      "step": 3012
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.3571010828018188,
+      "learning_rate": 0.00020697097494418255,
+      "loss": 0.1649,
+      "step": 3013
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 2.3821628093719482,
+      "learning_rate": 0.00020693376333416025,
+      "loss": 0.2691,
+      "step": 3014
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.663122296333313,
+      "learning_rate": 0.00020689655172413793,
+      "loss": 0.3526,
+      "step": 3015
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 3.3357670307159424,
+      "learning_rate": 0.00020685934011411557,
+      "loss": 0.3898,
+      "step": 3016
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 4.479585647583008,
+      "learning_rate": 0.00020682212850409328,
+      "loss": 0.2298,
+      "step": 3017
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 3.3683953285217285,
+      "learning_rate": 0.00020678491689407093,
+      "loss": 0.4032,
+      "step": 3018
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 12.573981285095215,
+      "learning_rate": 0.0002067477052840486,
+      "loss": 3.1628,
+      "step": 3019
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 2.957279682159424,
+      "learning_rate": 0.00020671049367402628,
+      "loss": 0.6773,
+      "step": 3020
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 2.875685453414917,
+      "learning_rate": 0.00020667328206400395,
+      "loss": 0.6258,
+      "step": 3021
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 4.08288049697876,
+      "learning_rate": 0.00020663607045398163,
+      "loss": 1.0724,
+      "step": 3022
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.861497163772583,
+      "learning_rate": 0.00020659885884395928,
+      "loss": 0.0636,
+      "step": 3023
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 3.3727078437805176,
+      "learning_rate": 0.00020656164723393695,
+      "loss": 0.5172,
+      "step": 3024
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 3.8272769451141357,
+      "learning_rate": 0.00020652443562391466,
+      "loss": 0.6818,
+      "step": 3025
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.6726584434509277,
+      "learning_rate": 0.0002064872240138923,
+      "loss": 0.0578,
+      "step": 3026
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 4.285344123840332,
+      "learning_rate": 0.00020645001240386998,
+      "loss": 0.7979,
+      "step": 3027
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 7.106195449829102,
+      "learning_rate": 0.00020641280079384768,
+      "loss": 3.3654,
+      "step": 3028
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.9665549993515015,
+      "learning_rate": 0.00020637558918382533,
+      "loss": 0.1718,
+      "step": 3029
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.6774803400039673,
+      "learning_rate": 0.000206338377573803,
+      "loss": 0.6851,
+      "step": 3030
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 2.010876417160034,
+      "learning_rate": 0.00020630116596378066,
+      "loss": 0.4386,
+      "step": 3031
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.4450249671936035,
+      "learning_rate": 0.00020626395435375836,
+      "loss": 0.3772,
+      "step": 3032
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 2.3104891777038574,
+      "learning_rate": 0.00020622674274373604,
+      "loss": 0.3893,
+      "step": 3033
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.3111859560012817,
+      "learning_rate": 0.00020618953113371368,
+      "loss": 0.2572,
+      "step": 3034
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 2.2951879501342773,
+      "learning_rate": 0.0002061523195236914,
+      "loss": 0.3824,
+      "step": 3035
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 2.6300010681152344,
+      "learning_rate": 0.00020611510791366904,
+      "loss": 0.5777,
+      "step": 3036
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 3.6638519763946533,
+      "learning_rate": 0.0002060778963036467,
+      "loss": 1.1108,
+      "step": 3037
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 2.498910427093506,
+      "learning_rate": 0.00020604068469362442,
+      "loss": 0.3768,
+      "step": 3038
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 2.4238033294677734,
+      "learning_rate": 0.00020600347308360206,
+      "loss": 0.5088,
+      "step": 3039
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.8531930446624756,
+      "learning_rate": 0.00020596626147357974,
+      "loss": 0.2276,
+      "step": 3040
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 2.1931774616241455,
+      "learning_rate": 0.00020592904986355742,
+      "loss": 0.3416,
+      "step": 3041
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.7940934896469116,
+      "learning_rate": 0.0002058918382535351,
+      "loss": 0.2504,
+      "step": 3042
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.3244794607162476,
+      "learning_rate": 0.00020585462664351277,
+      "loss": 0.1899,
+      "step": 3043
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 2.233159303665161,
+      "learning_rate": 0.00020581741503349042,
+      "loss": 0.4307,
+      "step": 3044
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 4.452759265899658,
+      "learning_rate": 0.0002057802034234681,
+      "loss": 1.5291,
+      "step": 3045
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 6.447901248931885,
+      "learning_rate": 0.0002057429918134458,
+      "loss": 0.3252,
+      "step": 3046
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 6.9542765617370605,
+      "learning_rate": 0.00020570578020342344,
+      "loss": 0.2644,
+      "step": 3047
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 3.4594507217407227,
+      "learning_rate": 0.00020566856859340112,
+      "loss": 0.6206,
+      "step": 3048
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 4.8815388679504395,
+      "learning_rate": 0.0002056313569833788,
+      "loss": 1.6332,
+      "step": 3049
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 4.452385902404785,
+      "learning_rate": 0.00020559414537335647,
+      "loss": 0.9805,
+      "step": 3050
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 3.0658645629882812,
+      "learning_rate": 0.00020555693376333415,
+      "loss": 0.5303,
+      "step": 3051
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 4.26853084564209,
+      "learning_rate": 0.0002055197221533118,
+      "loss": 0.3217,
+      "step": 3052
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 2.2705271244049072,
+      "learning_rate": 0.0002054825105432895,
+      "loss": 0.3074,
+      "step": 3053
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 2.20687198638916,
+      "learning_rate": 0.00020544529893326717,
+      "loss": 0.2464,
+      "step": 3054
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 3.222644090652466,
+      "learning_rate": 0.00020540808732324482,
+      "loss": 1.1983,
+      "step": 3055
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.951676368713379,
+      "learning_rate": 0.00020537087571322253,
+      "loss": 0.4934,
+      "step": 3056
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.844739317893982,
+      "learning_rate": 0.00020533366410320017,
+      "loss": 0.4015,
+      "step": 3057
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 2.413743019104004,
+      "learning_rate": 0.00020529645249317785,
+      "loss": 0.4277,
+      "step": 3058
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 2.190391778945923,
+      "learning_rate": 0.00020525924088315555,
+      "loss": 0.3995,
+      "step": 3059
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 3.3840878009796143,
+      "learning_rate": 0.0002052220292731332,
+      "loss": 0.5431,
+      "step": 3060
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 3.4068565368652344,
+      "learning_rate": 0.00020518481766311088,
+      "loss": 0.2978,
+      "step": 3061
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.8028383255004883,
+      "learning_rate": 0.00020514760605308853,
+      "loss": 0.5075,
+      "step": 3062
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 2.067969799041748,
+      "learning_rate": 0.00020511039444306623,
+      "loss": 0.3496,
+      "step": 3063
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 3.3436696529388428,
+      "learning_rate": 0.0002050731828330439,
+      "loss": 0.5107,
+      "step": 3064
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 2.0510003566741943,
+      "learning_rate": 0.00020503597122302155,
+      "loss": 0.3864,
+      "step": 3065
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.5820317268371582,
+      "learning_rate": 0.00020499875961299923,
+      "loss": 0.2344,
+      "step": 3066
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 2.2191803455352783,
+      "learning_rate": 0.00020496154800297693,
+      "loss": 0.493,
+      "step": 3067
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.7818882465362549,
+      "learning_rate": 0.00020492433639295458,
+      "loss": 0.1892,
+      "step": 3068
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 3.912038564682007,
+      "learning_rate": 0.00020488712478293226,
+      "loss": 0.6714,
+      "step": 3069
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 2.960134506225586,
+      "learning_rate": 0.00020484991317290993,
+      "loss": 0.264,
+      "step": 3070
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 2.104323625564575,
+      "learning_rate": 0.0002048127015628876,
+      "loss": 0.5416,
+      "step": 3071
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 4.2121148109436035,
+      "learning_rate": 0.00020477548995286528,
+      "loss": 1.9044,
+      "step": 3072
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 2.713250160217285,
+      "learning_rate": 0.00020473827834284293,
+      "loss": 0.3652,
+      "step": 3073
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.9541451334953308,
+      "learning_rate": 0.00020470106673282064,
+      "loss": 0.0514,
+      "step": 3074
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 9.565132141113281,
+      "learning_rate": 0.00020466385512279828,
+      "loss": 1.9225,
+      "step": 3075
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 3.1412031650543213,
+      "learning_rate": 0.00020462664351277596,
+      "loss": 0.3966,
+      "step": 3076
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.8707399368286133,
+      "learning_rate": 0.00020458943190275366,
+      "loss": 0.2158,
+      "step": 3077
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 2.589046001434326,
+      "learning_rate": 0.0002045522202927313,
+      "loss": 0.3275,
+      "step": 3078
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 4.2253804206848145,
+      "learning_rate": 0.000204515008682709,
+      "loss": 0.3441,
+      "step": 3079
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 3.184797525405884,
+      "learning_rate": 0.00020447779707268664,
+      "loss": 1.0592,
+      "step": 3080
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 2.5358669757843018,
+      "learning_rate": 0.00020444058546266434,
+      "loss": 0.6669,
+      "step": 3081
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 3.5697948932647705,
+      "learning_rate": 0.00020440337385264201,
+      "loss": 0.4798,
+      "step": 3082
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.44380521774292,
+      "learning_rate": 0.00020436616224261966,
+      "loss": 0.4084,
+      "step": 3083
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 12.404438972473145,
+      "learning_rate": 0.00020432895063259737,
+      "loss": 4.2819,
+      "step": 3084
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 2.316073417663574,
+      "learning_rate": 0.00020429173902257504,
+      "loss": 0.6902,
+      "step": 3085
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.3172087669372559,
+      "learning_rate": 0.0002042545274125527,
+      "loss": 0.2745,
+      "step": 3086
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 2.7705490589141846,
+      "learning_rate": 0.00020421731580253037,
+      "loss": 1.3201,
+      "step": 3087
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 2.1392929553985596,
+      "learning_rate": 0.00020418010419250804,
+      "loss": 0.2256,
+      "step": 3088
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 2.1735339164733887,
+      "learning_rate": 0.00020414289258248572,
+      "loss": 0.7259,
+      "step": 3089
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.7704854011535645,
+      "learning_rate": 0.0002041056809724634,
+      "loss": 0.3775,
+      "step": 3090
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 2.1959261894226074,
+      "learning_rate": 0.00020406846936244107,
+      "loss": 0.3028,
+      "step": 3091
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.4941906929016113,
+      "learning_rate": 0.00020403125775241875,
+      "loss": 0.2983,
+      "step": 3092
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.2572664022445679,
+      "learning_rate": 0.0002039940461423964,
+      "loss": 0.1432,
+      "step": 3093
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.722988247871399,
+      "learning_rate": 0.00020395683453237407,
+      "loss": 0.252,
+      "step": 3094
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.260048508644104,
+      "learning_rate": 0.00020391962292235177,
+      "loss": 0.2426,
+      "step": 3095
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.990040898323059,
+      "learning_rate": 0.00020388241131232942,
+      "loss": 0.2344,
+      "step": 3096
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 4.270902156829834,
+      "learning_rate": 0.0002038451997023071,
+      "loss": 0.445,
+      "step": 3097
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 6.141219615936279,
+      "learning_rate": 0.0002038079880922848,
+      "loss": 0.8813,
+      "step": 3098
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 4.40023946762085,
+      "learning_rate": 0.00020377077648226245,
+      "loss": 0.3991,
+      "step": 3099
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 2.3631558418273926,
+      "learning_rate": 0.00020373356487224012,
+      "loss": 0.2522,
+      "step": 3100
+    },
+    {
+      "epoch": 1.09,
+      "eval_loss": 0.5331594347953796,
+      "eval_runtime": 50.9883,
+      "eval_samples_per_second": 42.52,
+      "eval_steps_per_second": 10.63,
+      "eval_wer": 0.46116588825462723,
+      "step": 3100
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 6.387331485748291,
+      "learning_rate": 0.00020369635326221777,
+      "loss": 0.5592,
+      "step": 3101
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 3.025975227355957,
+      "learning_rate": 0.00020365914165219548,
+      "loss": 0.1448,
+      "step": 3102
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 5.364758014678955,
+      "learning_rate": 0.00020362193004217315,
+      "loss": 0.854,
+      "step": 3103
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 3.2311348915100098,
+      "learning_rate": 0.0002035847184321508,
+      "loss": 0.2019,
+      "step": 3104
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 2.2923085689544678,
+      "learning_rate": 0.0002035475068221285,
+      "loss": 0.7306,
+      "step": 3105
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.8082457780838013,
+      "learning_rate": 0.00020351029521210615,
+      "loss": 0.74,
+      "step": 3106
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 3.0178422927856445,
+      "learning_rate": 0.00020347308360208383,
+      "loss": 0.5268,
+      "step": 3107
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.8592593669891357,
+      "learning_rate": 0.0002034358719920615,
+      "loss": 0.3687,
+      "step": 3108
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 2.91344952583313,
+      "learning_rate": 0.00020339866038203918,
+      "loss": 0.3192,
+      "step": 3109
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 4.410944938659668,
+      "learning_rate": 0.00020336144877201686,
+      "loss": 0.3268,
+      "step": 3110
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 2.5860300064086914,
+      "learning_rate": 0.00020332423716199453,
+      "loss": 0.6365,
+      "step": 3111
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 2.230043411254883,
+      "learning_rate": 0.0002032870255519722,
+      "loss": 0.3995,
+      "step": 3112
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 2.6025543212890625,
+      "learning_rate": 0.00020324981394194988,
+      "loss": 0.534,
+      "step": 3113
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 3.757107973098755,
+      "learning_rate": 0.00020321260233192753,
+      "loss": 0.6327,
+      "step": 3114
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.6777108907699585,
+      "learning_rate": 0.0002031753907219052,
+      "loss": 0.1812,
+      "step": 3115
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 2.1241323947906494,
+      "learning_rate": 0.0002031381791118829,
+      "loss": 0.3462,
+      "step": 3116
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 2.4214656352996826,
+      "learning_rate": 0.00020310096750186056,
+      "loss": 0.5252,
+      "step": 3117
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.9669643640518188,
+      "learning_rate": 0.00020306375589183823,
+      "loss": 0.2442,
+      "step": 3118
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 2.5513925552368164,
+      "learning_rate": 0.00020302654428181588,
+      "loss": 0.7533,
+      "step": 3119
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 2.9641880989074707,
+      "learning_rate": 0.00020298933267179359,
+      "loss": 0.5389,
+      "step": 3120
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 2.38806414604187,
+      "learning_rate": 0.00020295212106177126,
+      "loss": 0.1863,
+      "step": 3121
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 3.5284969806671143,
+      "learning_rate": 0.0002029149094517489,
+      "loss": 0.6842,
+      "step": 3122
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 3.640097141265869,
+      "learning_rate": 0.00020287769784172661,
+      "loss": 0.9635,
+      "step": 3123
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 3.8287434577941895,
+      "learning_rate": 0.00020284048623170426,
+      "loss": 1.0185,
+      "step": 3124
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 2.4016950130462646,
+      "learning_rate": 0.00020280327462168194,
+      "loss": 0.3064,
+      "step": 3125
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 4.09964656829834,
+      "learning_rate": 0.00020276606301165964,
+      "loss": 0.3252,
+      "step": 3126
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.654166579246521,
+      "learning_rate": 0.0002027288514016373,
+      "loss": 0.2792,
+      "step": 3127
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 4.5107622146606445,
+      "learning_rate": 0.00020269163979161497,
+      "loss": 0.4426,
+      "step": 3128
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 2.8204357624053955,
+      "learning_rate": 0.00020265442818159264,
+      "loss": 0.2263,
+      "step": 3129
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 2.2634711265563965,
+      "learning_rate": 0.00020261721657157032,
+      "loss": 0.7353,
+      "step": 3130
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 2.142753839492798,
+      "learning_rate": 0.000202580004961548,
+      "loss": 0.6885,
+      "step": 3131
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.8485157489776611,
+      "learning_rate": 0.00020254279335152564,
+      "loss": 0.4566,
+      "step": 3132
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 2.232492685317993,
+      "learning_rate": 0.00020250558174150334,
+      "loss": 0.4872,
+      "step": 3133
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 2.160388231277466,
+      "learning_rate": 0.00020246837013148102,
+      "loss": 0.3397,
+      "step": 3134
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.7792160511016846,
+      "learning_rate": 0.00020243115852145867,
+      "loss": 0.5532,
+      "step": 3135
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.8902424573898315,
+      "learning_rate": 0.00020239394691143634,
+      "loss": 0.3345,
+      "step": 3136
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 2.7055909633636475,
+      "learning_rate": 0.00020235673530141402,
+      "loss": 0.5375,
+      "step": 3137
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.9411323070526123,
+      "learning_rate": 0.0002023195236913917,
+      "loss": 0.3124,
+      "step": 3138
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.9080373048782349,
+      "learning_rate": 0.00020228231208136937,
+      "loss": 0.2328,
+      "step": 3139
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 2.3703057765960693,
+      "learning_rate": 0.00020224510047134702,
+      "loss": 0.4152,
+      "step": 3140
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 3.085244655609131,
+      "learning_rate": 0.00020220788886132472,
+      "loss": 0.6424,
+      "step": 3141
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 4.0253801345825195,
+      "learning_rate": 0.0002021706772513024,
+      "loss": 0.3865,
+      "step": 3142
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 2.0360500812530518,
+      "learning_rate": 0.00020213346564128005,
+      "loss": 0.2447,
+      "step": 3143
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 2.123710870742798,
+      "learning_rate": 0.00020209625403125775,
+      "loss": 0.3863,
+      "step": 3144
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 2.0543999671936035,
+      "learning_rate": 0.0002020590424212354,
+      "loss": 0.3734,
+      "step": 3145
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 2.6033356189727783,
+      "learning_rate": 0.00020202183081121308,
+      "loss": 1.1586,
+      "step": 3146
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 2.208231210708618,
+      "learning_rate": 0.00020198461920119078,
+      "loss": 0.4899,
+      "step": 3147
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 2.316422700881958,
+      "learning_rate": 0.00020194740759116843,
+      "loss": 0.7458,
+      "step": 3148
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 2.228898525238037,
+      "learning_rate": 0.0002019101959811461,
+      "loss": 0.355,
+      "step": 3149
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 3.5317399501800537,
+      "learning_rate": 0.00020187298437112375,
+      "loss": 0.3847,
+      "step": 3150
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 2.578821897506714,
+      "learning_rate": 0.00020183577276110145,
+      "loss": 0.3668,
+      "step": 3151
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 4.172102928161621,
+      "learning_rate": 0.00020179856115107913,
+      "loss": 0.5389,
+      "step": 3152
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 2.7270286083221436,
+      "learning_rate": 0.00020176134954105678,
+      "loss": 0.4602,
+      "step": 3153
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 3.5921061038970947,
+      "learning_rate": 0.00020172413793103448,
+      "loss": 0.3265,
+      "step": 3154
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.7611098289489746,
+      "learning_rate": 0.00020168692632101216,
+      "loss": 0.359,
+      "step": 3155
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 2.818851947784424,
+      "learning_rate": 0.0002016497147109898,
+      "loss": 0.8464,
+      "step": 3156
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 2.597490072250366,
+      "learning_rate": 0.00020161250310096748,
+      "loss": 0.6396,
+      "step": 3157
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 4.247330665588379,
+      "learning_rate": 0.00020157529149094516,
+      "loss": 1.4697,
+      "step": 3158
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 2.6339824199676514,
+      "learning_rate": 0.00020153807988092283,
+      "loss": 0.6861,
+      "step": 3159
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.2723904848098755,
+      "learning_rate": 0.0002015008682709005,
+      "loss": 0.2326,
+      "step": 3160
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 2.37565016746521,
+      "learning_rate": 0.00020146365666087816,
+      "loss": 0.2286,
+      "step": 3161
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 3.4451630115509033,
+      "learning_rate": 0.00020142644505085586,
+      "loss": 0.9374,
+      "step": 3162
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 2.457343101501465,
+      "learning_rate": 0.0002013892334408335,
+      "loss": 0.3735,
+      "step": 3163
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 2.033566951751709,
+      "learning_rate": 0.00020135202183081119,
+      "loss": 0.2974,
+      "step": 3164
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.4532384872436523,
+      "learning_rate": 0.0002013148102207889,
+      "loss": 0.2587,
+      "step": 3165
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 9.672016143798828,
+      "learning_rate": 0.00020127759861076654,
+      "loss": 2.1447,
+      "step": 3166
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 3.0728158950805664,
+      "learning_rate": 0.0002012403870007442,
+      "loss": 0.4088,
+      "step": 3167
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 3.4193506240844727,
+      "learning_rate": 0.00020120317539072186,
+      "loss": 0.6485,
+      "step": 3168
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 2.32649302482605,
+      "learning_rate": 0.00020116596378069956,
+      "loss": 0.5016,
+      "step": 3169
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.4236040115356445,
+      "learning_rate": 0.00020112875217067724,
+      "loss": 0.2539,
+      "step": 3170
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 2.348944902420044,
+      "learning_rate": 0.0002010915405606549,
+      "loss": 0.3252,
+      "step": 3171
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.6811813712120056,
+      "learning_rate": 0.0002010543289506326,
+      "loss": 0.0438,
+      "step": 3172
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 2.597822427749634,
+      "learning_rate": 0.00020101711734061027,
+      "loss": 0.3622,
+      "step": 3173
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.7118090391159058,
+      "learning_rate": 0.00020097990573058792,
+      "loss": 0.2325,
+      "step": 3174
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 2.546022415161133,
+      "learning_rate": 0.00020094269412056562,
+      "loss": 0.2348,
+      "step": 3175
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.8402515649795532,
+      "learning_rate": 0.00020090548251054327,
+      "loss": 0.2659,
+      "step": 3176
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 3.6556687355041504,
+      "learning_rate": 0.00020086827090052094,
+      "loss": 0.2047,
+      "step": 3177
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 3.188159942626953,
+      "learning_rate": 0.00020083105929049862,
+      "loss": 0.3726,
+      "step": 3178
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": NaN,
+      "learning_rate": 0.00020083105929049862,
+      "loss": 0.0813,
+      "step": 3179
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 2.2536139488220215,
+      "learning_rate": 0.0002007938476804763,
+      "loss": 0.7854,
+      "step": 3180
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.8012816905975342,
+      "learning_rate": 0.00020075663607045397,
+      "loss": 0.406,
+      "step": 3181
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 2.1195740699768066,
+      "learning_rate": 0.00020071942446043162,
+      "loss": 0.309,
+      "step": 3182
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 2.053685188293457,
+      "learning_rate": 0.0002006822128504093,
+      "loss": 0.485,
+      "step": 3183
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 2.672041654586792,
+      "learning_rate": 0.000200645001240387,
+      "loss": 0.4533,
+      "step": 3184
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 3.1025266647338867,
+      "learning_rate": 0.00020060778963036465,
+      "loss": 0.9278,
+      "step": 3185
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.895678162574768,
+      "learning_rate": 0.00020057057802034232,
+      "loss": 0.3792,
+      "step": 3186
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 3.569613218307495,
+      "learning_rate": 0.00020053336641032003,
+      "loss": 0.9807,
+      "step": 3187
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 2.383301019668579,
+      "learning_rate": 0.00020049615480029767,
+      "loss": 0.4056,
+      "step": 3188
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 5.517592430114746,
+      "learning_rate": 0.00020045894319027535,
+      "loss": 0.4775,
+      "step": 3189
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 3.9130759239196777,
+      "learning_rate": 0.000200421731580253,
+      "loss": 1.5079,
+      "step": 3190
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.221430778503418,
+      "learning_rate": 0.0002003845199702307,
+      "loss": 0.1815,
+      "step": 3191
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.0000752210617065,
+      "learning_rate": 0.00020034730836020838,
+      "loss": 0.1435,
+      "step": 3192
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 2.36952543258667,
+      "learning_rate": 0.00020031009675018603,
+      "loss": 0.4677,
+      "step": 3193
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.628010869026184,
+      "learning_rate": 0.00020027288514016373,
+      "loss": 0.2069,
+      "step": 3194
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 3.232687473297119,
+      "learning_rate": 0.00020023567353014138,
+      "loss": 0.4343,
+      "step": 3195
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 2.015259027481079,
+      "learning_rate": 0.00020019846192011905,
+      "loss": 0.3253,
+      "step": 3196
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.6881203651428223,
+      "learning_rate": 0.00020016125031009676,
+      "loss": 0.2276,
+      "step": 3197
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 2.478654384613037,
+      "learning_rate": 0.0002001240387000744,
+      "loss": 0.4578,
+      "step": 3198
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 3.1310973167419434,
+      "learning_rate": 0.00020008682709005208,
+      "loss": 0.3007,
+      "step": 3199
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 2.1835124492645264,
+      "learning_rate": 0.00020004961548002976,
+      "loss": 0.4325,
+      "step": 3200
+    },
+    {
+      "epoch": 1.12,
+      "eval_loss": 0.5288262367248535,
+      "eval_runtime": 51.0509,
+      "eval_samples_per_second": 42.467,
+      "eval_steps_per_second": 10.617,
+      "eval_wer": 0.4644525168655942,
+      "step": 3200
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 5.164034843444824,
+      "learning_rate": 0.00020001240387000743,
+      "loss": 0.4793,
+      "step": 3201
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 4.324225902557373,
+      "learning_rate": 0.0001999751922599851,
+      "loss": 1.9038,
+      "step": 3202
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 3.0840883255004883,
+      "learning_rate": 0.00019993798064996276,
+      "loss": 0.2772,
+      "step": 3203
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 4.304738521575928,
+      "learning_rate": 0.00019990076903994043,
+      "loss": 0.4444,
+      "step": 3204
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 2.658390998840332,
+      "learning_rate": 0.00019986355742991814,
+      "loss": 1.3376,
+      "step": 3205
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.7802234888076782,
+      "learning_rate": 0.00019982634581989578,
+      "loss": 0.5389,
+      "step": 3206
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 2.279994010925293,
+      "learning_rate": 0.00019978913420987346,
+      "loss": 0.5139,
+      "step": 3207
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 2.162747621536255,
+      "learning_rate": 0.0001997519225998511,
+      "loss": 0.508,
+      "step": 3208
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 2.005580186843872,
+      "learning_rate": 0.0001997147109898288,
+      "loss": 0.5523,
+      "step": 3209
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 3.2639243602752686,
+      "learning_rate": 0.0001996774993798065,
+      "loss": 0.7956,
+      "step": 3210
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.1911425590515137,
+      "learning_rate": 0.00019964028776978414,
+      "loss": 0.1047,
+      "step": 3211
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 2.8164703845977783,
+      "learning_rate": 0.00019960307615976184,
+      "loss": 0.3521,
+      "step": 3212
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.5338577032089233,
+      "learning_rate": 0.0001995658645497395,
+      "loss": 0.3544,
+      "step": 3213
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 2.2264373302459717,
+      "learning_rate": 0.00019952865293971716,
+      "loss": 0.5603,
+      "step": 3214
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 2.001485824584961,
+      "learning_rate": 0.00019949144132969487,
+      "loss": 0.3046,
+      "step": 3215
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 4.817356586456299,
+      "learning_rate": 0.00019945422971967252,
+      "loss": 0.527,
+      "step": 3216
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 2.8212592601776123,
+      "learning_rate": 0.0001994170181096502,
+      "loss": 0.595,
+      "step": 3217
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 4.72152853012085,
+      "learning_rate": 0.0001993798064996279,
+      "loss": 0.9974,
+      "step": 3218
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 3.7662367820739746,
+      "learning_rate": 0.00019934259488960554,
+      "loss": 0.8654,
+      "step": 3219
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.8878252506256104,
+      "learning_rate": 0.00019930538327958322,
+      "loss": 0.0876,
+      "step": 3220
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 4.1150922775268555,
+      "learning_rate": 0.00019926817166956087,
+      "loss": 0.3632,
+      "step": 3221
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 2.1186084747314453,
+      "learning_rate": 0.00019923096005953857,
+      "loss": 0.4102,
+      "step": 3222
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 2.5478599071502686,
+      "learning_rate": 0.00019919374844951625,
+      "loss": 0.2458,
+      "step": 3223
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 2.566505193710327,
+      "learning_rate": 0.0001991565368394939,
+      "loss": 0.2762,
+      "step": 3224
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 2.5866873264312744,
+      "learning_rate": 0.00019911932522947157,
+      "loss": 0.8733,
+      "step": 3225
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.6467078924179077,
+      "learning_rate": 0.00019908211361944925,
+      "loss": 0.1229,
+      "step": 3226
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 7.527315139770508,
+      "learning_rate": 0.00019904490200942692,
+      "loss": 0.5953,
+      "step": 3227
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 2.10206937789917,
+      "learning_rate": 0.0001990076903994046,
+      "loss": 0.4737,
+      "step": 3228
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.532116174697876,
+      "learning_rate": 0.00019897047878938225,
+      "loss": 0.1381,
+      "step": 3229
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.7178963422775269,
+      "learning_rate": 0.00019893326717935995,
+      "loss": 0.6589,
+      "step": 3230
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.9948376417160034,
+      "learning_rate": 0.00019889605556933763,
+      "loss": 0.4115,
+      "step": 3231
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.7607015371322632,
+      "learning_rate": 0.00019885884395931527,
+      "loss": 0.3731,
+      "step": 3232
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.7697004079818726,
+      "learning_rate": 0.00019882163234929298,
+      "loss": 0.2709,
+      "step": 3233
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.821341633796692,
+      "learning_rate": 0.00019878442073927063,
+      "loss": 0.2842,
+      "step": 3234
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 2.076326370239258,
+      "learning_rate": 0.0001987472091292483,
+      "loss": 0.3179,
+      "step": 3235
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.9342252016067505,
+      "learning_rate": 0.000198709997519226,
+      "loss": 0.2932,
+      "step": 3236
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 2.633678674697876,
+      "learning_rate": 0.00019867278590920365,
+      "loss": 0.5614,
+      "step": 3237
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 2.345721483230591,
+      "learning_rate": 0.00019863557429918133,
+      "loss": 0.3905,
+      "step": 3238
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 5.671786785125732,
+      "learning_rate": 0.00019859836268915898,
+      "loss": 0.373,
+      "step": 3239
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 3.5498249530792236,
+      "learning_rate": 0.00019856115107913668,
+      "loss": 0.3899,
+      "step": 3240
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 2.3076889514923096,
+      "learning_rate": 0.00019852393946911436,
+      "loss": 0.3927,
+      "step": 3241
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.8654396533966064,
+      "learning_rate": 0.000198486727859092,
+      "loss": 0.4113,
+      "step": 3242
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 3.3518593311309814,
+      "learning_rate": 0.0001984495162490697,
+      "loss": 0.4291,
+      "step": 3243
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 3.1347527503967285,
+      "learning_rate": 0.00019841230463904738,
+      "loss": 0.5388,
+      "step": 3244
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.9862194061279297,
+      "learning_rate": 0.00019837509302902503,
+      "loss": 0.3185,
+      "step": 3245
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 3.0931460857391357,
+      "learning_rate": 0.0001983378814190027,
+      "loss": 0.2293,
+      "step": 3246
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 3.0222811698913574,
+      "learning_rate": 0.00019830066980898038,
+      "loss": 0.675,
+      "step": 3247
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 4.591670989990234,
+      "learning_rate": 0.00019826345819895806,
+      "loss": 1.6013,
+      "step": 3248
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 2.511385679244995,
+      "learning_rate": 0.00019822624658893574,
+      "loss": 0.3192,
+      "step": 3249
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 2.967205047607422,
+      "learning_rate": 0.00019818903497891338,
+      "loss": 0.2893,
+      "step": 3250
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.8851898908615112,
+      "learning_rate": 0.0001981518233688911,
+      "loss": 0.221,
+      "step": 3251
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 4.387383460998535,
+      "learning_rate": 0.00019811461175886874,
+      "loss": 0.3816,
+      "step": 3252
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.66984224319458,
+      "learning_rate": 0.0001980774001488464,
+      "loss": 0.1179,
+      "step": 3253
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 2.261944055557251,
+      "learning_rate": 0.00019804018853882411,
+      "loss": 0.23,
+      "step": 3254
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 2.3139607906341553,
+      "learning_rate": 0.00019800297692880176,
+      "loss": 0.6849,
+      "step": 3255
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 3.3277676105499268,
+      "learning_rate": 0.00019796576531877944,
+      "loss": 0.821,
+      "step": 3256
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.5088117122650146,
+      "learning_rate": 0.0001979285537087571,
+      "loss": 0.3421,
+      "step": 3257
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 26.52508544921875,
+      "learning_rate": 0.0001978913420987348,
+      "loss": 5.8404,
+      "step": 3258
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 2.2505085468292236,
+      "learning_rate": 0.00019785413048871247,
+      "loss": 0.2756,
+      "step": 3259
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.61557137966156,
+      "learning_rate": 0.00019781691887869012,
+      "loss": 0.2101,
+      "step": 3260
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.9748115539550781,
+      "learning_rate": 0.00019777970726866782,
+      "loss": 0.3818,
+      "step": 3261
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.2797274589538574,
+      "learning_rate": 0.0001977424956586455,
+      "loss": 0.2122,
+      "step": 3262
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 2.0891008377075195,
+      "learning_rate": 0.00019770528404862314,
+      "loss": 0.3929,
+      "step": 3263
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.7179592847824097,
+      "learning_rate": 0.00019766807243860085,
+      "loss": 0.2509,
+      "step": 3264
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 2.773587703704834,
+      "learning_rate": 0.0001976308608285785,
+      "loss": 0.2744,
+      "step": 3265
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.902330756187439,
+      "learning_rate": 0.00019759364921855617,
+      "loss": 0.3719,
+      "step": 3266
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 3.543816566467285,
+      "learning_rate": 0.00019755643760853385,
+      "loss": 0.6176,
+      "step": 3267
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 2.187521457672119,
+      "learning_rate": 0.00019751922599851152,
+      "loss": 0.2295,
+      "step": 3268
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 2.898231029510498,
+      "learning_rate": 0.0001974820143884892,
+      "loss": 0.3294,
+      "step": 3269
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.8710620403289795,
+      "learning_rate": 0.00019744480277846685,
+      "loss": 0.2361,
+      "step": 3270
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.4173840284347534,
+      "learning_rate": 0.00019740759116844452,
+      "loss": 0.2267,
+      "step": 3271
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 3.5871973037719727,
+      "learning_rate": 0.00019737037955842222,
+      "loss": 0.6848,
+      "step": 3272
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 3.1234724521636963,
+      "learning_rate": 0.00019733316794839987,
+      "loss": 0.3783,
+      "step": 3273
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.742729902267456,
+      "learning_rate": 0.00019729595633837755,
+      "loss": 0.3094,
+      "step": 3274
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 3.235287666320801,
+      "learning_rate": 0.00019725874472835525,
+      "loss": 0.5631,
+      "step": 3275
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 5.217351913452148,
+      "learning_rate": 0.0001972215331183329,
+      "loss": 0.5498,
+      "step": 3276
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.3385629653930664,
+      "learning_rate": 0.00019718432150831058,
+      "loss": 0.066,
+      "step": 3277
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 5.248405933380127,
+      "learning_rate": 0.00019714710989828823,
+      "loss": 0.769,
+      "step": 3278
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 3.6078615188598633,
+      "learning_rate": 0.00019710989828826593,
+      "loss": 1.5544,
+      "step": 3279
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 3.464466094970703,
+      "learning_rate": 0.0001970726866782436,
+      "loss": 0.6935,
+      "step": 3280
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.388893723487854,
+      "learning_rate": 0.00019703547506822125,
+      "loss": 0.196,
+      "step": 3281
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 2.4973855018615723,
+      "learning_rate": 0.00019699826345819896,
+      "loss": 1.0041,
+      "step": 3282
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.7962448596954346,
+      "learning_rate": 0.0001969610518481766,
+      "loss": 0.3729,
+      "step": 3283
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 2.4133572578430176,
+      "learning_rate": 0.00019692384023815428,
+      "loss": 0.4041,
+      "step": 3284
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.40854012966156,
+      "learning_rate": 0.00019688662862813198,
+      "loss": 0.2944,
+      "step": 3285
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 2.9127371311187744,
+      "learning_rate": 0.00019684941701810963,
+      "loss": 0.1197,
+      "step": 3286
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.9602265357971191,
+      "learning_rate": 0.0001968122054080873,
+      "loss": 0.3084,
+      "step": 3287
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 3.038182258605957,
+      "learning_rate": 0.00019677499379806498,
+      "loss": 0.7004,
+      "step": 3288
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.9776606559753418,
+      "learning_rate": 0.00019673778218804266,
+      "loss": 0.5882,
+      "step": 3289
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 2.051262855529785,
+      "learning_rate": 0.00019670057057802033,
+      "loss": 0.3358,
+      "step": 3290
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 2.713244676589966,
+      "learning_rate": 0.00019666335896799798,
+      "loss": 0.6035,
+      "step": 3291
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.6439521312713623,
+      "learning_rate": 0.00019662614735797566,
+      "loss": 0.1498,
+      "step": 3292
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 2.486056327819824,
+      "learning_rate": 0.00019658893574795336,
+      "loss": 0.5642,
+      "step": 3293
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.6842329502105713,
+      "learning_rate": 0.000196551724137931,
+      "loss": 0.3027,
+      "step": 3294
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 5.799333095550537,
+      "learning_rate": 0.0001965145125279087,
+      "loss": 0.771,
+      "step": 3295
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 3.132808208465576,
+      "learning_rate": 0.00019647730091788636,
+      "loss": 0.5863,
+      "step": 3296
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.502746343612671,
+      "learning_rate": 0.00019644008930786404,
+      "loss": 0.1146,
+      "step": 3297
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.2371695041656494,
+      "learning_rate": 0.00019640287769784171,
+      "loss": 0.1923,
+      "step": 3298
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.7083606719970703,
+      "learning_rate": 0.00019636566608781936,
+      "loss": 0.2605,
+      "step": 3299
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.663743019104004,
+      "learning_rate": 0.00019632845447779707,
+      "loss": 0.2404,
+      "step": 3300
+    },
+    {
+      "epoch": 1.16,
+      "eval_loss": 0.53449547290802,
+      "eval_runtime": 51.0865,
+      "eval_samples_per_second": 42.438,
+      "eval_steps_per_second": 10.609,
+      "eval_wer": 0.48806434872859367,
+      "step": 3300
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.146204710006714,
+      "learning_rate": 0.00019629124286777474,
+      "loss": 0.3883,
+      "step": 3301
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 4.3042311668396,
+      "learning_rate": 0.0001962540312577524,
+      "loss": 0.5085,
+      "step": 3302
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.9416069984436035,
+      "learning_rate": 0.0001962168196477301,
+      "loss": 0.1635,
+      "step": 3303
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.1905418634414673,
+      "learning_rate": 0.00019617960803770774,
+      "loss": 0.0755,
+      "step": 3304
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.9481916427612305,
+      "learning_rate": 0.00019614239642768542,
+      "loss": 1.2842,
+      "step": 3305
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 4.018986701965332,
+      "learning_rate": 0.00019610518481766312,
+      "loss": 1.239,
+      "step": 3306
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.8045568466186523,
+      "learning_rate": 0.00019606797320764077,
+      "loss": 0.7349,
+      "step": 3307
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.780566692352295,
+      "learning_rate": 0.00019603076159761844,
+      "loss": 0.6553,
+      "step": 3308
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.8182471990585327,
+      "learning_rate": 0.0001959935499875961,
+      "loss": 0.3462,
+      "step": 3309
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.2723684310913086,
+      "learning_rate": 0.0001959563383775738,
+      "loss": 0.3089,
+      "step": 3310
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.788958787918091,
+      "learning_rate": 0.00019591912676755147,
+      "loss": 0.439,
+      "step": 3311
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.5581282377243042,
+      "learning_rate": 0.00019588191515752912,
+      "loss": 0.1266,
+      "step": 3312
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.770595073699951,
+      "learning_rate": 0.0001958447035475068,
+      "loss": 0.5998,
+      "step": 3313
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.167032480239868,
+      "learning_rate": 0.00019580749193748447,
+      "loss": 0.4174,
+      "step": 3314
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 3.6844260692596436,
+      "learning_rate": 0.00019577028032746215,
+      "loss": 0.574,
+      "step": 3315
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.024935722351074,
+      "learning_rate": 0.00019573306871743982,
+      "loss": 0.2714,
+      "step": 3316
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 3.656796932220459,
+      "learning_rate": 0.0001956958571074175,
+      "loss": 0.7618,
+      "step": 3317
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 3.2108490467071533,
+      "learning_rate": 0.00019565864549739518,
+      "loss": 0.4265,
+      "step": 3318
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.514299750328064,
+      "learning_rate": 0.00019562143388737285,
+      "loss": 0.1267,
+      "step": 3319
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 4.003231048583984,
+      "learning_rate": 0.0001955842222773505,
+      "loss": 0.4855,
+      "step": 3320
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.6469032764434814,
+      "learning_rate": 0.0001955470106673282,
+      "loss": 0.4387,
+      "step": 3321
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.6862777471542358,
+      "learning_rate": 0.00019550979905730585,
+      "loss": 0.2019,
+      "step": 3322
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 3.7266769409179688,
+      "learning_rate": 0.00019547258744728353,
+      "loss": 0.6177,
+      "step": 3323
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.3731110095977783,
+      "learning_rate": 0.00019543537583726123,
+      "loss": 0.2896,
+      "step": 3324
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 3.063096761703491,
+      "learning_rate": 0.00019539816422723888,
+      "loss": 0.3411,
+      "step": 3325
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 2.101358413696289,
+      "learning_rate": 0.00019536095261721655,
+      "loss": 0.3822,
+      "step": 3326
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 4.259075164794922,
+      "learning_rate": 0.0001953237410071942,
+      "loss": 1.7024,
+      "step": 3327
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 4.619799613952637,
+      "learning_rate": 0.0001952865293971719,
+      "loss": 0.7057,
+      "step": 3328
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.7613863945007324,
+      "learning_rate": 0.00019524931778714958,
+      "loss": 0.1737,
+      "step": 3329
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 2.4438929557800293,
+      "learning_rate": 0.00019521210617712723,
+      "loss": 0.759,
+      "step": 3330
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 2.7754616737365723,
+      "learning_rate": 0.00019517489456710493,
+      "loss": 0.7182,
+      "step": 3331
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 2.5432474613189697,
+      "learning_rate": 0.0001951376829570826,
+      "loss": 0.3178,
+      "step": 3332
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.662652850151062,
+      "learning_rate": 0.00019510047134706026,
+      "loss": 0.268,
+      "step": 3333
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.8840018510818481,
+      "learning_rate": 0.00019506325973703793,
+      "loss": 0.5437,
+      "step": 3334
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 2.192028760910034,
+      "learning_rate": 0.0001950260481270156,
+      "loss": 0.2108,
+      "step": 3335
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.9820222854614258,
+      "learning_rate": 0.00019498883651699329,
+      "loss": 0.2483,
+      "step": 3336
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 3.4582228660583496,
+      "learning_rate": 0.00019495162490697096,
+      "loss": 0.6822,
+      "step": 3337
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.7434332370758057,
+      "learning_rate": 0.00019491441329694864,
+      "loss": 0.2123,
+      "step": 3338
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 2.4500560760498047,
+      "learning_rate": 0.0001948772016869263,
+      "loss": 0.482,
+      "step": 3339
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 2.2425453662872314,
+      "learning_rate": 0.00019483999007690396,
+      "loss": 0.3816,
+      "step": 3340
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 3.76282000541687,
+      "learning_rate": 0.00019480277846688164,
+      "loss": 0.3388,
+      "step": 3341
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.8158988952636719,
+      "learning_rate": 0.00019476556685685934,
+      "loss": 0.2258,
+      "step": 3342
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.5334137678146362,
+      "learning_rate": 0.000194728355246837,
+      "loss": 0.217,
+      "step": 3343
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 3.0049142837524414,
+      "learning_rate": 0.00019469114363681466,
+      "loss": 0.3094,
+      "step": 3344
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.7647264003753662,
+      "learning_rate": 0.00019465393202679237,
+      "loss": 0.0804,
+      "step": 3345
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 2.254091739654541,
+      "learning_rate": 0.00019461672041677002,
+      "loss": 0.2055,
+      "step": 3346
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 3.365161657333374,
+      "learning_rate": 0.0001945795088067477,
+      "loss": 0.6031,
+      "step": 3347
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 2.869905471801758,
+      "learning_rate": 0.00019454229719672534,
+      "loss": 0.2699,
+      "step": 3348
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.7060647010803223,
+      "learning_rate": 0.00019450508558670304,
+      "loss": 0.1983,
+      "step": 3349
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 3.431352376937866,
+      "learning_rate": 0.00019446787397668072,
+      "loss": 0.2818,
+      "step": 3350
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 3.3158788681030273,
+      "learning_rate": 0.00019443066236665837,
+      "loss": 0.6713,
+      "step": 3351
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 2.5465288162231445,
+      "learning_rate": 0.00019439345075663607,
+      "loss": 0.1836,
+      "step": 3352
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 7.218588352203369,
+      "learning_rate": 0.00019435623914661372,
+      "loss": 0.5528,
+      "step": 3353
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 3.608271360397339,
+      "learning_rate": 0.0001943190275365914,
+      "loss": 0.2997,
+      "step": 3354
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 2.334040403366089,
+      "learning_rate": 0.00019428181592656907,
+      "loss": 0.8903,
+      "step": 3355
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 3.217416524887085,
+      "learning_rate": 0.00019424460431654675,
+      "loss": 0.8583,
+      "step": 3356
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 2.663219928741455,
+      "learning_rate": 0.00019420739270652442,
+      "loss": 1.0071,
+      "step": 3357
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 2.4265856742858887,
+      "learning_rate": 0.00019417018109650207,
+      "loss": 0.5831,
+      "step": 3358
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 3.747819185256958,
+      "learning_rate": 0.00019413296948647977,
+      "loss": 0.9428,
+      "step": 3359
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.7516096830368042,
+      "learning_rate": 0.00019409575787645745,
+      "loss": 0.3003,
+      "step": 3360
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 2.066317558288574,
+      "learning_rate": 0.0001940585462664351,
+      "loss": 0.4669,
+      "step": 3361
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.9735511541366577,
+      "learning_rate": 0.00019402133465641277,
+      "loss": 0.5892,
+      "step": 3362
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 3.3901679515838623,
+      "learning_rate": 0.00019398412304639048,
+      "loss": 0.6881,
+      "step": 3363
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 6.7699785232543945,
+      "learning_rate": 0.00019394691143636813,
+      "loss": 0.9592,
+      "step": 3364
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 3.4974114894866943,
+      "learning_rate": 0.0001939096998263458,
+      "loss": 0.4567,
+      "step": 3365
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.7345564365386963,
+      "learning_rate": 0.00019387248821632345,
+      "loss": 0.2935,
+      "step": 3366
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 2.886187791824341,
+      "learning_rate": 0.00019383527660630115,
+      "loss": 0.3215,
+      "step": 3367
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 2.7109858989715576,
+      "learning_rate": 0.00019379806499627883,
+      "loss": 0.6187,
+      "step": 3368
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 3.9955689907073975,
+      "learning_rate": 0.00019376085338625648,
+      "loss": 0.9804,
+      "step": 3369
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 2.5214648246765137,
+      "learning_rate": 0.00019372364177623418,
+      "loss": 0.509,
+      "step": 3370
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.90560781955719,
+      "learning_rate": 0.00019368643016621183,
+      "loss": 0.1936,
+      "step": 3371
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 5.796487331390381,
+      "learning_rate": 0.0001936492185561895,
+      "loss": 1.6628,
+      "step": 3372
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 2.390986919403076,
+      "learning_rate": 0.0001936120069461672,
+      "loss": 0.1252,
+      "step": 3373
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.6030224561691284,
+      "learning_rate": 0.00019357479533614486,
+      "loss": 0.1826,
+      "step": 3374
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.6612952947616577,
+      "learning_rate": 0.00019353758372612253,
+      "loss": 0.3292,
+      "step": 3375
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 4.129511833190918,
+      "learning_rate": 0.0001935003721161002,
+      "loss": 0.5278,
+      "step": 3376
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 3.8274450302124023,
+      "learning_rate": 0.00019346316050607788,
+      "loss": 0.537,
+      "step": 3377
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.8161903619766235,
+      "learning_rate": 0.00019342594889605556,
+      "loss": 0.2818,
+      "step": 3378
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 3.630079746246338,
+      "learning_rate": 0.0001933887372860332,
+      "loss": 0.1704,
+      "step": 3379
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 3.014974594116211,
+      "learning_rate": 0.0001933515256760109,
+      "loss": 1.1119,
+      "step": 3380
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 6.352705478668213,
+      "learning_rate": 0.0001933143140659886,
+      "loss": 1.7637,
+      "step": 3381
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.94754159450531,
+      "learning_rate": 0.00019327710245596624,
+      "loss": 0.5617,
+      "step": 3382
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.7361313104629517,
+      "learning_rate": 0.0001932398908459439,
+      "loss": 0.4192,
+      "step": 3383
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.9336076974868774,
+      "learning_rate": 0.0001932026792359216,
+      "loss": 0.5007,
+      "step": 3384
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.9910989999771118,
+      "learning_rate": 0.00019316546762589926,
+      "loss": 0.6072,
+      "step": 3385
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.9290117025375366,
+      "learning_rate": 0.00019312825601587694,
+      "loss": 0.3207,
+      "step": 3386
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 2.000905990600586,
+      "learning_rate": 0.0001930910444058546,
+      "loss": 0.6482,
+      "step": 3387
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 2.0836875438690186,
+      "learning_rate": 0.0001930538327958323,
+      "loss": 0.3071,
+      "step": 3388
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.299101710319519,
+      "learning_rate": 0.00019301662118580997,
+      "loss": 0.2281,
+      "step": 3389
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 2.1644299030303955,
+      "learning_rate": 0.00019297940957578762,
+      "loss": 0.4635,
+      "step": 3390
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.7756919860839844,
+      "learning_rate": 0.00019294219796576532,
+      "loss": 0.3381,
+      "step": 3391
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.1819500923156738,
+      "learning_rate": 0.00019290498635574297,
+      "loss": 0.0619,
+      "step": 3392
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 2.2042393684387207,
+      "learning_rate": 0.00019286777474572064,
+      "loss": 0.3169,
+      "step": 3393
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.324364423751831,
+      "learning_rate": 0.00019283056313569835,
+      "loss": 0.0623,
+      "step": 3394
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 3.631133556365967,
+      "learning_rate": 0.000192793351525676,
+      "loss": 0.4106,
+      "step": 3395
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 4.347462177276611,
+      "learning_rate": 0.00019275613991565367,
+      "loss": 0.4728,
+      "step": 3396
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 3.753833770751953,
+      "learning_rate": 0.00019271892830563132,
+      "loss": 0.6239,
+      "step": 3397
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 3.2457504272460938,
+      "learning_rate": 0.00019268171669560902,
+      "loss": 0.4568,
+      "step": 3398
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 3.2867891788482666,
+      "learning_rate": 0.0001926445050855867,
+      "loss": 0.8455,
+      "step": 3399
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 2.0727713108062744,
+      "learning_rate": 0.00019260729347556435,
+      "loss": 0.2855,
+      "step": 3400
+    },
+    {
+      "epoch": 1.19,
+      "eval_loss": 0.5019727349281311,
+      "eval_runtime": 51.094,
+      "eval_samples_per_second": 42.432,
+      "eval_steps_per_second": 10.608,
+      "eval_wer": 0.48131811105345096,
+      "step": 3400
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.6999847292900085,
+      "learning_rate": 0.00019257008186554205,
+      "loss": 0.0542,
+      "step": 3401
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 2.255366563796997,
+      "learning_rate": 0.0001925328702555197,
+      "loss": 0.2321,
+      "step": 3402
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 3.2004432678222656,
+      "learning_rate": 0.00019249565864549737,
+      "loss": 0.5873,
+      "step": 3403
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 7.720293998718262,
+      "learning_rate": 0.00019245844703547505,
+      "loss": 1.3688,
+      "step": 3404
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 2.9082274436950684,
+      "learning_rate": 0.00019242123542545273,
+      "loss": 0.8094,
+      "step": 3405
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 24.3272762298584,
+      "learning_rate": 0.0001923840238154304,
+      "loss": 5.8724,
+      "step": 3406
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.7688156366348267,
+      "learning_rate": 0.00019234681220540808,
+      "loss": 0.2914,
+      "step": 3407
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 2.372715950012207,
+      "learning_rate": 0.00019230960059538573,
+      "loss": 0.2986,
+      "step": 3408
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 2.354346990585327,
+      "learning_rate": 0.00019227238898536343,
+      "loss": 0.4223,
+      "step": 3409
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.976027250289917,
+      "learning_rate": 0.00019223517737534108,
+      "loss": 0.2763,
+      "step": 3410
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.1134535074234009,
+      "learning_rate": 0.00019219796576531875,
+      "loss": 0.1037,
+      "step": 3411
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 2.9392096996307373,
+      "learning_rate": 0.00019216075415529646,
+      "loss": 0.4426,
+      "step": 3412
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.128270149230957,
+      "learning_rate": 0.0001921235425452741,
+      "loss": 0.1267,
+      "step": 3413
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.8088712692260742,
+      "learning_rate": 0.00019208633093525178,
+      "loss": 0.0562,
+      "step": 3414
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": NaN,
+      "learning_rate": 0.00019208633093525178,
+      "loss": 0.4185,
+      "step": 3415
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 4.43418550491333,
+      "learning_rate": 0.00019204911932522943,
+      "loss": 0.8585,
+      "step": 3416
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 2.2035839557647705,
+      "learning_rate": 0.00019201190771520713,
+      "loss": 0.2187,
+      "step": 3417
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 2.9955644607543945,
+      "learning_rate": 0.0001919746961051848,
+      "loss": 0.2888,
+      "step": 3418
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 3.610004186630249,
+      "learning_rate": 0.00019193748449516246,
+      "loss": 0.8794,
+      "step": 3419
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 5.0635809898376465,
+      "learning_rate": 0.00019190027288514016,
+      "loss": 0.5365,
+      "step": 3420
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 2.6513617038726807,
+      "learning_rate": 0.00019186306127511784,
+      "loss": 0.2151,
+      "step": 3421
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 3.7692182064056396,
+      "learning_rate": 0.00019182584966509548,
+      "loss": 0.5967,
+      "step": 3422
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 4.396376609802246,
+      "learning_rate": 0.0001917886380550732,
+      "loss": 0.6936,
+      "step": 3423
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 5.046158313751221,
+      "learning_rate": 0.00019175142644505084,
+      "loss": 0.7558,
+      "step": 3424
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 2.674578905105591,
+      "learning_rate": 0.0001917142148350285,
+      "loss": 0.2655,
+      "step": 3425
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 3.748619556427002,
+      "learning_rate": 0.0001916770032250062,
+      "loss": 0.7096,
+      "step": 3426
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 4.511143684387207,
+      "learning_rate": 0.00019163979161498386,
+      "loss": 0.3599,
+      "step": 3427
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 7.0941481590271,
+      "learning_rate": 0.00019160258000496154,
+      "loss": 0.2173,
+      "step": 3428
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": NaN,
+      "learning_rate": 0.00019160258000496154,
+      "loss": 0.0813,
+      "step": 3429
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 2.511932134628296,
+      "learning_rate": 0.0001915653683949392,
+      "loss": 0.896,
+      "step": 3430
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.8741555213928223,
+      "learning_rate": 0.00019152815678491686,
+      "loss": 0.4229,
+      "step": 3431
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.6796114444732666,
+      "learning_rate": 0.00019149094517489457,
+      "loss": 0.4073,
+      "step": 3432
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 2.2420263290405273,
+      "learning_rate": 0.00019145373356487222,
+      "loss": 0.2323,
+      "step": 3433
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.184206485748291,
+      "learning_rate": 0.0001914165219548499,
+      "loss": 0.1977,
+      "step": 3434
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.924786925315857,
+      "learning_rate": 0.0001913793103448276,
+      "loss": 0.3857,
+      "step": 3435
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 2.2253284454345703,
+      "learning_rate": 0.00019134209873480524,
+      "loss": 0.3105,
+      "step": 3436
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 2.1328001022338867,
+      "learning_rate": 0.00019130488712478292,
+      "loss": 0.3734,
+      "step": 3437
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 2.7778384685516357,
+      "learning_rate": 0.00019126767551476057,
+      "loss": 0.2954,
+      "step": 3438
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 2.7506439685821533,
+      "learning_rate": 0.00019123046390473827,
+      "loss": 0.2154,
+      "step": 3439
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.9881768226623535,
+      "learning_rate": 0.00019119325229471595,
+      "loss": 0.2684,
+      "step": 3440
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 2.7313220500946045,
+      "learning_rate": 0.0001911560406846936,
+      "loss": 0.2496,
+      "step": 3441
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 2.945786237716675,
+      "learning_rate": 0.0001911188290746713,
+      "loss": 0.7178,
+      "step": 3442
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.2393323183059692,
+      "learning_rate": 0.00019108161746464895,
+      "loss": 0.1755,
+      "step": 3443
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 2.6016125679016113,
+      "learning_rate": 0.00019104440585462662,
+      "loss": 0.4143,
+      "step": 3444
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 4.074510097503662,
+      "learning_rate": 0.00019100719424460432,
+      "loss": 0.4698,
+      "step": 3445
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.408915400505066,
+      "learning_rate": 0.00019096998263458197,
+      "loss": 0.1266,
+      "step": 3446
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 2.3694286346435547,
+      "learning_rate": 0.00019093277102455965,
+      "loss": 0.3139,
+      "step": 3447
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 2.525958776473999,
+      "learning_rate": 0.0001908955594145373,
+      "loss": 0.4195,
+      "step": 3448
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 2.69555401802063,
+      "learning_rate": 0.000190858347804515,
+      "loss": 0.3823,
+      "step": 3449
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.620243787765503,
+      "learning_rate": 0.00019082113619449268,
+      "loss": 0.1779,
+      "step": 3450
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 4.35145902633667,
+      "learning_rate": 0.00019078392458447033,
+      "loss": 1.6517,
+      "step": 3451
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 3.985114336013794,
+      "learning_rate": 0.000190746712974448,
+      "loss": 1.7988,
+      "step": 3452
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 4.146234512329102,
+      "learning_rate": 0.0001907095013644257,
+      "loss": 0.2545,
+      "step": 3453
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 7.112207412719727,
+      "learning_rate": 0.00019067228975440335,
+      "loss": 0.6047,
+      "step": 3454
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 2.409411668777466,
+      "learning_rate": 0.00019063507814438103,
+      "loss": 0.9475,
+      "step": 3455
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 3.0187294483184814,
+      "learning_rate": 0.00019059786653435868,
+      "loss": 0.7817,
+      "step": 3456
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.8108159303665161,
+      "learning_rate": 0.00019056065492433638,
+      "loss": 0.5218,
+      "step": 3457
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.8068140745162964,
+      "learning_rate": 0.00019052344331431406,
+      "loss": 0.4357,
+      "step": 3458
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 2.0581214427948,
+      "learning_rate": 0.0001904862317042917,
+      "loss": 0.4965,
+      "step": 3459
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 2.0165770053863525,
+      "learning_rate": 0.0001904490200942694,
+      "loss": 0.3724,
+      "step": 3460
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.243851900100708,
+      "learning_rate": 0.00019041180848424706,
+      "loss": 0.2379,
+      "step": 3461
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 2.6917412281036377,
+      "learning_rate": 0.00019037459687422473,
+      "loss": 0.7907,
+      "step": 3462
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.7246170043945312,
+      "learning_rate": 0.00019033738526420243,
+      "loss": 0.4098,
+      "step": 3463
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.3708009719848633,
+      "learning_rate": 0.00019030017365418008,
+      "loss": 0.182,
+      "step": 3464
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.1372557878494263,
+      "learning_rate": 0.00019026296204415776,
+      "loss": 0.2405,
+      "step": 3465
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 2.3582069873809814,
+      "learning_rate": 0.00019022575043413546,
+      "loss": 0.3401,
+      "step": 3466
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 5.7276835441589355,
+      "learning_rate": 0.0001901885388241131,
+      "loss": 0.4506,
+      "step": 3467
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.6258736848831177,
+      "learning_rate": 0.0001901513272140908,
+      "loss": 0.2368,
+      "step": 3468
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 3.5020952224731445,
+      "learning_rate": 0.00019011411560406844,
+      "loss": 0.7847,
+      "step": 3469
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 4.4619526863098145,
+      "learning_rate": 0.00019007690399404614,
+      "loss": 0.3962,
+      "step": 3470
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.9173682928085327,
+      "learning_rate": 0.00019003969238402381,
+      "loss": 0.194,
+      "step": 3471
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 2.3916375637054443,
+      "learning_rate": 0.00019000248077400146,
+      "loss": 0.3269,
+      "step": 3472
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.9122830629348755,
+      "learning_rate": 0.00018996526916397914,
+      "loss": 0.0976,
+      "step": 3473
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.4700841903686523,
+      "learning_rate": 0.00018992805755395681,
+      "loss": 0.0848,
+      "step": 3474
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 5.40512228012085,
+      "learning_rate": 0.0001898908459439345,
+      "loss": 0.3826,
+      "step": 3475
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 9.25135612487793,
+      "learning_rate": 0.00018985363433391217,
+      "loss": 2.3233,
+      "step": 3476
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 4.043542861938477,
+      "learning_rate": 0.00018981642272388981,
+      "loss": 0.726,
+      "step": 3477
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.8716025948524475,
+      "learning_rate": 0.00018977921111386752,
+      "loss": 0.0542,
+      "step": 3478
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.9265203475952148,
+      "learning_rate": 0.0001897419995038452,
+      "loss": 0.2385,
+      "step": 3479
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 3.297703981399536,
+      "learning_rate": 0.00018970478789382284,
+      "loss": 0.9332,
+      "step": 3480
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 4.390806674957275,
+      "learning_rate": 0.00018966757628380054,
+      "loss": 0.9762,
+      "step": 3481
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 2.7843103408813477,
+      "learning_rate": 0.0001896303646737782,
+      "loss": 0.3961,
+      "step": 3482
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 2.2936036586761475,
+      "learning_rate": 0.00018959315306375587,
+      "loss": 0.515,
+      "step": 3483
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.33071768283844,
+      "learning_rate": 0.00018955594145373357,
+      "loss": 0.1201,
+      "step": 3484
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.876550555229187,
+      "learning_rate": 0.00018951872984371122,
+      "loss": 0.391,
+      "step": 3485
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.878441333770752,
+      "learning_rate": 0.0001894815182336889,
+      "loss": 0.2876,
+      "step": 3486
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 2.9820950031280518,
+      "learning_rate": 0.00018944430662366655,
+      "loss": 0.5285,
+      "step": 3487
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 2.3062541484832764,
+      "learning_rate": 0.00018940709501364425,
+      "loss": 0.3399,
+      "step": 3488
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 2.2896065711975098,
+      "learning_rate": 0.00018936988340362192,
+      "loss": 0.3287,
+      "step": 3489
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 2.6173689365386963,
+      "learning_rate": 0.00018933267179359957,
+      "loss": 0.7003,
+      "step": 3490
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.3826167583465576,
+      "learning_rate": 0.00018929546018357728,
+      "loss": 0.1447,
+      "step": 3491
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.945378303527832,
+      "learning_rate": 0.00018925824857355495,
+      "loss": 0.1798,
+      "step": 3492
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 3.251554489135742,
+      "learning_rate": 0.0001892210369635326,
+      "loss": 0.4669,
+      "step": 3493
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 3.7647817134857178,
+      "learning_rate": 0.00018918382535351028,
+      "loss": 0.3134,
+      "step": 3494
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.6382604837417603,
+      "learning_rate": 0.00018914661374348795,
+      "loss": 0.168,
+      "step": 3495
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 4.437539100646973,
+      "learning_rate": 0.00018910940213346563,
+      "loss": 0.4604,
+      "step": 3496
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 19.861299514770508,
+      "learning_rate": 0.0001890721905234433,
+      "loss": 1.9396,
+      "step": 3497
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.6166614294052124,
+      "learning_rate": 0.00018903497891342095,
+      "loss": 0.0805,
+      "step": 3498
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.6526031494140625,
+      "learning_rate": 0.00018899776730339865,
+      "loss": 0.2469,
+      "step": 3499
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 3.7113420963287354,
+      "learning_rate": 0.0001889605556933763,
+      "loss": 0.3856,
+      "step": 3500
+    },
+    {
+      "epoch": 1.23,
+      "eval_loss": 0.48135173320770264,
+      "eval_runtime": 51.3793,
+      "eval_samples_per_second": 42.196,
+      "eval_steps_per_second": 10.549,
+      "eval_wer": 0.4328835841549905,
+      "step": 3500
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 6.976255893707275,
+      "learning_rate": 0.00018892334408335398,
+      "loss": 2.395,
+      "step": 3501
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 3.2706868648529053,
+      "learning_rate": 0.00018888613247333168,
+      "loss": 0.5762,
+      "step": 3502
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 2.8764538764953613,
+      "learning_rate": 0.00018884892086330933,
+      "loss": 0.2334,
+      "step": 3503
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 4.566110610961914,
+      "learning_rate": 0.000188811709253287,
+      "loss": 1.624,
+      "step": 3504
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 3.793048620223999,
+      "learning_rate": 0.00018877449764326466,
+      "loss": 1.0498,
+      "step": 3505
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 2.507688283920288,
+      "learning_rate": 0.00018873728603324236,
+      "loss": 0.4346,
+      "step": 3506
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 2.692643404006958,
+      "learning_rate": 0.00018870007442322003,
+      "loss": 0.4612,
+      "step": 3507
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.4150710105895996,
+      "learning_rate": 0.00018866286281319768,
+      "loss": 0.2225,
+      "step": 3508
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 2.229931116104126,
+      "learning_rate": 0.00018862565120317539,
+      "loss": 0.386,
+      "step": 3509
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.482277750968933,
+      "learning_rate": 0.00018858843959315306,
+      "loss": 0.4351,
+      "step": 3510
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 2.131032943725586,
+      "learning_rate": 0.0001885512279831307,
+      "loss": 0.2765,
+      "step": 3511
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.5943416357040405,
+      "learning_rate": 0.0001885140163731084,
+      "loss": 0.3301,
+      "step": 3512
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.9223726987838745,
+      "learning_rate": 0.00018847680476308606,
+      "loss": 0.3836,
+      "step": 3513
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.1175265312194824,
+      "learning_rate": 0.00018843959315306374,
+      "loss": 0.208,
+      "step": 3514
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 2.3698058128356934,
+      "learning_rate": 0.0001884023815430414,
+      "loss": 0.422,
+      "step": 3515
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 3.7978971004486084,
+      "learning_rate": 0.0001883651699330191,
+      "loss": 1.4153,
+      "step": 3516
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 2.1298134326934814,
+      "learning_rate": 0.00018832795832299676,
+      "loss": 0.5128,
+      "step": 3517
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.7395635843276978,
+      "learning_rate": 0.00018829074671297441,
+      "loss": 0.1592,
+      "step": 3518
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.4399739503860474,
+      "learning_rate": 0.0001882535351029521,
+      "loss": 0.0962,
+      "step": 3519
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 2.0876457691192627,
+      "learning_rate": 0.0001882163234929298,
+      "loss": 0.5794,
+      "step": 3520
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 2.5795669555664062,
+      "learning_rate": 0.00018817911188290744,
+      "loss": 0.45,
+      "step": 3521
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.8200573921203613,
+      "learning_rate": 0.00018814190027288512,
+      "loss": 0.0697,
+      "step": 3522
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.8458431959152222,
+      "learning_rate": 0.00018810468866286282,
+      "loss": 0.1935,
+      "step": 3523
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.1191134452819824,
+      "learning_rate": 0.00018806747705284047,
+      "loss": 0.0881,
+      "step": 3524
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 4.453530311584473,
+      "learning_rate": 0.00018803026544281814,
+      "loss": 0.5468,
+      "step": 3525
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.0060765743255615,
+      "learning_rate": 0.0001879930538327958,
+      "loss": 0.122,
+      "step": 3526
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 2.2647805213928223,
+      "learning_rate": 0.0001879558422227735,
+      "loss": 0.1266,
+      "step": 3527
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 3.9961869716644287,
+      "learning_rate": 0.00018791863061275117,
+      "loss": 0.118,
+      "step": 3528
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.9355319142341614,
+      "learning_rate": 0.00018788141900272882,
+      "loss": 0.0574,
+      "step": 3529
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 3.0905239582061768,
+      "learning_rate": 0.00018784420739270652,
+      "loss": 1.0581,
+      "step": 3530
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 2.1791467666625977,
+      "learning_rate": 0.00018780699578268417,
+      "loss": 0.6661,
+      "step": 3531
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 2.5521254539489746,
+      "learning_rate": 0.00018776978417266185,
+      "loss": 0.4533,
+      "step": 3532
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 2.4842238426208496,
+      "learning_rate": 0.00018773257256263955,
+      "loss": 0.6154,
+      "step": 3533
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.9282567501068115,
+      "learning_rate": 0.0001876953609526172,
+      "loss": 0.3092,
+      "step": 3534
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 2.4563050270080566,
+      "learning_rate": 0.00018765814934259488,
+      "loss": 0.3415,
+      "step": 3535
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.9763050079345703,
+      "learning_rate": 0.00018762093773257255,
+      "loss": 0.4392,
+      "step": 3536
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 3.0104451179504395,
+      "learning_rate": 0.00018758372612255023,
+      "loss": 0.3869,
+      "step": 3537
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 2.270270824432373,
+      "learning_rate": 0.0001875465145125279,
+      "loss": 0.2269,
+      "step": 3538
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 4.0361008644104,
+      "learning_rate": 0.00018750930290250555,
+      "loss": 0.2886,
+      "step": 3539
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 2.7986207008361816,
+      "learning_rate": 0.00018747209129248323,
+      "loss": 0.4975,
+      "step": 3540
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 2.5085690021514893,
+      "learning_rate": 0.00018743487968246093,
+      "loss": 0.3265,
+      "step": 3541
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 2.8387558460235596,
+      "learning_rate": 0.00018739766807243858,
+      "loss": 0.2783,
+      "step": 3542
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 3.8573365211486816,
+      "learning_rate": 0.00018736045646241625,
+      "loss": 0.6503,
+      "step": 3543
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 4.418932914733887,
+      "learning_rate": 0.00018732324485239393,
+      "loss": 0.5577,
+      "step": 3544
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.9382532835006714,
+      "learning_rate": 0.0001872860332423716,
+      "loss": 0.4881,
+      "step": 3545
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 2.3090128898620605,
+      "learning_rate": 0.00018724882163234928,
+      "loss": 0.4069,
+      "step": 3546
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 2.0739097595214844,
+      "learning_rate": 0.00018721161002232693,
+      "loss": 0.3043,
+      "step": 3547
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 2.7553725242614746,
+      "learning_rate": 0.00018717439841230463,
+      "loss": 0.299,
+      "step": 3548
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 4.5599212646484375,
+      "learning_rate": 0.00018713718680228228,
+      "loss": 1.5577,
+      "step": 3549
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 4.312692642211914,
+      "learning_rate": 0.00018709997519225996,
+      "loss": 0.4333,
+      "step": 3550
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 2.080211877822876,
+      "learning_rate": 0.00018706276358223766,
+      "loss": 0.1648,
+      "step": 3551
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 4.01967191696167,
+      "learning_rate": 0.0001870255519722153,
+      "loss": 0.4049,
+      "step": 3552
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 2.141727924346924,
+      "learning_rate": 0.00018698834036219299,
+      "loss": 0.265,
+      "step": 3553
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.7477151155471802,
+      "learning_rate": 0.0001869511287521707,
+      "loss": 0.2846,
+      "step": 3554
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 2.3710789680480957,
+      "learning_rate": 0.00018691391714214834,
+      "loss": 0.4719,
+      "step": 3555
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.525882601737976,
+      "learning_rate": 0.000186876705532126,
+      "loss": 0.4626,
+      "step": 3556
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.5245758295059204,
+      "learning_rate": 0.00018683949392210366,
+      "loss": 0.367,
+      "step": 3557
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 3.189377784729004,
+      "learning_rate": 0.00018680228231208136,
+      "loss": 0.8119,
+      "step": 3558
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.608099102973938,
+      "learning_rate": 0.00018676507070205904,
+      "loss": 0.3054,
+      "step": 3559
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.2353789806365967,
+      "learning_rate": 0.0001867278590920367,
+      "loss": 0.1744,
+      "step": 3560
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.7148106098175049,
+      "learning_rate": 0.00018669064748201436,
+      "loss": 0.4698,
+      "step": 3561
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 2.018219470977783,
+      "learning_rate": 0.00018665343587199204,
+      "loss": 0.2835,
+      "step": 3562
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.603143572807312,
+      "learning_rate": 0.00018661622426196972,
+      "loss": 0.2637,
+      "step": 3563
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.5139524936676025,
+      "learning_rate": 0.0001865790126519474,
+      "loss": 0.3385,
+      "step": 3564
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 2.398815393447876,
+      "learning_rate": 0.00018654180104192507,
+      "loss": 0.195,
+      "step": 3565
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 2.6348488330841064,
+      "learning_rate": 0.00018650458943190274,
+      "loss": 0.8485,
+      "step": 3566
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 3.205677032470703,
+      "learning_rate": 0.00018646737782188042,
+      "loss": 0.6398,
+      "step": 3567
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.6468408107757568,
+      "learning_rate": 0.00018643016621185807,
+      "loss": 0.2491,
+      "step": 3568
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.7327412366867065,
+      "learning_rate": 0.00018639295460183577,
+      "loss": 0.2545,
+      "step": 3569
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 2.0780012607574463,
+      "learning_rate": 0.00018635574299181342,
+      "loss": 0.3401,
+      "step": 3570
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.9040396213531494,
+      "learning_rate": 0.0001863185313817911,
+      "loss": 0.2863,
+      "step": 3571
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.8664028644561768,
+      "learning_rate": 0.0001862813197717688,
+      "loss": 0.1769,
+      "step": 3572
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 2.1530826091766357,
+      "learning_rate": 0.00018624410816174645,
+      "loss": 0.2482,
+      "step": 3573
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 4.735560417175293,
+      "learning_rate": 0.00018620689655172412,
+      "loss": 1.8875,
+      "step": 3574
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.7484613656997681,
+      "learning_rate": 0.00018616968494170177,
+      "loss": 0.0572,
+      "step": 3575
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 2.2754640579223633,
+      "learning_rate": 0.00018613247333167947,
+      "loss": 0.12,
+      "step": 3576
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 2.0318729877471924,
+      "learning_rate": 0.00018609526172165715,
+      "loss": 0.0707,
+      "step": 3577
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 4.152217388153076,
+      "learning_rate": 0.0001860580501116348,
+      "loss": 1.4776,
+      "step": 3578
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 6.380149841308594,
+      "learning_rate": 0.0001860208385016125,
+      "loss": 0.3457,
+      "step": 3579
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 2.7456815242767334,
+      "learning_rate": 0.00018598362689159018,
+      "loss": 0.8919,
+      "step": 3580
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.7026749849319458,
+      "learning_rate": 0.00018594641528156783,
+      "loss": 0.424,
+      "step": 3581
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.725861668586731,
+      "learning_rate": 0.0001859092036715455,
+      "loss": 0.3761,
+      "step": 3582
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.5452537536621094,
+      "learning_rate": 0.00018587199206152318,
+      "loss": 0.3575,
+      "step": 3583
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 2.4589481353759766,
+      "learning_rate": 0.00018583478045150085,
+      "loss": 0.4449,
+      "step": 3584
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.2494957447052002,
+      "learning_rate": 0.00018579756884147853,
+      "loss": 0.2364,
+      "step": 3585
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.3550739288330078,
+      "learning_rate": 0.0001857603572314562,
+      "loss": 0.2196,
+      "step": 3586
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 2.515165090560913,
+      "learning_rate": 0.00018572314562143388,
+      "loss": 0.3064,
+      "step": 3587
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.4528539180755615,
+      "learning_rate": 0.00018568593401141153,
+      "loss": 0.2979,
+      "step": 3588
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 2.131699800491333,
+      "learning_rate": 0.0001856487224013892,
+      "loss": 0.3577,
+      "step": 3589
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 2.56373929977417,
+      "learning_rate": 0.0001856115107913669,
+      "loss": 0.6694,
+      "step": 3590
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 3.1758902072906494,
+      "learning_rate": 0.00018557429918134456,
+      "loss": 0.7342,
+      "step": 3591
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.6790735721588135,
+      "learning_rate": 0.00018553708757132223,
+      "loss": 0.2676,
+      "step": 3592
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 3.171738624572754,
+      "learning_rate": 0.00018549987596129988,
+      "loss": 0.6191,
+      "step": 3593
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 3.72245454788208,
+      "learning_rate": 0.00018546266435127758,
+      "loss": 0.9003,
+      "step": 3594
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 3.158806085586548,
+      "learning_rate": 0.00018542545274125526,
+      "loss": 0.5151,
+      "step": 3595
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 2.433006763458252,
+      "learning_rate": 0.0001853882411312329,
+      "loss": 0.3028,
+      "step": 3596
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 2.634091377258301,
+      "learning_rate": 0.0001853510295212106,
+      "loss": 0.3578,
+      "step": 3597
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.5862232446670532,
+      "learning_rate": 0.0001853138179111883,
+      "loss": 0.2275,
+      "step": 3598
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 3.65346097946167,
+      "learning_rate": 0.00018527660630116594,
+      "loss": 0.4811,
+      "step": 3599
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 2.1591145992279053,
+      "learning_rate": 0.00018523939469114364,
+      "loss": 0.3709,
+      "step": 3600
+    },
+    {
+      "epoch": 1.26,
+      "eval_loss": 0.5617594122886658,
+      "eval_runtime": 51.0249,
+      "eval_samples_per_second": 42.489,
+      "eval_steps_per_second": 10.622,
+      "eval_wer": 0.5672029060716139,
+      "step": 3600
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 4.291797161102295,
+      "learning_rate": 0.0001852021830811213,
+      "loss": 0.7502,
+      "step": 3601
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 2.1709883213043213,
+      "learning_rate": 0.00018516497147109896,
+      "loss": 0.6292,
+      "step": 3602
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 2.368156909942627,
+      "learning_rate": 0.00018512775986107664,
+      "loss": 0.3489,
+      "step": 3603
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 4.229372024536133,
+      "learning_rate": 0.00018509054825105432,
+      "loss": 1.2459,
+      "step": 3604
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 2.126002788543701,
+      "learning_rate": 0.000185053336641032,
+      "loss": 0.7528,
+      "step": 3605
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 2.2286715507507324,
+      "learning_rate": 0.00018501612503100964,
+      "loss": 0.6335,
+      "step": 3606
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 2.8708016872406006,
+      "learning_rate": 0.00018497891342098734,
+      "loss": 0.6489,
+      "step": 3607
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.7687532901763916,
+      "learning_rate": 0.00018494170181096502,
+      "loss": 0.6581,
+      "step": 3608
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.8210864067077637,
+      "learning_rate": 0.00018490449020094267,
+      "loss": 0.4258,
+      "step": 3609
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.6010420322418213,
+      "learning_rate": 0.00018486727859092034,
+      "loss": 0.2422,
+      "step": 3610
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.8562211990356445,
+      "learning_rate": 0.00018483006698089805,
+      "loss": 0.3653,
+      "step": 3611
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 2.263174533843994,
+      "learning_rate": 0.0001847928553708757,
+      "loss": 0.5212,
+      "step": 3612
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 2.848961591720581,
+      "learning_rate": 0.00018475564376085337,
+      "loss": 0.383,
+      "step": 3613
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 4.881875514984131,
+      "learning_rate": 0.00018471843215083102,
+      "loss": 0.6873,
+      "step": 3614
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 3.0648579597473145,
+      "learning_rate": 0.00018468122054080872,
+      "loss": 0.8545,
+      "step": 3615
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 2.5609729290008545,
+      "learning_rate": 0.0001846440089307864,
+      "loss": 0.3276,
+      "step": 3616
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 2.5366320610046387,
+      "learning_rate": 0.00018460679732076405,
+      "loss": 0.3974,
+      "step": 3617
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.6741987466812134,
+      "learning_rate": 0.00018456958571074175,
+      "loss": 0.3688,
+      "step": 3618
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 2.0672824382781982,
+      "learning_rate": 0.0001845323741007194,
+      "loss": 0.2186,
+      "step": 3619
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 3.1901285648345947,
+      "learning_rate": 0.00018449516249069707,
+      "loss": 0.3451,
+      "step": 3620
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 2.1494998931884766,
+      "learning_rate": 0.00018445795088067478,
+      "loss": 0.2902,
+      "step": 3621
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 5.410013675689697,
+      "learning_rate": 0.00018442073927065243,
+      "loss": 0.2924,
+      "step": 3622
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.7226850986480713,
+      "learning_rate": 0.0001843835276606301,
+      "loss": 0.2274,
+      "step": 3623
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.2159602642059326,
+      "learning_rate": 0.00018434631605060778,
+      "loss": 0.0521,
+      "step": 3624
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 6.059643268585205,
+      "learning_rate": 0.00018430910444058545,
+      "loss": 0.7462,
+      "step": 3625
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.91728937625885,
+      "learning_rate": 0.00018427189283056313,
+      "loss": 0.1399,
+      "step": 3626
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 2.763751745223999,
+      "learning_rate": 0.00018423468122054078,
+      "loss": 0.9764,
+      "step": 3627
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 3.757922887802124,
+      "learning_rate": 0.00018419746961051848,
+      "loss": 0.4835,
+      "step": 3628
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 4.211630821228027,
+      "learning_rate": 0.00018416025800049616,
+      "loss": 0.4143,
+      "step": 3629
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 4.662005424499512,
+      "learning_rate": 0.0001841230463904738,
+      "loss": 0.9232,
+      "step": 3630
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.8224451541900635,
+      "learning_rate": 0.00018408583478045148,
+      "loss": 0.2985,
+      "step": 3631
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.7688192129135132,
+      "learning_rate": 0.00018404862317042916,
+      "loss": 0.2334,
+      "step": 3632
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.983191967010498,
+      "learning_rate": 0.00018401141156040683,
+      "loss": 0.3987,
+      "step": 3633
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 2.6247076988220215,
+      "learning_rate": 0.0001839741999503845,
+      "loss": 0.5811,
+      "step": 3634
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.944804310798645,
+      "learning_rate": 0.00018393698834036216,
+      "loss": 0.5739,
+      "step": 3635
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.3238880634307861,
+      "learning_rate": 0.00018389977673033986,
+      "loss": 0.2322,
+      "step": 3636
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 2.190138339996338,
+      "learning_rate": 0.0001838625651203175,
+      "loss": 0.4987,
+      "step": 3637
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.9091345071792603,
+      "learning_rate": 0.00018382535351029518,
+      "loss": 0.3517,
+      "step": 3638
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.6717495918273926,
+      "learning_rate": 0.0001837881419002729,
+      "loss": 0.1578,
+      "step": 3639
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 2.4669203758239746,
+      "learning_rate": 0.00018375093029025054,
+      "loss": 0.3245,
+      "step": 3640
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.4179534912109375,
+      "learning_rate": 0.0001837137186802282,
+      "loss": 0.2772,
+      "step": 3641
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 2.291625738143921,
+      "learning_rate": 0.00018367650707020591,
+      "loss": 0.7771,
+      "step": 3642
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 2.39475417137146,
+      "learning_rate": 0.00018363929546018356,
+      "loss": 0.3678,
+      "step": 3643
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.2829241752624512,
+      "learning_rate": 0.00018360208385016124,
+      "loss": 0.3058,
+      "step": 3644
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.9139297008514404,
+      "learning_rate": 0.0001835648722401389,
+      "loss": 0.1881,
+      "step": 3645
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 4.035134792327881,
+      "learning_rate": 0.0001835276606301166,
+      "loss": 0.2359,
+      "step": 3646
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 3.865679979324341,
+      "learning_rate": 0.00018349044902009427,
+      "loss": 0.4176,
+      "step": 3647
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 7.729085445404053,
+      "learning_rate": 0.00018345323741007191,
+      "loss": 1.612,
+      "step": 3648
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.2197033166885376,
+      "learning_rate": 0.00018341602580004962,
+      "loss": 0.1282,
+      "step": 3649
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.467267394065857,
+      "learning_rate": 0.00018337881419002727,
+      "loss": 0.1326,
+      "step": 3650
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 4.615586757659912,
+      "learning_rate": 0.00018334160258000494,
+      "loss": 0.674,
+      "step": 3651
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.724073052406311,
+      "learning_rate": 0.00018330439096998262,
+      "loss": 0.268,
+      "step": 3652
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 3.465282440185547,
+      "learning_rate": 0.0001832671793599603,
+      "loss": 0.5606,
+      "step": 3653
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 2.1217899322509766,
+      "learning_rate": 0.00018322996774993797,
+      "loss": 0.1697,
+      "step": 3654
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.2871822118759155,
+      "learning_rate": 0.00018319275613991565,
+      "loss": 0.4938,
+      "step": 3655
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 2.0756819248199463,
+      "learning_rate": 0.0001831555445298933,
+      "loss": 0.5783,
+      "step": 3656
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 3.0986368656158447,
+      "learning_rate": 0.000183118332919871,
+      "loss": 0.5325,
+      "step": 3657
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.4617551565170288,
+      "learning_rate": 0.00018308112130984865,
+      "loss": 0.3135,
+      "step": 3658
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.108564019203186,
+      "learning_rate": 0.00018304390969982632,
+      "loss": 0.1062,
+      "step": 3659
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 3.5125279426574707,
+      "learning_rate": 0.00018300669808980402,
+      "loss": 0.6693,
+      "step": 3660
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 2.014686346054077,
+      "learning_rate": 0.00018296948647978167,
+      "loss": 0.4059,
+      "step": 3661
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.8392510414123535,
+      "learning_rate": 0.00018293227486975935,
+      "loss": 0.4116,
+      "step": 3662
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 2.073396921157837,
+      "learning_rate": 0.000182895063259737,
+      "loss": 0.3753,
+      "step": 3663
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 2.183483362197876,
+      "learning_rate": 0.0001828578516497147,
+      "loss": 0.5035,
+      "step": 3664
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 2.8279621601104736,
+      "learning_rate": 0.00018282064003969238,
+      "loss": 0.3977,
+      "step": 3665
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 2.0163662433624268,
+      "learning_rate": 0.00018278342842967002,
+      "loss": 0.2057,
+      "step": 3666
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 2.163269519805908,
+      "learning_rate": 0.00018274621681964773,
+      "loss": 0.5138,
+      "step": 3667
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.980743169784546,
+      "learning_rate": 0.0001827090052096254,
+      "loss": 0.3655,
+      "step": 3668
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 2.201828956604004,
+      "learning_rate": 0.00018267179359960305,
+      "loss": 0.6443,
+      "step": 3669
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.6139568090438843,
+      "learning_rate": 0.00018263458198958075,
+      "loss": 0.1735,
+      "step": 3670
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 3.5710642337799072,
+      "learning_rate": 0.0001825973703795584,
+      "loss": 0.6014,
+      "step": 3671
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.6693230867385864,
+      "learning_rate": 0.00018256015876953608,
+      "loss": 0.3088,
+      "step": 3672
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 2.734638214111328,
+      "learning_rate": 0.00018252294715951376,
+      "loss": 0.4733,
+      "step": 3673
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 3.7495346069335938,
+      "learning_rate": 0.00018248573554949143,
+      "loss": 0.6562,
+      "step": 3674
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.6632982492446899,
+      "learning_rate": 0.0001824485239394691,
+      "loss": 0.0649,
+      "step": 3675
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 2.975598096847534,
+      "learning_rate": 0.00018241131232944676,
+      "loss": 0.3478,
+      "step": 3676
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.9812325239181519,
+      "learning_rate": 0.00018237410071942443,
+      "loss": 0.3126,
+      "step": 3677
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 4.1589274406433105,
+      "learning_rate": 0.00018233688910940213,
+      "loss": 0.3591,
+      "step": 3678
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.129032015800476,
+      "learning_rate": 0.00018229967749937978,
+      "loss": 0.0812,
+      "step": 3679
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.9920796155929565,
+      "learning_rate": 0.00018226246588935746,
+      "loss": 1.0188,
+      "step": 3680
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 2.480633020401001,
+      "learning_rate": 0.0001822252542793351,
+      "loss": 0.8319,
+      "step": 3681
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.5719470977783203,
+      "learning_rate": 0.0001821880426693128,
+      "loss": 0.446,
+      "step": 3682
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 2.4275617599487305,
+      "learning_rate": 0.00018215083105929049,
+      "loss": 0.601,
+      "step": 3683
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.5678902864456177,
+      "learning_rate": 0.00018211361944926813,
+      "loss": 0.326,
+      "step": 3684
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.8893762826919556,
+      "learning_rate": 0.00018207640783924584,
+      "loss": 0.3836,
+      "step": 3685
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.8527387380599976,
+      "learning_rate": 0.0001820391962292235,
+      "loss": 0.3873,
+      "step": 3686
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 2.947082757949829,
+      "learning_rate": 0.00018200198461920116,
+      "loss": 0.2627,
+      "step": 3687
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.0232598781585693,
+      "learning_rate": 0.00018196477300917886,
+      "loss": 0.1291,
+      "step": 3688
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.9064313173294067,
+      "learning_rate": 0.00018192756139915651,
+      "loss": 0.2368,
+      "step": 3689
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.997897982597351,
+      "learning_rate": 0.0001818903497891342,
+      "loss": 0.17,
+      "step": 3690
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 8.157362937927246,
+      "learning_rate": 0.0001818531381791119,
+      "loss": 1.9281,
+      "step": 3691
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 4.2168660163879395,
+      "learning_rate": 0.00018181592656908954,
+      "loss": 0.4399,
+      "step": 3692
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 2.208404064178467,
+      "learning_rate": 0.00018177871495906722,
+      "loss": 0.1804,
+      "step": 3693
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 2.2321760654449463,
+      "learning_rate": 0.00018174150334904487,
+      "loss": 0.2453,
+      "step": 3694
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.9782792329788208,
+      "learning_rate": 0.00018170429173902257,
+      "loss": 0.1628,
+      "step": 3695
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 2.479062795639038,
+      "learning_rate": 0.00018166708012900024,
+      "loss": 0.3915,
+      "step": 3696
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.956723928451538,
+      "learning_rate": 0.0001816298685189779,
+      "loss": 0.2848,
+      "step": 3697
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 4.0798211097717285,
+      "learning_rate": 0.00018159265690895557,
+      "loss": 1.0078,
+      "step": 3698
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 2.6179561614990234,
+      "learning_rate": 0.00018155544529893327,
+      "loss": 0.3553,
+      "step": 3699
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 2.044512987136841,
+      "learning_rate": 0.00018151823368891092,
+      "loss": 0.1905,
+      "step": 3700
+    },
+    {
+      "epoch": 1.3,
+      "eval_loss": 0.46060672402381897,
+      "eval_runtime": 51.537,
+      "eval_samples_per_second": 42.067,
+      "eval_steps_per_second": 10.517,
+      "eval_wer": 0.44300294066770457,
+      "step": 3700
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 3.8095672130584717,
+      "learning_rate": 0.0001814810220788886,
+      "loss": 1.5776,
+      "step": 3701
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 4.374634265899658,
+      "learning_rate": 0.00018144381046886624,
+      "loss": 1.4929,
+      "step": 3702
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.16568922996521,
+      "learning_rate": 0.00018140659885884395,
+      "loss": 0.1051,
+      "step": 3703
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.5767412185668945,
+      "learning_rate": 0.00018136938724882162,
+      "loss": 0.1765,
+      "step": 3704
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 4.894369125366211,
+      "learning_rate": 0.00018133217563879927,
+      "loss": 0.9859,
+      "step": 3705
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 3.5991573333740234,
+      "learning_rate": 0.00018129496402877698,
+      "loss": 0.5492,
+      "step": 3706
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 4.314338684082031,
+      "learning_rate": 0.00018125775241875462,
+      "loss": 0.9102,
+      "step": 3707
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 2.0932729244232178,
+      "learning_rate": 0.0001812205408087323,
+      "loss": 0.5376,
+      "step": 3708
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.41013765335083,
+      "learning_rate": 0.00018118332919871,
+      "loss": 0.2054,
+      "step": 3709
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 2.404813528060913,
+      "learning_rate": 0.00018114611758868765,
+      "loss": 0.2452,
+      "step": 3710
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.5884853601455688,
+      "learning_rate": 0.00018110890597866533,
+      "loss": 0.233,
+      "step": 3711
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.3731337785720825,
+      "learning_rate": 0.00018107169436864303,
+      "loss": 0.1386,
+      "step": 3712
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 3.145470142364502,
+      "learning_rate": 0.00018103448275862068,
+      "loss": 0.6139,
+      "step": 3713
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.8240619897842407,
+      "learning_rate": 0.00018099727114859835,
+      "loss": 0.4299,
+      "step": 3714
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.6262075901031494,
+      "learning_rate": 0.000180960059538576,
+      "loss": 0.2562,
+      "step": 3715
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 2.166367769241333,
+      "learning_rate": 0.0001809228479285537,
+      "loss": 0.3563,
+      "step": 3716
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 2.9389724731445312,
+      "learning_rate": 0.00018088563631853138,
+      "loss": 1.4989,
+      "step": 3717
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.0228006839752197,
+      "learning_rate": 0.00018084842470850903,
+      "loss": 0.0534,
+      "step": 3718
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 3.372701406478882,
+      "learning_rate": 0.0001808112130984867,
+      "loss": 0.4041,
+      "step": 3719
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 2.5104494094848633,
+      "learning_rate": 0.00018077400148846438,
+      "loss": 0.3649,
+      "step": 3720
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 3.1328682899475098,
+      "learning_rate": 0.00018073678987844206,
+      "loss": 0.5192,
+      "step": 3721
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.5079821348190308,
+      "learning_rate": 0.00018069957826841973,
+      "loss": 0.1187,
+      "step": 3722
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 2.341155529022217,
+      "learning_rate": 0.00018066236665839738,
+      "loss": 0.2129,
+      "step": 3723
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 3.211752414703369,
+      "learning_rate": 0.00018062515504837509,
+      "loss": 0.3416,
+      "step": 3724
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.5088387727737427,
+      "learning_rate": 0.00018058794343835276,
+      "loss": 0.2176,
+      "step": 3725
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 2.9890217781066895,
+      "learning_rate": 0.0001805507318283304,
+      "loss": 0.437,
+      "step": 3726
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.218353271484375,
+      "learning_rate": 0.0001805135202183081,
+      "loss": 0.1558,
+      "step": 3727
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 2.0937466621398926,
+      "learning_rate": 0.00018047630860828576,
+      "loss": 0.3612,
+      "step": 3728
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 2.505051612854004,
+      "learning_rate": 0.00018043909699826344,
+      "loss": 0.4007,
+      "step": 3729
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 2.501373529434204,
+      "learning_rate": 0.00018040188538824114,
+      "loss": 0.8542,
+      "step": 3730
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.9108080863952637,
+      "learning_rate": 0.0001803646737782188,
+      "loss": 0.6732,
+      "step": 3731
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.495678424835205,
+      "learning_rate": 0.00018032746216819646,
+      "loss": 0.2758,
+      "step": 3732
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.6871167421340942,
+      "learning_rate": 0.0001802902505581741,
+      "loss": 0.2974,
+      "step": 3733
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.7430815696716309,
+      "learning_rate": 0.00018025303894815182,
+      "loss": 0.3686,
+      "step": 3734
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 3.15686297416687,
+      "learning_rate": 0.0001802158273381295,
+      "loss": 0.4121,
+      "step": 3735
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.959276556968689,
+      "learning_rate": 0.00018017861572810714,
+      "loss": 0.0906,
+      "step": 3736
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.973922610282898,
+      "learning_rate": 0.00018014140411808484,
+      "loss": 0.3729,
+      "step": 3737
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.6108390092849731,
+      "learning_rate": 0.0001801041925080625,
+      "loss": 0.2135,
+      "step": 3738
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.7713210582733154,
+      "learning_rate": 0.00018006698089804017,
+      "loss": 0.447,
+      "step": 3739
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 2.696568727493286,
+      "learning_rate": 0.00018002976928801784,
+      "loss": 0.3003,
+      "step": 3740
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.4664280414581299,
+      "learning_rate": 0.00017999255767799552,
+      "loss": 0.174,
+      "step": 3741
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 3.218578815460205,
+      "learning_rate": 0.0001799553460679732,
+      "loss": 0.2536,
+      "step": 3742
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 3.3548967838287354,
+      "learning_rate": 0.00017991813445795087,
+      "loss": 0.3933,
+      "step": 3743
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 3.9141340255737305,
+      "learning_rate": 0.00017988092284792852,
+      "loss": 0.6092,
+      "step": 3744
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 2.043869972229004,
+      "learning_rate": 0.00017984371123790622,
+      "loss": 0.3214,
+      "step": 3745
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 2.997690200805664,
+      "learning_rate": 0.00017980649962788387,
+      "loss": 1.3812,
+      "step": 3746
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.9347894191741943,
+      "learning_rate": 0.00017976928801786155,
+      "loss": 0.185,
+      "step": 3747
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 2.9342029094696045,
+      "learning_rate": 0.00017973207640783925,
+      "loss": 0.5739,
+      "step": 3748
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 2.7313919067382812,
+      "learning_rate": 0.0001796948647978169,
+      "loss": 0.3957,
+      "step": 3749
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 3.2632312774658203,
+      "learning_rate": 0.00017965765318779457,
+      "loss": 0.1384,
+      "step": 3750
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 2.596008539199829,
+      "learning_rate": 0.00017962044157777222,
+      "loss": 0.4888,
+      "step": 3751
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 2.564819574356079,
+      "learning_rate": 0.00017958322996774993,
+      "loss": 0.2528,
+      "step": 3752
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 2.4152514934539795,
+      "learning_rate": 0.0001795460183577276,
+      "loss": 0.2479,
+      "step": 3753
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.620201587677002,
+      "learning_rate": 0.00017950880674770525,
+      "loss": 0.1761,
+      "step": 3754
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 2.0693604946136475,
+      "learning_rate": 0.00017947159513768295,
+      "loss": 0.9116,
+      "step": 3755
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.889093041419983,
+      "learning_rate": 0.00017943438352766063,
+      "loss": 0.417,
+      "step": 3756
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 2.0303893089294434,
+      "learning_rate": 0.00017939717191763828,
+      "loss": 0.3572,
+      "step": 3757
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.2891769409179688,
+      "learning_rate": 0.00017935996030761598,
+      "loss": 0.1608,
+      "step": 3758
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.4122023582458496,
+      "learning_rate": 0.00017932274869759363,
+      "loss": 0.2832,
+      "step": 3759
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 2.285872459411621,
+      "learning_rate": 0.0001792855370875713,
+      "loss": 0.4338,
+      "step": 3760
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.5522655248641968,
+      "learning_rate": 0.00017924832547754898,
+      "loss": 0.3917,
+      "step": 3761
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 5.722376823425293,
+      "learning_rate": 0.00017921111386752666,
+      "loss": 0.201,
+      "step": 3762
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 2.5768380165100098,
+      "learning_rate": 0.00017917390225750433,
+      "loss": 0.2258,
+      "step": 3763
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 3.3789002895355225,
+      "learning_rate": 0.00017913669064748198,
+      "loss": 0.6568,
+      "step": 3764
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 2.571653127670288,
+      "learning_rate": 0.00017909947903745966,
+      "loss": 0.3319,
+      "step": 3765
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 2.023444414138794,
+      "learning_rate": 0.00017906226742743736,
+      "loss": 0.2127,
+      "step": 3766
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 2.665778875350952,
+      "learning_rate": 0.000179025055817415,
+      "loss": 0.3803,
+      "step": 3767
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 2.8000216484069824,
+      "learning_rate": 0.00017898784420739268,
+      "loss": 0.3656,
+      "step": 3768
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 5.576625347137451,
+      "learning_rate": 0.0001789506325973704,
+      "loss": 2.0108,
+      "step": 3769
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 4.2713165283203125,
+      "learning_rate": 0.00017891342098734804,
+      "loss": 0.5785,
+      "step": 3770
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 2.837256908416748,
+      "learning_rate": 0.0001788762093773257,
+      "loss": 0.744,
+      "step": 3771
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 2.32977557182312,
+      "learning_rate": 0.00017883899776730336,
+      "loss": 0.1912,
+      "step": 3772
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.8392449617385864,
+      "learning_rate": 0.00017880178615728106,
+      "loss": 0.4553,
+      "step": 3773
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 2.949434757232666,
+      "learning_rate": 0.00017876457454725874,
+      "loss": 0.2178,
+      "step": 3774
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.4645886421203613,
+      "learning_rate": 0.0001787273629372364,
+      "loss": 0.1202,
+      "step": 3775
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.1854076385498047,
+      "learning_rate": 0.0001786901513272141,
+      "loss": 0.1576,
+      "step": 3776
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.7854406833648682,
+      "learning_rate": 0.00017865293971719174,
+      "loss": 0.2579,
+      "step": 3777
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 3.848768472671509,
+      "learning_rate": 0.00017861572810716942,
+      "loss": 1.4119,
+      "step": 3778
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 2.103977918624878,
+      "learning_rate": 0.00017857851649714712,
+      "loss": 0.2911,
+      "step": 3779
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 2.866246461868286,
+      "learning_rate": 0.00017854130488712477,
+      "loss": 0.9159,
+      "step": 3780
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.2970139980316162,
+      "learning_rate": 0.00017850409327710244,
+      "loss": 0.4276,
+      "step": 3781
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.2081961631774902,
+      "learning_rate": 0.0001784668816670801,
+      "loss": 0.3139,
+      "step": 3782
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 2.1453487873077393,
+      "learning_rate": 0.0001784296700570578,
+      "loss": 0.7616,
+      "step": 3783
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.9024728536605835,
+      "learning_rate": 0.00017839245844703547,
+      "loss": 0.2492,
+      "step": 3784
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.2727264165878296,
+      "learning_rate": 0.00017835524683701312,
+      "loss": 0.2043,
+      "step": 3785
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.4936034679412842,
+      "learning_rate": 0.0001783180352269908,
+      "loss": 0.3883,
+      "step": 3786
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.377806544303894,
+      "learning_rate": 0.0001782808236169685,
+      "loss": 0.273,
+      "step": 3787
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 2.399864435195923,
+      "learning_rate": 0.00017824361200694615,
+      "loss": 0.3119,
+      "step": 3788
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.5656849145889282,
+      "learning_rate": 0.00017820640039692382,
+      "loss": 0.2332,
+      "step": 3789
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 15.384991645812988,
+      "learning_rate": 0.0001781691887869015,
+      "loss": 5.3366,
+      "step": 3790
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 3.5615835189819336,
+      "learning_rate": 0.00017813197717687917,
+      "loss": 0.8114,
+      "step": 3791
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 2.1784439086914062,
+      "learning_rate": 0.00017809476556685685,
+      "loss": 0.5916,
+      "step": 3792
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.5509488582611084,
+      "learning_rate": 0.0001780575539568345,
+      "loss": 0.2626,
+      "step": 3793
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.8482944369316101,
+      "learning_rate": 0.0001780203423468122,
+      "loss": 0.066,
+      "step": 3794
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.3398417234420776,
+      "learning_rate": 0.00017798313073678985,
+      "loss": 0.1493,
+      "step": 3795
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.738661766052246,
+      "learning_rate": 0.00017794591912676753,
+      "loss": 0.198,
+      "step": 3796
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 4.550568580627441,
+      "learning_rate": 0.00017790870751674523,
+      "loss": 1.3028,
+      "step": 3797
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 2.300915002822876,
+      "learning_rate": 0.00017787149590672288,
+      "loss": 0.2393,
+      "step": 3798
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.3625463247299194,
+      "learning_rate": 0.00017783428429670055,
+      "loss": 0.1016,
+      "step": 3799
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 2.7032511234283447,
+      "learning_rate": 0.00017779707268667826,
+      "loss": 0.4724,
+      "step": 3800
+    },
+    {
+      "epoch": 1.33,
+      "eval_loss": 0.49983254075050354,
+      "eval_runtime": 51.5096,
+      "eval_samples_per_second": 42.089,
+      "eval_steps_per_second": 10.522,
+      "eval_wer": 0.4850371907974399,
+      "step": 3800
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 3.71159029006958,
+      "learning_rate": 0.0001777598610766559,
+      "loss": 0.169,
+      "step": 3801
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 2.478241205215454,
+      "learning_rate": 0.00017772264946663358,
+      "loss": 0.2778,
+      "step": 3802
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 3.812366247177124,
+      "learning_rate": 0.00017768543785661123,
+      "loss": 0.4634,
+      "step": 3803
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.6730952262878418,
+      "learning_rate": 0.00017764822624658893,
+      "loss": 0.0989,
+      "step": 3804
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 3.086575508117676,
+      "learning_rate": 0.0001776110146365666,
+      "loss": 0.9503,
+      "step": 3805
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.9270522594451904,
+      "learning_rate": 0.00017757380302654426,
+      "loss": 0.7548,
+      "step": 3806
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.6181902885437012,
+      "learning_rate": 0.00017753659141652193,
+      "loss": 0.3449,
+      "step": 3807
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 2.112922430038452,
+      "learning_rate": 0.0001774993798064996,
+      "loss": 0.4004,
+      "step": 3808
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 2.618039846420288,
+      "learning_rate": 0.00017746216819647728,
+      "loss": 0.4503,
+      "step": 3809
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 2.6346006393432617,
+      "learning_rate": 0.00017742495658645496,
+      "loss": 0.6283,
+      "step": 3810
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 2.166271448135376,
+      "learning_rate": 0.00017738774497643264,
+      "loss": 0.21,
+      "step": 3811
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.9964041709899902,
+      "learning_rate": 0.0001773505333664103,
+      "loss": 0.1659,
+      "step": 3812
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.5653916597366333,
+      "learning_rate": 0.000177313321756388,
+      "loss": 0.2786,
+      "step": 3813
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 3.1856682300567627,
+      "learning_rate": 0.00017727611014636564,
+      "loss": 0.3468,
+      "step": 3814
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.068045973777771,
+      "learning_rate": 0.00017723889853634334,
+      "loss": 0.1729,
+      "step": 3815
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 2.467663526535034,
+      "learning_rate": 0.000177201686926321,
+      "loss": 0.3148,
+      "step": 3816
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 2.52803373336792,
+      "learning_rate": 0.00017716447531629866,
+      "loss": 0.6002,
+      "step": 3817
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.5784509181976318,
+      "learning_rate": 0.00017712726370627637,
+      "loss": 0.1783,
+      "step": 3818
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 5.923018455505371,
+      "learning_rate": 0.00017709005209625401,
+      "loss": 0.2747,
+      "step": 3819
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.4094946384429932,
+      "learning_rate": 0.0001770528404862317,
+      "loss": 0.0858,
+      "step": 3820
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.2953375577926636,
+      "learning_rate": 0.00017701562887620934,
+      "loss": 0.0638,
+      "step": 3821
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 4.134175777435303,
+      "learning_rate": 0.00017697841726618704,
+      "loss": 0.4833,
+      "step": 3822
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 4.836775302886963,
+      "learning_rate": 0.00017694120565616472,
+      "loss": 0.8205,
+      "step": 3823
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 3.950936794281006,
+      "learning_rate": 0.00017690399404614237,
+      "loss": 0.4628,
+      "step": 3824
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 2.5039966106414795,
+      "learning_rate": 0.00017686678243612007,
+      "loss": 0.2418,
+      "step": 3825
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.359667420387268,
+      "learning_rate": 0.00017682957082609772,
+      "loss": 0.1673,
+      "step": 3826
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 3.17059326171875,
+      "learning_rate": 0.0001767923592160754,
+      "loss": 0.2301,
+      "step": 3827
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 3.165222406387329,
+      "learning_rate": 0.00017675514760605307,
+      "loss": 0.2882,
+      "step": 3828
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 2.6665470600128174,
+      "learning_rate": 0.00017671793599603075,
+      "loss": 0.1528,
+      "step": 3829
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 3.1193861961364746,
+      "learning_rate": 0.00017668072438600842,
+      "loss": 0.8316,
+      "step": 3830
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 3.0916197299957275,
+      "learning_rate": 0.0001766435127759861,
+      "loss": 0.7043,
+      "step": 3831
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 2.5814852714538574,
+      "learning_rate": 0.00017660630116596377,
+      "loss": 0.8129,
+      "step": 3832
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 2.543747663497925,
+      "learning_rate": 0.00017656908955594145,
+      "loss": 0.3409,
+      "step": 3833
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 4.633474826812744,
+      "learning_rate": 0.0001765318779459191,
+      "loss": 1.6709,
+      "step": 3834
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.9206253290176392,
+      "learning_rate": 0.00017649466633589677,
+      "loss": 0.4245,
+      "step": 3835
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 2.735995054244995,
+      "learning_rate": 0.00017645745472587448,
+      "loss": 0.3843,
+      "step": 3836
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.561367392539978,
+      "learning_rate": 0.00017642024311585212,
+      "loss": 0.3513,
+      "step": 3837
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 2.5527138710021973,
+      "learning_rate": 0.0001763830315058298,
+      "loss": 0.2287,
+      "step": 3838
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 2.1322858333587646,
+      "learning_rate": 0.00017634581989580745,
+      "loss": 0.465,
+      "step": 3839
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 3.6257131099700928,
+      "learning_rate": 0.00017630860828578515,
+      "loss": 0.6081,
+      "step": 3840
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 2.3494338989257812,
+      "learning_rate": 0.00017627139667576283,
+      "loss": 0.5918,
+      "step": 3841
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.8620758056640625,
+      "learning_rate": 0.00017623418506574048,
+      "loss": 0.3641,
+      "step": 3842
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.5785250663757324,
+      "learning_rate": 0.00017619697345571818,
+      "loss": 0.1248,
+      "step": 3843
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.321878433227539,
+      "learning_rate": 0.00017615976184569586,
+      "loss": 0.1409,
+      "step": 3844
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.8503745794296265,
+      "learning_rate": 0.0001761225502356735,
+      "loss": 0.2243,
+      "step": 3845
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.8740498423576355,
+      "learning_rate": 0.0001760853386256512,
+      "loss": 0.0779,
+      "step": 3846
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 2.1911394596099854,
+      "learning_rate": 0.00017604812701562886,
+      "loss": 0.2616,
+      "step": 3847
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.2762500047683716,
+      "learning_rate": 0.00017601091540560653,
+      "loss": 0.0597,
+      "step": 3848
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 3.6884820461273193,
+      "learning_rate": 0.0001759737037955842,
+      "loss": 0.4311,
+      "step": 3849
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.9794673919677734,
+      "learning_rate": 0.00017593649218556188,
+      "loss": 0.1456,
+      "step": 3850
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 18.375457763671875,
+      "learning_rate": 0.00017589928057553956,
+      "loss": 0.7519,
+      "step": 3851
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 9.318000793457031,
+      "learning_rate": 0.0001758620689655172,
+      "loss": 1.2229,
+      "step": 3852
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.8230860233306885,
+      "learning_rate": 0.0001758248573554949,
+      "loss": 0.0594,
+      "step": 3853
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 3.9834702014923096,
+      "learning_rate": 0.00017578764574547259,
+      "loss": 1.0047,
+      "step": 3854
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 2.5470404624938965,
+      "learning_rate": 0.00017575043413545023,
+      "loss": 0.7612,
+      "step": 3855
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 2.304075241088867,
+      "learning_rate": 0.0001757132225254279,
+      "loss": 0.4069,
+      "step": 3856
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 2.3937854766845703,
+      "learning_rate": 0.0001756760109154056,
+      "loss": 0.4114,
+      "step": 3857
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.7787152528762817,
+      "learning_rate": 0.00017563879930538326,
+      "loss": 0.3035,
+      "step": 3858
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.8002721071243286,
+      "learning_rate": 0.00017560158769536094,
+      "loss": 0.6163,
+      "step": 3859
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.9406236410140991,
+      "learning_rate": 0.0001755643760853386,
+      "loss": 0.2896,
+      "step": 3860
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 2.4129421710968018,
+      "learning_rate": 0.0001755271644753163,
+      "loss": 0.6502,
+      "step": 3861
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 2.7634332180023193,
+      "learning_rate": 0.00017548995286529397,
+      "loss": 0.6027,
+      "step": 3862
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.0977592468261719,
+      "learning_rate": 0.00017545274125527161,
+      "loss": 0.1788,
+      "step": 3863
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 3.4504082202911377,
+      "learning_rate": 0.00017541552964524932,
+      "loss": 0.3887,
+      "step": 3864
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.857100009918213,
+      "learning_rate": 0.00017537831803522697,
+      "loss": 0.2107,
+      "step": 3865
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 4.293605327606201,
+      "learning_rate": 0.00017534110642520464,
+      "loss": 1.4237,
+      "step": 3866
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 2.1752376556396484,
+      "learning_rate": 0.00017530389481518234,
+      "loss": 0.3186,
+      "step": 3867
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 2.9012653827667236,
+      "learning_rate": 0.00017526668320516,
+      "loss": 0.3952,
+      "step": 3868
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.6721316576004028,
+      "learning_rate": 0.00017522947159513767,
+      "loss": 0.1556,
+      "step": 3869
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 3.3207669258117676,
+      "learning_rate": 0.00017519225998511532,
+      "loss": 0.1144,
+      "step": 3870
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 2.309567928314209,
+      "learning_rate": 0.00017515504837509302,
+      "loss": 0.2904,
+      "step": 3871
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 3.172945261001587,
+      "learning_rate": 0.0001751178367650707,
+      "loss": 0.4218,
+      "step": 3872
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.9167137145996094,
+      "learning_rate": 0.00017508062515504834,
+      "loss": 0.1754,
+      "step": 3873
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.7879188656806946,
+      "learning_rate": 0.00017504341354502605,
+      "loss": 0.0516,
+      "step": 3874
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 10.358311653137207,
+      "learning_rate": 0.00017500620193500372,
+      "loss": 0.6033,
+      "step": 3875
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 3.172363758087158,
+      "learning_rate": 0.00017496899032498137,
+      "loss": 0.2935,
+      "step": 3876
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 3.7427735328674316,
+      "learning_rate": 0.00017493177871495905,
+      "loss": 0.2327,
+      "step": 3877
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 5.786536693572998,
+      "learning_rate": 0.00017489456710493672,
+      "loss": 1.0866,
+      "step": 3878
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 5.266422748565674,
+      "learning_rate": 0.0001748573554949144,
+      "loss": 1.0328,
+      "step": 3879
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.993481993675232,
+      "learning_rate": 0.00017482014388489208,
+      "loss": 0.6055,
+      "step": 3880
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 2.0876619815826416,
+      "learning_rate": 0.00017478293227486972,
+      "loss": 0.6218,
+      "step": 3881
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 2.1874656677246094,
+      "learning_rate": 0.00017474572066484743,
+      "loss": 0.3442,
+      "step": 3882
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.9037868976593018,
+      "learning_rate": 0.00017470850905482508,
+      "loss": 0.1977,
+      "step": 3883
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 2.1550540924072266,
+      "learning_rate": 0.00017467129744480275,
+      "loss": 0.3056,
+      "step": 3884
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 2.475119113922119,
+      "learning_rate": 0.00017463408583478045,
+      "loss": 0.4085,
+      "step": 3885
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 2.8540899753570557,
+      "learning_rate": 0.0001745968742247581,
+      "loss": 0.2703,
+      "step": 3886
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.23556649684906,
+      "learning_rate": 0.00017455966261473578,
+      "loss": 0.1271,
+      "step": 3887
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 3.1399030685424805,
+      "learning_rate": 0.00017452245100471348,
+      "loss": 0.3749,
+      "step": 3888
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 2.0435614585876465,
+      "learning_rate": 0.00017448523939469113,
+      "loss": 0.3553,
+      "step": 3889
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 2.5815916061401367,
+      "learning_rate": 0.0001744480277846688,
+      "loss": 0.3135,
+      "step": 3890
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 2.8995308876037598,
+      "learning_rate": 0.00017441081617464645,
+      "loss": 0.3687,
+      "step": 3891
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 3.48101544380188,
+      "learning_rate": 0.00017437360456462416,
+      "loss": 0.5593,
+      "step": 3892
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 2.921421766281128,
+      "learning_rate": 0.00017433639295460183,
+      "loss": 0.3693,
+      "step": 3893
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.2923214435577393,
+      "learning_rate": 0.00017429918134457948,
+      "loss": 0.0802,
+      "step": 3894
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 2.212675094604492,
+      "learning_rate": 0.00017426196973455719,
+      "loss": 0.3009,
+      "step": 3895
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 4.238613128662109,
+      "learning_rate": 0.00017422475812453483,
+      "loss": 0.2686,
+      "step": 3896
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 2.2763991355895996,
+      "learning_rate": 0.0001741875465145125,
+      "loss": 0.234,
+      "step": 3897
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 2.234372615814209,
+      "learning_rate": 0.00017415033490449019,
+      "loss": 0.3564,
+      "step": 3898
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 4.635880470275879,
+      "learning_rate": 0.00017411312329446786,
+      "loss": 0.1816,
+      "step": 3899
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 2.752310037612915,
+      "learning_rate": 0.00017407591168444554,
+      "loss": 0.2585,
+      "step": 3900
+    },
+    {
+      "epoch": 1.37,
+      "eval_loss": 0.5356029272079468,
+      "eval_runtime": 51.7497,
+      "eval_samples_per_second": 41.894,
+      "eval_steps_per_second": 10.473,
+      "eval_wer": 0.4876318975955717,
+      "step": 3900
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 2.9280731678009033,
+      "learning_rate": 0.0001740387000744232,
+      "loss": 0.4685,
+      "step": 3901
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 2.367094039916992,
+      "learning_rate": 0.00017400148846440086,
+      "loss": 0.1438,
+      "step": 3902
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 2.750795602798462,
+      "learning_rate": 0.00017396427685437856,
+      "loss": 0.3067,
+      "step": 3903
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 2.7507309913635254,
+      "learning_rate": 0.0001739270652443562,
+      "loss": 0.4385,
+      "step": 3904
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.9318255186080933,
+      "learning_rate": 0.0001738898536343339,
+      "loss": 0.5685,
+      "step": 3905
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.8872981071472168,
+      "learning_rate": 0.0001738526420243116,
+      "loss": 0.5847,
+      "step": 3906
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.6942192316055298,
+      "learning_rate": 0.00017381543041428924,
+      "loss": 0.3749,
+      "step": 3907
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.610882043838501,
+      "learning_rate": 0.00017377821880426692,
+      "loss": 0.3559,
+      "step": 3908
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.635751485824585,
+      "learning_rate": 0.00017374100719424456,
+      "loss": 0.1369,
+      "step": 3909
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 2.099339723587036,
+      "learning_rate": 0.00017370379558422227,
+      "loss": 0.6477,
+      "step": 3910
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.2585214376449585,
+      "learning_rate": 0.00017366658397419994,
+      "loss": 0.272,
+      "step": 3911
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 3.2351205348968506,
+      "learning_rate": 0.0001736293723641776,
+      "loss": 0.5484,
+      "step": 3912
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 2.1207034587860107,
+      "learning_rate": 0.0001735921607541553,
+      "loss": 0.3723,
+      "step": 3913
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 2.07908034324646,
+      "learning_rate": 0.00017355494914413297,
+      "loss": 0.314,
+      "step": 3914
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 2.0071959495544434,
+      "learning_rate": 0.00017351773753411062,
+      "loss": 0.6595,
+      "step": 3915
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 2.149886131286621,
+      "learning_rate": 0.00017348052592408832,
+      "loss": 0.6534,
+      "step": 3916
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.7911567687988281,
+      "learning_rate": 0.00017344331431406597,
+      "loss": 0.2316,
+      "step": 3917
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 3.381746292114258,
+      "learning_rate": 0.00017340610270404365,
+      "loss": 1.408,
+      "step": 3918
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 4.066563606262207,
+      "learning_rate": 0.00017336889109402132,
+      "loss": 0.5341,
+      "step": 3919
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 2.2210659980773926,
+      "learning_rate": 0.000173331679483999,
+      "loss": 0.2397,
+      "step": 3920
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 2.28897762298584,
+      "learning_rate": 0.00017329446787397667,
+      "loss": 0.277,
+      "step": 3921
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 2.897461175918579,
+      "learning_rate": 0.00017325725626395432,
+      "loss": 0.5583,
+      "step": 3922
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 2.140532970428467,
+      "learning_rate": 0.000173220044653932,
+      "loss": 0.4073,
+      "step": 3923
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 3.024677276611328,
+      "learning_rate": 0.0001731828330439097,
+      "loss": 0.5048,
+      "step": 3924
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 2.735630750656128,
+      "learning_rate": 0.00017314562143388735,
+      "loss": 0.374,
+      "step": 3925
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 5.665718078613281,
+      "learning_rate": 0.00017310840982386503,
+      "loss": 0.7075,
+      "step": 3926
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 4.23260498046875,
+      "learning_rate": 0.00017307119821384268,
+      "loss": 1.46,
+      "step": 3927
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.9788540601730347,
+      "learning_rate": 0.00017303398660382038,
+      "loss": 0.3469,
+      "step": 3928
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.4005393981933594,
+      "learning_rate": 0.00017299677499379805,
+      "loss": 0.1381,
+      "step": 3929
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.7616915702819824,
+      "learning_rate": 0.0001729595633837757,
+      "loss": 0.7645,
+      "step": 3930
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 2.4172654151916504,
+      "learning_rate": 0.0001729223517737534,
+      "loss": 0.5808,
+      "step": 3931
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.8812202215194702,
+      "learning_rate": 0.00017288514016373108,
+      "loss": 0.3962,
+      "step": 3932
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 2.0839879512786865,
+      "learning_rate": 0.00017284792855370873,
+      "loss": 0.3107,
+      "step": 3933
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.7547316551208496,
+      "learning_rate": 0.00017281071694368643,
+      "loss": 0.4774,
+      "step": 3934
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.7336246967315674,
+      "learning_rate": 0.00017277350533366408,
+      "loss": 0.297,
+      "step": 3935
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.464845895767212,
+      "learning_rate": 0.00017273629372364176,
+      "loss": 0.2388,
+      "step": 3936
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.2136765718460083,
+      "learning_rate": 0.00017269908211361946,
+      "loss": 0.0873,
+      "step": 3937
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.4514065980911255,
+      "learning_rate": 0.0001726618705035971,
+      "loss": 0.1335,
+      "step": 3938
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 3.2778918743133545,
+      "learning_rate": 0.00017262465889357478,
+      "loss": 0.3828,
+      "step": 3939
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.513120174407959,
+      "learning_rate": 0.00017258744728355243,
+      "loss": 0.1283,
+      "step": 3940
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.7113932371139526,
+      "learning_rate": 0.00017255023567353014,
+      "loss": 0.1633,
+      "step": 3941
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 2.0901389122009277,
+      "learning_rate": 0.0001725130240635078,
+      "loss": 0.2138,
+      "step": 3942
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 4.5583720207214355,
+      "learning_rate": 0.00017247581245348546,
+      "loss": 0.7136,
+      "step": 3943
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.8743005990982056,
+      "learning_rate": 0.00017243860084346314,
+      "loss": 0.2668,
+      "step": 3944
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 3.3022148609161377,
+      "learning_rate": 0.00017240138923344084,
+      "loss": 1.001,
+      "step": 3945
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 3.1593058109283447,
+      "learning_rate": 0.0001723641776234185,
+      "loss": 0.358,
+      "step": 3946
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 4.342871189117432,
+      "learning_rate": 0.00017232696601339616,
+      "loss": 0.8455,
+      "step": 3947
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 4.121707439422607,
+      "learning_rate": 0.0001722897544033738,
+      "loss": 0.511,
+      "step": 3948
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 2.737363815307617,
+      "learning_rate": 0.00017225254279335152,
+      "loss": 0.2577,
+      "step": 3949
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 4.477611541748047,
+      "learning_rate": 0.0001722153311833292,
+      "loss": 1.0024,
+      "step": 3950
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 2.959623336791992,
+      "learning_rate": 0.00017217811957330684,
+      "loss": 0.4979,
+      "step": 3951
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 3.6896047592163086,
+      "learning_rate": 0.00017214090796328454,
+      "loss": 0.502,
+      "step": 3952
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 2.5401012897491455,
+      "learning_rate": 0.0001721036963532622,
+      "loss": 0.4048,
+      "step": 3953
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 2.559720039367676,
+      "learning_rate": 0.00017206648474323987,
+      "loss": 0.201,
+      "step": 3954
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.9754453897476196,
+      "learning_rate": 0.00017202927313321757,
+      "loss": 0.6484,
+      "step": 3955
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.5338460206985474,
+      "learning_rate": 0.00017199206152319522,
+      "loss": 0.3546,
+      "step": 3956
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.5986213684082031,
+      "learning_rate": 0.0001719548499131729,
+      "loss": 0.3799,
+      "step": 3957
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 2.644622802734375,
+      "learning_rate": 0.0001719176383031506,
+      "loss": 0.5743,
+      "step": 3958
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 2.5025296211242676,
+      "learning_rate": 0.00017188042669312825,
+      "loss": 0.3493,
+      "step": 3959
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.100515604019165,
+      "learning_rate": 0.00017184321508310592,
+      "loss": 0.0687,
+      "step": 3960
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.8527333736419678,
+      "learning_rate": 0.00017180600347308357,
+      "loss": 0.172,
+      "step": 3961
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 2.1646475791931152,
+      "learning_rate": 0.00017176879186306127,
+      "loss": 0.3352,
+      "step": 3962
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.9406778812408447,
+      "learning_rate": 0.00017173158025303895,
+      "loss": 0.2927,
+      "step": 3963
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.5747565031051636,
+      "learning_rate": 0.0001716943686430166,
+      "loss": 0.2809,
+      "step": 3964
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 2.524975299835205,
+      "learning_rate": 0.00017165715703299427,
+      "loss": 0.3335,
+      "step": 3965
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 4.1112823486328125,
+      "learning_rate": 0.00017161994542297195,
+      "loss": 0.3868,
+      "step": 3966
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 2.326080799102783,
+      "learning_rate": 0.00017158273381294963,
+      "loss": 0.4065,
+      "step": 3967
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.6009414196014404,
+      "learning_rate": 0.0001715455222029273,
+      "loss": 0.1864,
+      "step": 3968
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 14.051066398620605,
+      "learning_rate": 0.00017150831059290495,
+      "loss": 2.8598,
+      "step": 3969
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 3.3767621517181396,
+      "learning_rate": 0.00017147109898288265,
+      "loss": 0.192,
+      "step": 3970
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 4.84522008895874,
+      "learning_rate": 0.0001714338873728603,
+      "loss": 0.5361,
+      "step": 3971
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 3.2274253368377686,
+      "learning_rate": 0.00017139667576283798,
+      "loss": 0.5027,
+      "step": 3972
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.2869607210159302,
+      "learning_rate": 0.00017135946415281568,
+      "loss": 0.1095,
+      "step": 3973
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 2.0447194576263428,
+      "learning_rate": 0.00017132225254279333,
+      "loss": 0.2067,
+      "step": 3974
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 4.841538906097412,
+      "learning_rate": 0.000171285040932771,
+      "loss": 0.6107,
+      "step": 3975
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.8944703340530396,
+      "learning_rate": 0.0001712478293227487,
+      "loss": 0.1473,
+      "step": 3976
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 4.2880659103393555,
+      "learning_rate": 0.00017121061771272636,
+      "loss": 1.5288,
+      "step": 3977
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 3.7833964824676514,
+      "learning_rate": 0.00017117340610270403,
+      "loss": 0.4495,
+      "step": 3978
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 5.308502674102783,
+      "learning_rate": 0.00017113619449268168,
+      "loss": 1.1916,
+      "step": 3979
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.6576392650604248,
+      "learning_rate": 0.00017109898288265938,
+      "loss": 0.5451,
+      "step": 3980
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.7152270078659058,
+      "learning_rate": 0.00017106177127263706,
+      "loss": 0.4996,
+      "step": 3981
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 2.689882278442383,
+      "learning_rate": 0.0001710245596626147,
+      "loss": 0.8473,
+      "step": 3982
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 2.522495746612549,
+      "learning_rate": 0.0001709873480525924,
+      "loss": 0.6978,
+      "step": 3983
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.8638933897018433,
+      "learning_rate": 0.00017095013644257006,
+      "loss": 0.3669,
+      "step": 3984
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.6137725114822388,
+      "learning_rate": 0.00017091292483254774,
+      "loss": 0.2279,
+      "step": 3985
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 3.917705774307251,
+      "learning_rate": 0.0001708757132225254,
+      "loss": 1.6021,
+      "step": 3986
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 3.742889165878296,
+      "learning_rate": 0.0001708385016125031,
+      "loss": 0.6855,
+      "step": 3987
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 2.922379493713379,
+      "learning_rate": 0.00017080129000248076,
+      "loss": 0.2816,
+      "step": 3988
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.8310586214065552,
+      "learning_rate": 0.00017076407839245844,
+      "loss": 0.3215,
+      "step": 3989
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.7077147960662842,
+      "learning_rate": 0.0001707268667824361,
+      "loss": 0.4342,
+      "step": 3990
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.127012014389038,
+      "learning_rate": 0.0001706896551724138,
+      "loss": 0.1672,
+      "step": 3991
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 2.1023213863372803,
+      "learning_rate": 0.00017065244356239144,
+      "loss": 0.4928,
+      "step": 3992
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.650435209274292,
+      "learning_rate": 0.00017061523195236911,
+      "loss": 0.3392,
+      "step": 3993
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.628867268562317,
+      "learning_rate": 0.00017057802034234682,
+      "loss": 0.1326,
+      "step": 3994
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 3.0410878658294678,
+      "learning_rate": 0.00017054080873232447,
+      "loss": 0.3564,
+      "step": 3995
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 4.636041164398193,
+      "learning_rate": 0.00017050359712230214,
+      "loss": 0.5428,
+      "step": 3996
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 2.0328681468963623,
+      "learning_rate": 0.0001704663855122798,
+      "loss": 0.1978,
+      "step": 3997
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 2.5405378341674805,
+      "learning_rate": 0.0001704291739022575,
+      "loss": 0.341,
+      "step": 3998
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 2.484320640563965,
+      "learning_rate": 0.00017039196229223517,
+      "loss": 0.3324,
+      "step": 3999
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 2.7092721462249756,
+      "learning_rate": 0.00017035475068221282,
+      "loss": 0.3742,
+      "step": 4000
+    },
+    {
+      "epoch": 1.4,
+      "eval_loss": 0.4177427291870117,
+      "eval_runtime": 50.8405,
+      "eval_samples_per_second": 42.643,
+      "eval_steps_per_second": 10.661,
+      "eval_wer": 0.3851409790693652,
+      "step": 4000
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.0885396003723145,
+      "learning_rate": 0.00017031753907219052,
+      "loss": 0.0431,
+      "step": 4001
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 3.04752254486084,
+      "learning_rate": 0.0001702803274621682,
+      "loss": 0.2688,
+      "step": 4002
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 2.453718900680542,
+      "learning_rate": 0.00017024311585214585,
+      "loss": 0.1588,
+      "step": 4003
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 2.611881732940674,
+      "learning_rate": 0.00017020590424212355,
+      "loss": 0.1296,
+      "step": 4004
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.520546555519104,
+      "learning_rate": 0.0001701686926321012,
+      "loss": 0.4584,
+      "step": 4005
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.637783408164978,
+      "learning_rate": 0.00017013148102207887,
+      "loss": 0.4303,
+      "step": 4006
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.7446013689041138,
+      "learning_rate": 0.00017009426941205655,
+      "loss": 0.2818,
+      "step": 4007
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.9278056621551514,
+      "learning_rate": 0.00017005705780203422,
+      "loss": 0.1499,
+      "step": 4008
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.8733441829681396,
+      "learning_rate": 0.0001700198461920119,
+      "loss": 0.3069,
+      "step": 4009
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.9108270406723022,
+      "learning_rate": 0.00016998263458198955,
+      "loss": 0.3054,
+      "step": 4010
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.8484750986099243,
+      "learning_rate": 0.00016994542297196722,
+      "loss": 0.2867,
+      "step": 4011
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.9013972878456116,
+      "learning_rate": 0.00016990821136194493,
+      "loss": 0.129,
+      "step": 4012
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.9117753505706787,
+      "learning_rate": 0.00016987099975192258,
+      "loss": 0.2353,
+      "step": 4013
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.5956077575683594,
+      "learning_rate": 0.00016983378814190025,
+      "loss": 0.1636,
+      "step": 4014
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 3.0155153274536133,
+      "learning_rate": 0.00016979657653187793,
+      "loss": 1.5728,
+      "step": 4015
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.7413406372070312,
+      "learning_rate": 0.0001697593649218556,
+      "loss": 0.2977,
+      "step": 4016
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 3.3247792720794678,
+      "learning_rate": 0.00016972215331183328,
+      "loss": 0.3402,
+      "step": 4017
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.7041586637496948,
+      "learning_rate": 0.00016968494170181093,
+      "loss": 0.1643,
+      "step": 4018
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 2.3175644874572754,
+      "learning_rate": 0.00016964773009178863,
+      "loss": 0.2752,
+      "step": 4019
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 2.561286211013794,
+      "learning_rate": 0.0001696105184817663,
+      "loss": 0.3803,
+      "step": 4020
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.3216911554336548,
+      "learning_rate": 0.00016957330687174396,
+      "loss": 0.1224,
+      "step": 4021
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 2.120414972305298,
+      "learning_rate": 0.00016953609526172166,
+      "loss": 0.2079,
+      "step": 4022
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 2.5523006916046143,
+      "learning_rate": 0.0001694988836516993,
+      "loss": 0.3881,
+      "step": 4023
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 2.9489386081695557,
+      "learning_rate": 0.00016946167204167698,
+      "loss": 0.3729,
+      "step": 4024
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 2.491358757019043,
+      "learning_rate": 0.00016942446043165469,
+      "loss": 0.4037,
+      "step": 4025
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.5692845582962036,
+      "learning_rate": 0.00016938724882163233,
+      "loss": 0.044,
+      "step": 4026
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 5.157808780670166,
+      "learning_rate": 0.00016935003721161,
+      "loss": 0.2437,
+      "step": 4027
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 3.8554892539978027,
+      "learning_rate": 0.00016931282560158766,
+      "loss": 1.2543,
+      "step": 4028
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 4.396365642547607,
+      "learning_rate": 0.00016927561399156536,
+      "loss": 0.8959,
+      "step": 4029
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 5.370511054992676,
+      "learning_rate": 0.00016923840238154304,
+      "loss": 1.3415,
+      "step": 4030
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 2.6273608207702637,
+      "learning_rate": 0.0001692011907715207,
+      "loss": 0.7875,
+      "step": 4031
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.6899088621139526,
+      "learning_rate": 0.00016916397916149836,
+      "loss": 0.4165,
+      "step": 4032
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 2.6857593059539795,
+      "learning_rate": 0.00016912676755147607,
+      "loss": 0.552,
+      "step": 4033
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 3.3171510696411133,
+      "learning_rate": 0.00016908955594145371,
+      "loss": 0.2746,
+      "step": 4034
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 3.1243948936462402,
+      "learning_rate": 0.0001690523443314314,
+      "loss": 1.4537,
+      "step": 4035
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 2.551290273666382,
+      "learning_rate": 0.00016901513272140907,
+      "loss": 0.7787,
+      "step": 4036
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.3610817193984985,
+      "learning_rate": 0.00016897792111138674,
+      "loss": 0.2018,
+      "step": 4037
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 2.2377007007598877,
+      "learning_rate": 0.00016894070950136442,
+      "loss": 0.6747,
+      "step": 4038
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 3.3656833171844482,
+      "learning_rate": 0.00016890349789134207,
+      "loss": 1.3106,
+      "step": 4039
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.9152419567108154,
+      "learning_rate": 0.00016886628628131977,
+      "loss": 0.4056,
+      "step": 4040
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 2.9426541328430176,
+      "learning_rate": 0.00016882907467129742,
+      "loss": 0.7818,
+      "step": 4041
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.2641545534133911,
+      "learning_rate": 0.0001687918630612751,
+      "loss": 0.218,
+      "step": 4042
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.596787691116333,
+      "learning_rate": 0.0001687546514512528,
+      "loss": 0.2985,
+      "step": 4043
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.177599549293518,
+      "learning_rate": 0.00016871743984123044,
+      "loss": 0.1941,
+      "step": 4044
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 2.4323856830596924,
+      "learning_rate": 0.00016868022823120812,
+      "loss": 0.4303,
+      "step": 4045
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.3808494806289673,
+      "learning_rate": 0.00016864301662118582,
+      "loss": 0.1963,
+      "step": 4046
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 3.3002772331237793,
+      "learning_rate": 0.00016860580501116347,
+      "loss": 0.4449,
+      "step": 4047
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 3.7071423530578613,
+      "learning_rate": 0.00016856859340114115,
+      "loss": 1.0698,
+      "step": 4048
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 3.5901901721954346,
+      "learning_rate": 0.0001685313817911188,
+      "loss": 0.4192,
+      "step": 4049
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.7595155239105225,
+      "learning_rate": 0.0001684941701810965,
+      "loss": 0.1192,
+      "step": 4050
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 2.3339569568634033,
+      "learning_rate": 0.00016845695857107418,
+      "loss": 0.2746,
+      "step": 4051
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 2.930159568786621,
+      "learning_rate": 0.00016841974696105182,
+      "loss": 0.6843,
+      "step": 4052
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.9335921406745911,
+      "learning_rate": 0.0001683825353510295,
+      "loss": 0.0758,
+      "step": 4053
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 9.361223220825195,
+      "learning_rate": 0.00016834532374100718,
+      "loss": 0.8905,
+      "step": 4054
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 2.8475027084350586,
+      "learning_rate": 0.00016830811213098485,
+      "loss": 0.7053,
+      "step": 4055
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 3.661919593811035,
+      "learning_rate": 0.00016827090052096253,
+      "loss": 0.4094,
+      "step": 4056
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.8695873022079468,
+      "learning_rate": 0.0001682336889109402,
+      "loss": 0.1678,
+      "step": 4057
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 2.414778232574463,
+      "learning_rate": 0.00016819647730091788,
+      "loss": 0.5948,
+      "step": 4058
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 2.0762922763824463,
+      "learning_rate": 0.00016815926569089553,
+      "loss": 0.2875,
+      "step": 4059
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 2.1993393898010254,
+      "learning_rate": 0.0001681220540808732,
+      "loss": 0.5876,
+      "step": 4060
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 3.719446897506714,
+      "learning_rate": 0.0001680848424708509,
+      "loss": 0.5155,
+      "step": 4061
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 3.0953850746154785,
+      "learning_rate": 0.00016804763086082855,
+      "loss": 0.4639,
+      "step": 4062
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 2.066967487335205,
+      "learning_rate": 0.00016801041925080623,
+      "loss": 0.3151,
+      "step": 4063
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 2.6755409240722656,
+      "learning_rate": 0.00016797320764078393,
+      "loss": 0.4324,
+      "step": 4064
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.722860336303711,
+      "learning_rate": 0.00016793599603076158,
+      "loss": 0.3173,
+      "step": 4065
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 3.7338409423828125,
+      "learning_rate": 0.00016789878442073926,
+      "loss": 0.8751,
+      "step": 4066
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.5438848733901978,
+      "learning_rate": 0.0001678615728107169,
+      "loss": 0.164,
+      "step": 4067
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 2.704420328140259,
+      "learning_rate": 0.0001678243612006946,
+      "loss": 0.2626,
+      "step": 4068
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 3.400644063949585,
+      "learning_rate": 0.00016778714959067229,
+      "loss": 0.5246,
+      "step": 4069
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 2.035137414932251,
+      "learning_rate": 0.00016774993798064993,
+      "loss": 0.3621,
+      "step": 4070
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 3.7264976501464844,
+      "learning_rate": 0.00016771272637062764,
+      "loss": 0.6366,
+      "step": 4071
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 2.768700361251831,
+      "learning_rate": 0.00016767551476060529,
+      "loss": 0.8278,
+      "step": 4072
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 2.690241813659668,
+      "learning_rate": 0.00016763830315058296,
+      "loss": 0.3648,
+      "step": 4073
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 5.137421607971191,
+      "learning_rate": 0.00016760109154056064,
+      "loss": 0.3702,
+      "step": 4074
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.9374441504478455,
+      "learning_rate": 0.0001675638799305383,
+      "loss": 0.0988,
+      "step": 4075
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.983991265296936,
+      "learning_rate": 0.000167526668320516,
+      "loss": 0.1474,
+      "step": 4076
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.5250822305679321,
+      "learning_rate": 0.00016748945671049366,
+      "loss": 0.1217,
+      "step": 4077
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.7774677276611328,
+      "learning_rate": 0.00016745224510047134,
+      "loss": 0.1447,
+      "step": 4078
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 6.5276994705200195,
+      "learning_rate": 0.00016741503349044902,
+      "loss": 0.3185,
+      "step": 4079
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 2.6374850273132324,
+      "learning_rate": 0.00016737782188042666,
+      "loss": 1.071,
+      "step": 4080
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.5644630193710327,
+      "learning_rate": 0.00016734061027040434,
+      "loss": 0.4552,
+      "step": 4081
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 2.4700703620910645,
+      "learning_rate": 0.00016730339866038204,
+      "loss": 0.6185,
+      "step": 4082
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.9310435056686401,
+      "learning_rate": 0.0001672661870503597,
+      "loss": 0.3552,
+      "step": 4083
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.529681921005249,
+      "learning_rate": 0.00016722897544033737,
+      "loss": 0.2221,
+      "step": 4084
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 2.560516357421875,
+      "learning_rate": 0.00016719176383031502,
+      "loss": 0.2798,
+      "step": 4085
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 2.354257106781006,
+      "learning_rate": 0.00016715455222029272,
+      "loss": 0.2105,
+      "step": 4086
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 2.342054843902588,
+      "learning_rate": 0.0001671173406102704,
+      "loss": 0.2796,
+      "step": 4087
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.957754135131836,
+      "learning_rate": 0.00016708012900024804,
+      "loss": 0.2997,
+      "step": 4088
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 2.821141004562378,
+      "learning_rate": 0.00016704291739022575,
+      "loss": 0.7008,
+      "step": 4089
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 12.810015678405762,
+      "learning_rate": 0.00016700570578020342,
+      "loss": 3.4665,
+      "step": 4090
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 2.314800500869751,
+      "learning_rate": 0.00016696849417018107,
+      "loss": 0.689,
+      "step": 4091
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 3.096869468688965,
+      "learning_rate": 0.00016693128256015877,
+      "loss": 0.7017,
+      "step": 4092
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 3.2854676246643066,
+      "learning_rate": 0.00016689407095013642,
+      "loss": 0.5166,
+      "step": 4093
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.9120439291000366,
+      "learning_rate": 0.0001668568593401141,
+      "loss": 0.2128,
+      "step": 4094
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 2.307483434677124,
+      "learning_rate": 0.00016681964773009177,
+      "loss": 0.2255,
+      "step": 4095
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 3.0140573978424072,
+      "learning_rate": 0.00016678243612006945,
+      "loss": 0.3603,
+      "step": 4096
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.1581931114196777,
+      "learning_rate": 0.00016674522451004713,
+      "loss": 0.1719,
+      "step": 4097
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.792546033859253,
+      "learning_rate": 0.00016670801290002478,
+      "loss": 0.1403,
+      "step": 4098
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.0819088220596313,
+      "learning_rate": 0.00016667080129000248,
+      "loss": 0.0878,
+      "step": 4099
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 5.966635227203369,
+      "learning_rate": 0.00016663358967998015,
+      "loss": 0.6863,
+      "step": 4100
+    },
+    {
+      "epoch": 1.44,
+      "eval_loss": 0.4655866026878357,
+      "eval_runtime": 51.618,
+      "eval_samples_per_second": 42.001,
+      "eval_steps_per_second": 10.5,
+      "eval_wer": 0.4545061408060889,
+      "step": 4100
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.8284623622894287,
+      "learning_rate": 0.0001665963780699578,
+      "loss": 0.0954,
+      "step": 4101
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.2109270095825195,
+      "learning_rate": 0.00016655916645993548,
+      "loss": 0.0594,
+      "step": 4102
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 2.24116587638855,
+      "learning_rate": 0.00016652195484991315,
+      "loss": 0.2425,
+      "step": 4103
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 6.5993475914001465,
+      "learning_rate": 0.00016648474323989083,
+      "loss": 1.3286,
+      "step": 4104
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 2.824436902999878,
+      "learning_rate": 0.0001664475316298685,
+      "loss": 0.8653,
+      "step": 4105
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 2.3655171394348145,
+      "learning_rate": 0.00016641032001984615,
+      "loss": 0.6443,
+      "step": 4106
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 2.7993006706237793,
+      "learning_rate": 0.00016637310840982386,
+      "loss": 0.4293,
+      "step": 4107
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.109576940536499,
+      "learning_rate": 0.00016633589679980153,
+      "loss": 0.3577,
+      "step": 4108
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.4774690866470337,
+      "learning_rate": 0.00016629868518977918,
+      "loss": 0.2693,
+      "step": 4109
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.442692518234253,
+      "learning_rate": 0.00016626147357975688,
+      "loss": 0.3693,
+      "step": 4110
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.9737093448638916,
+      "learning_rate": 0.00016622426196973453,
+      "loss": 0.2922,
+      "step": 4111
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 3.433197498321533,
+      "learning_rate": 0.0001661870503597122,
+      "loss": 0.2324,
+      "step": 4112
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.9645071029663086,
+      "learning_rate": 0.0001661498387496899,
+      "loss": 0.2494,
+      "step": 4113
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 2.4682822227478027,
+      "learning_rate": 0.00016611262713966756,
+      "loss": 0.3718,
+      "step": 4114
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.6847212314605713,
+      "learning_rate": 0.00016607541552964524,
+      "loss": 0.315,
+      "step": 4115
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.7860573530197144,
+      "learning_rate": 0.00016603820391962289,
+      "loss": 0.4285,
+      "step": 4116
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 2.275067090988159,
+      "learning_rate": 0.0001660009923096006,
+      "loss": 0.2831,
+      "step": 4117
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 2.7724900245666504,
+      "learning_rate": 0.00016596378069957826,
+      "loss": 0.2465,
+      "step": 4118
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 2.4586076736450195,
+      "learning_rate": 0.0001659265690895559,
+      "loss": 0.2474,
+      "step": 4119
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 2.156881332397461,
+      "learning_rate": 0.00016588935747953362,
+      "loss": 0.1468,
+      "step": 4120
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 2.4893946647644043,
+      "learning_rate": 0.0001658521458695113,
+      "loss": 0.2606,
+      "step": 4121
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.4689332246780396,
+      "learning_rate": 0.00016581493425948894,
+      "loss": 0.0906,
+      "step": 4122
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 4.48578405380249,
+      "learning_rate": 0.00016577772264946662,
+      "loss": 1.7996,
+      "step": 4123
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.9185426235198975,
+      "learning_rate": 0.0001657405110394443,
+      "loss": 0.1204,
+      "step": 4124
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 3.5373518466949463,
+      "learning_rate": 0.00016570329942942197,
+      "loss": 0.6004,
+      "step": 4125
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.8519726991653442,
+      "learning_rate": 0.00016566608781939964,
+      "loss": 0.0576,
+      "step": 4126
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.851152777671814,
+      "learning_rate": 0.0001656288762093773,
+      "loss": 0.2595,
+      "step": 4127
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.1237139701843262,
+      "learning_rate": 0.000165591664599355,
+      "loss": 0.0662,
+      "step": 4128
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.562714695930481,
+      "learning_rate": 0.00016555445298933264,
+      "loss": 0.1051,
+      "step": 4129
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.522381067276001,
+      "learning_rate": 0.00016551724137931032,
+      "loss": 0.4156,
+      "step": 4130
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.2748173475265503,
+      "learning_rate": 0.00016548002976928802,
+      "loss": 0.2003,
+      "step": 4131
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 2.538360118865967,
+      "learning_rate": 0.00016544281815926567,
+      "loss": 0.963,
+      "step": 4132
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 2.971334934234619,
+      "learning_rate": 0.00016540560654924335,
+      "loss": 0.6725,
+      "step": 4133
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 2.554715633392334,
+      "learning_rate": 0.00016536839493922105,
+      "loss": 0.8091,
+      "step": 4134
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.7067142724990845,
+      "learning_rate": 0.0001653311833291987,
+      "loss": 0.4307,
+      "step": 4135
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 2.621626377105713,
+      "learning_rate": 0.00016529397171917637,
+      "loss": 0.2681,
+      "step": 4136
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 4.470788478851318,
+      "learning_rate": 0.00016525676010915402,
+      "loss": 0.9887,
+      "step": 4137
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.222482442855835,
+      "learning_rate": 0.00016521954849913173,
+      "loss": 0.1786,
+      "step": 4138
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 2.5366249084472656,
+      "learning_rate": 0.0001651823368891094,
+      "loss": 0.4021,
+      "step": 4139
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.220999836921692,
+      "learning_rate": 0.00016514512527908705,
+      "loss": 0.1173,
+      "step": 4140
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.6525065898895264,
+      "learning_rate": 0.00016510791366906475,
+      "loss": 0.249,
+      "step": 4141
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 3.0360629558563232,
+      "learning_rate": 0.0001650707020590424,
+      "loss": 1.3031,
+      "step": 4142
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 2.506528854370117,
+      "learning_rate": 0.00016503349044902008,
+      "loss": 0.3216,
+      "step": 4143
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 2.794973611831665,
+      "learning_rate": 0.00016499627883899775,
+      "loss": 0.3352,
+      "step": 4144
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.864272117614746,
+      "learning_rate": 0.00016495906722897543,
+      "loss": 0.1704,
+      "step": 4145
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.7695359587669373,
+      "learning_rate": 0.0001649218556189531,
+      "loss": 0.0561,
+      "step": 4146
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.7436617612838745,
+      "learning_rate": 0.00016488464400893078,
+      "loss": 0.3065,
+      "step": 4147
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.8266197443008423,
+      "learning_rate": 0.00016484743239890843,
+      "loss": 0.1391,
+      "step": 4148
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 4.190832614898682,
+      "learning_rate": 0.00016481022078888613,
+      "loss": 0.4169,
+      "step": 4149
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 3.211941719055176,
+      "learning_rate": 0.00016477300917886378,
+      "loss": 0.2127,
+      "step": 4150
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 3.884977340698242,
+      "learning_rate": 0.00016473579756884146,
+      "loss": 1.0903,
+      "step": 4151
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 2.969237804412842,
+      "learning_rate": 0.00016469858595881916,
+      "loss": 1.1619,
+      "step": 4152
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 4.25085973739624,
+      "learning_rate": 0.0001646613743487968,
+      "loss": 0.3401,
+      "step": 4153
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.1101423501968384,
+      "learning_rate": 0.00016462416273877448,
+      "loss": 0.0716,
+      "step": 4154
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.9234617948532104,
+      "learning_rate": 0.00016458695112875213,
+      "loss": 0.8407,
+      "step": 4155
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 2.631096601486206,
+      "learning_rate": 0.00016454973951872984,
+      "loss": 0.7553,
+      "step": 4156
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.5557128190994263,
+      "learning_rate": 0.0001645125279087075,
+      "loss": 0.4707,
+      "step": 4157
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.641202449798584,
+      "learning_rate": 0.00016447531629868516,
+      "loss": 0.3575,
+      "step": 4158
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 2.3638205528259277,
+      "learning_rate": 0.00016443810468866286,
+      "loss": 0.3766,
+      "step": 4159
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 2.019489288330078,
+      "learning_rate": 0.0001644008930786405,
+      "loss": 0.412,
+      "step": 4160
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.5674275159835815,
+      "learning_rate": 0.0001643636814686182,
+      "loss": 0.1787,
+      "step": 4161
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.2668960094451904,
+      "learning_rate": 0.0001643264698585959,
+      "loss": 0.1301,
+      "step": 4162
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.8856401443481445,
+      "learning_rate": 0.00016428925824857354,
+      "loss": 0.3409,
+      "step": 4163
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 2.119640350341797,
+      "learning_rate": 0.00016425204663855121,
+      "loss": 0.2735,
+      "step": 4164
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 3.6939704418182373,
+      "learning_rate": 0.0001642148350285289,
+      "loss": 0.4351,
+      "step": 4165
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 3.8741753101348877,
+      "learning_rate": 0.00016417762341850657,
+      "loss": 0.4203,
+      "step": 4166
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 2.0240747928619385,
+      "learning_rate": 0.00016414041180848424,
+      "loss": 0.2846,
+      "step": 4167
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.6168068647384644,
+      "learning_rate": 0.0001641032001984619,
+      "loss": 0.3109,
+      "step": 4168
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.7785304188728333,
+      "learning_rate": 0.00016406598858843957,
+      "loss": 0.0582,
+      "step": 4169
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 2.258286476135254,
+      "learning_rate": 0.00016402877697841727,
+      "loss": 0.1708,
+      "step": 4170
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.3507694005966187,
+      "learning_rate": 0.00016399156536839492,
+      "loss": 0.1883,
+      "step": 4171
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 3.4873201847076416,
+      "learning_rate": 0.0001639543537583726,
+      "loss": 0.3557,
+      "step": 4172
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 2.767658233642578,
+      "learning_rate": 0.00016391714214835024,
+      "loss": 0.3405,
+      "step": 4173
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 4.61103630065918,
+      "learning_rate": 0.00016387993053832795,
+      "loss": 0.5315,
+      "step": 4174
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 4.691863536834717,
+      "learning_rate": 0.00016384271892830562,
+      "loss": 0.2525,
+      "step": 4175
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 2.6262426376342773,
+      "learning_rate": 0.00016380550731828327,
+      "loss": 0.3521,
+      "step": 4176
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 4.043594837188721,
+      "learning_rate": 0.00016376829570826097,
+      "loss": 0.7037,
+      "step": 4177
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 3.2441458702087402,
+      "learning_rate": 0.00016373108409823865,
+      "loss": 0.4664,
+      "step": 4178
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.617117166519165,
+      "learning_rate": 0.0001636938724882163,
+      "loss": 0.0479,
+      "step": 4179
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.4325096607208252,
+      "learning_rate": 0.000163656660878194,
+      "loss": 0.3984,
+      "step": 4180
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 2.373211622238159,
+      "learning_rate": 0.00016361944926817165,
+      "loss": 0.5353,
+      "step": 4181
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.8118925094604492,
+      "learning_rate": 0.00016358223765814932,
+      "loss": 0.3167,
+      "step": 4182
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 2.375030279159546,
+      "learning_rate": 0.00016354502604812703,
+      "loss": 0.173,
+      "step": 4183
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 2.043342113494873,
+      "learning_rate": 0.00016350781443810468,
+      "loss": 0.5591,
+      "step": 4184
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 2.062807321548462,
+      "learning_rate": 0.00016347060282808235,
+      "loss": 0.3204,
+      "step": 4185
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.876054286956787,
+      "learning_rate": 0.00016343339121806,
+      "loss": 0.2098,
+      "step": 4186
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 2.224905014038086,
+      "learning_rate": 0.0001633961796080377,
+      "loss": 0.5262,
+      "step": 4187
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 2.370276689529419,
+      "learning_rate": 0.00016335896799801538,
+      "loss": 0.6072,
+      "step": 4188
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 3.29032301902771,
+      "learning_rate": 0.00016332175638799303,
+      "loss": 0.7222,
+      "step": 4189
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 2.3209869861602783,
+      "learning_rate": 0.0001632845447779707,
+      "loss": 0.4061,
+      "step": 4190
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 2.168501853942871,
+      "learning_rate": 0.0001632473331679484,
+      "loss": 0.2453,
+      "step": 4191
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.7172929048538208,
+      "learning_rate": 0.00016321012155792606,
+      "loss": 0.4018,
+      "step": 4192
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 3.087212562561035,
+      "learning_rate": 0.00016317290994790373,
+      "loss": 0.2529,
+      "step": 4193
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 2.2027931213378906,
+      "learning_rate": 0.00016313569833788138,
+      "loss": 0.3247,
+      "step": 4194
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.7877609729766846,
+      "learning_rate": 0.00016309848672785908,
+      "loss": 0.1612,
+      "step": 4195
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 2.64384126663208,
+      "learning_rate": 0.00016306127511783676,
+      "loss": 0.4328,
+      "step": 4196
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 3.098459243774414,
+      "learning_rate": 0.0001630240635078144,
+      "loss": 0.4273,
+      "step": 4197
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 2.489880084991455,
+      "learning_rate": 0.0001629868518977921,
+      "loss": 0.3342,
+      "step": 4198
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 2.079407215118408,
+      "learning_rate": 0.00016294964028776976,
+      "loss": 0.1652,
+      "step": 4199
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 9.008259773254395,
+      "learning_rate": 0.00016291242867774743,
+      "loss": 0.281,
+      "step": 4200
+    },
+    {
+      "epoch": 1.47,
+      "eval_loss": 0.4405984878540039,
+      "eval_runtime": 51.6554,
+      "eval_samples_per_second": 41.97,
+      "eval_steps_per_second": 10.493,
+      "eval_wer": 0.41567202906071615,
+      "step": 4200
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 2.8630118370056152,
+      "learning_rate": 0.00016287521706772514,
+      "loss": 0.3071,
+      "step": 4201
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 2.293118476867676,
+      "learning_rate": 0.0001628380054577028,
+      "loss": 0.2474,
+      "step": 4202
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 4.106233596801758,
+      "learning_rate": 0.00016280079384768046,
+      "loss": 0.094,
+      "step": 4203
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.9723445177078247,
+      "learning_rate": 0.0001627635822376581,
+      "loss": 0.1708,
+      "step": 4204
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.9736557006835938,
+      "learning_rate": 0.00016272637062763581,
+      "loss": 0.8252,
+      "step": 4205
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.89078688621521,
+      "learning_rate": 0.0001626891590176135,
+      "loss": 0.516,
+      "step": 4206
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 2.3267147541046143,
+      "learning_rate": 0.00016265194740759114,
+      "loss": 0.7264,
+      "step": 4207
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.9280672073364258,
+      "learning_rate": 0.00016261473579756884,
+      "loss": 0.462,
+      "step": 4208
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.9770915508270264,
+      "learning_rate": 0.00016257752418754652,
+      "loss": 0.2252,
+      "step": 4209
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.2686558961868286,
+      "learning_rate": 0.00016254031257752417,
+      "loss": 0.154,
+      "step": 4210
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 2.1990997791290283,
+      "learning_rate": 0.00016250310096750184,
+      "loss": 0.1505,
+      "step": 4211
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.4449461698532104,
+      "learning_rate": 0.00016246588935747952,
+      "loss": 0.218,
+      "step": 4212
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 2.5558810234069824,
+      "learning_rate": 0.0001624286777474572,
+      "loss": 0.6625,
+      "step": 4213
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 2.122541666030884,
+      "learning_rate": 0.00016239146613743487,
+      "loss": 0.3919,
+      "step": 4214
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 2.2353134155273438,
+      "learning_rate": 0.00016235425452741252,
+      "loss": 0.4296,
+      "step": 4215
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.7417476177215576,
+      "learning_rate": 0.00016231704291739022,
+      "loss": 0.4243,
+      "step": 4216
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.6109976768493652,
+      "learning_rate": 0.00016227983130736787,
+      "loss": 0.169,
+      "step": 4217
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.284348487854004,
+      "learning_rate": 0.00016224261969734555,
+      "loss": 0.2186,
+      "step": 4218
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.6397535800933838,
+      "learning_rate": 0.00016220540808732325,
+      "loss": 0.3277,
+      "step": 4219
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 3.001840114593506,
+      "learning_rate": 0.0001621681964773009,
+      "loss": 0.158,
+      "step": 4220
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.503371000289917,
+      "learning_rate": 0.00016213098486727857,
+      "loss": 0.2988,
+      "step": 4221
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.4237070083618164,
+      "learning_rate": 0.00016209377325725628,
+      "loss": 0.0898,
+      "step": 4222
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 3.1129143238067627,
+      "learning_rate": 0.00016205656164723392,
+      "loss": 0.2174,
+      "step": 4223
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.9691246747970581,
+      "learning_rate": 0.0001620193500372116,
+      "loss": 0.1042,
+      "step": 4224
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 2.2304303646087646,
+      "learning_rate": 0.00016198213842718925,
+      "loss": 0.1921,
+      "step": 4225
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 2.0455522537231445,
+      "learning_rate": 0.00016194492681716695,
+      "loss": 0.1548,
+      "step": 4226
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 3.2553882598876953,
+      "learning_rate": 0.00016190771520714463,
+      "loss": 0.523,
+      "step": 4227
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.433052897453308,
+      "learning_rate": 0.00016187050359712228,
+      "loss": 0.1224,
+      "step": 4228
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 4.60850191116333,
+      "learning_rate": 0.00016183329198709998,
+      "loss": 0.7773,
+      "step": 4229
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 3.2903518676757812,
+      "learning_rate": 0.00016179608037707763,
+      "loss": 0.5722,
+      "step": 4230
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 3.1476831436157227,
+      "learning_rate": 0.0001617588687670553,
+      "loss": 0.5969,
+      "step": 4231
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.1613272428512573,
+      "learning_rate": 0.00016172165715703298,
+      "loss": 0.1067,
+      "step": 4232
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.8969131708145142,
+      "learning_rate": 0.00016168444554701065,
+      "loss": 0.4415,
+      "step": 4233
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 2.0695786476135254,
+      "learning_rate": 0.00016164723393698833,
+      "loss": 0.3461,
+      "step": 4234
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 4.001595973968506,
+      "learning_rate": 0.000161610022326966,
+      "loss": 0.687,
+      "step": 4235
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 2.7482621669769287,
+      "learning_rate": 0.00016157281071694366,
+      "loss": 0.4221,
+      "step": 4236
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 3.1002767086029053,
+      "learning_rate": 0.00016153559910692136,
+      "loss": 0.2666,
+      "step": 4237
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.9085137844085693,
+      "learning_rate": 0.000161498387496899,
+      "loss": 0.4164,
+      "step": 4238
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 2.563671112060547,
+      "learning_rate": 0.00016146117588687668,
+      "loss": 0.3175,
+      "step": 4239
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 3.7906312942504883,
+      "learning_rate": 0.00016142396427685439,
+      "loss": 1.8977,
+      "step": 4240
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 4.173704147338867,
+      "learning_rate": 0.00016138675266683203,
+      "loss": 0.4102,
+      "step": 4241
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 4.604090690612793,
+      "learning_rate": 0.0001613495410568097,
+      "loss": 1.782,
+      "step": 4242
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 2.0995211601257324,
+      "learning_rate": 0.00016131232944678736,
+      "loss": 0.2092,
+      "step": 4243
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 3.1292269229888916,
+      "learning_rate": 0.00016127511783676506,
+      "loss": 0.341,
+      "step": 4244
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 2.0066728591918945,
+      "learning_rate": 0.00016123790622674274,
+      "loss": 0.3798,
+      "step": 4245
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 2.5838475227355957,
+      "learning_rate": 0.00016120069461672039,
+      "loss": 0.3337,
+      "step": 4246
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 3.64768385887146,
+      "learning_rate": 0.0001611634830066981,
+      "loss": 0.5525,
+      "step": 4247
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 5.899421215057373,
+      "learning_rate": 0.00016112627139667574,
+      "loss": 0.2683,
+      "step": 4248
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 2.8585946559906006,
+      "learning_rate": 0.0001610890597866534,
+      "loss": 0.3204,
+      "step": 4249
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.4531258344650269,
+      "learning_rate": 0.00016105184817663112,
+      "loss": 0.1442,
+      "step": 4250
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 2.3642122745513916,
+      "learning_rate": 0.00016101463656660876,
+      "loss": 0.2424,
+      "step": 4251
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 2.2466611862182617,
+      "learning_rate": 0.00016097742495658644,
+      "loss": 0.1768,
+      "step": 4252
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.5451269149780273,
+      "learning_rate": 0.00016094021334656412,
+      "loss": 0.065,
+      "step": 4253
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 3.8907992839813232,
+      "learning_rate": 0.0001609030017365418,
+      "loss": 0.3097,
+      "step": 4254
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 2.250171422958374,
+      "learning_rate": 0.00016086579012651947,
+      "loss": 0.908,
+      "step": 4255
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 5.258478164672852,
+      "learning_rate": 0.00016082857851649712,
+      "loss": 1.5232,
+      "step": 4256
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 2.6386005878448486,
+      "learning_rate": 0.0001607913669064748,
+      "loss": 0.4984,
+      "step": 4257
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.1076130867004395,
+      "learning_rate": 0.0001607541552964525,
+      "loss": 0.242,
+      "step": 4258
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.5664863586425781,
+      "learning_rate": 0.00016071694368643014,
+      "loss": 0.1759,
+      "step": 4259
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 2.852780342102051,
+      "learning_rate": 0.00016067973207640782,
+      "loss": 0.3636,
+      "step": 4260
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.7188012599945068,
+      "learning_rate": 0.0001606425204663855,
+      "loss": 0.2137,
+      "step": 4261
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.521380066871643,
+      "learning_rate": 0.00016060530885636317,
+      "loss": 0.2769,
+      "step": 4262
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.657301902770996,
+      "learning_rate": 0.00016056809724634085,
+      "loss": 0.3082,
+      "step": 4263
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 2.6540775299072266,
+      "learning_rate": 0.0001605308856363185,
+      "loss": 0.4584,
+      "step": 4264
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.5712755918502808,
+      "learning_rate": 0.0001604936740262962,
+      "loss": 0.1983,
+      "step": 4265
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 2.1187055110931396,
+      "learning_rate": 0.00016045646241627387,
+      "loss": 0.3515,
+      "step": 4266
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 4.760385036468506,
+      "learning_rate": 0.00016041925080625152,
+      "loss": 2.0062,
+      "step": 4267
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 12.318647384643555,
+      "learning_rate": 0.00016038203919622923,
+      "loss": 0.6059,
+      "step": 4268
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.8167134523391724,
+      "learning_rate": 0.00016034482758620688,
+      "loss": 0.1795,
+      "step": 4269
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 3.7228429317474365,
+      "learning_rate": 0.00016030761597618455,
+      "loss": 0.4189,
+      "step": 4270
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 2.039379596710205,
+      "learning_rate": 0.00016027040436616225,
+      "loss": 0.2122,
+      "step": 4271
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 3.154689073562622,
+      "learning_rate": 0.0001602331927561399,
+      "loss": 0.3161,
+      "step": 4272
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 2.978641986846924,
+      "learning_rate": 0.00016019598114611758,
+      "loss": 0.4632,
+      "step": 4273
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.5532993078231812,
+      "learning_rate": 0.00016015876953609523,
+      "loss": 0.2329,
+      "step": 4274
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.4337552785873413,
+      "learning_rate": 0.00016012155792607293,
+      "loss": 0.2356,
+      "step": 4275
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 2.861510992050171,
+      "learning_rate": 0.0001600843463160506,
+      "loss": 0.6937,
+      "step": 4276
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 2.7515218257904053,
+      "learning_rate": 0.00016004713470602825,
+      "loss": 0.2255,
+      "step": 4277
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 2.600706100463867,
+      "learning_rate": 0.00016000992309600593,
+      "loss": 0.2059,
+      "step": 4278
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 5.621941566467285,
+      "learning_rate": 0.00015997271148598363,
+      "loss": 0.5214,
+      "step": 4279
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 2.779236078262329,
+      "learning_rate": 0.00015993549987596128,
+      "loss": 0.9023,
+      "step": 4280
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 2.4109227657318115,
+      "learning_rate": 0.00015989828826593896,
+      "loss": 0.2555,
+      "step": 4281
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.9273884296417236,
+      "learning_rate": 0.00015986107665591663,
+      "loss": 0.3355,
+      "step": 4282
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.6707803010940552,
+      "learning_rate": 0.0001598238650458943,
+      "loss": 0.295,
+      "step": 4283
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 14.171672821044922,
+      "learning_rate": 0.00015978665343587198,
+      "loss": 4.537,
+      "step": 4284
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 2.2641332149505615,
+      "learning_rate": 0.00015974944182584963,
+      "loss": 0.502,
+      "step": 4285
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.5345269441604614,
+      "learning_rate": 0.00015971223021582734,
+      "loss": 0.4606,
+      "step": 4286
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.9413623809814453,
+      "learning_rate": 0.00015967501860580499,
+      "loss": 0.3375,
+      "step": 4287
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.7889741659164429,
+      "learning_rate": 0.00015963780699578266,
+      "loss": 0.2946,
+      "step": 4288
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 3.114777088165283,
+      "learning_rate": 0.00015960059538576036,
+      "loss": 0.5715,
+      "step": 4289
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 2.282841920852661,
+      "learning_rate": 0.000159563383775738,
+      "loss": 0.3345,
+      "step": 4290
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.8913458585739136,
+      "learning_rate": 0.0001595261721657157,
+      "loss": 0.3601,
+      "step": 4291
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.8621262311935425,
+      "learning_rate": 0.00015948896055569334,
+      "loss": 0.0981,
+      "step": 4292
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 3.931448221206665,
+      "learning_rate": 0.00015945174894567104,
+      "loss": 0.4209,
+      "step": 4293
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.0255603790283203,
+      "learning_rate": 0.00015941453733564872,
+      "loss": 0.1596,
+      "step": 4294
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 2.5541460514068604,
+      "learning_rate": 0.00015937732572562636,
+      "loss": 0.5015,
+      "step": 4295
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.8127018213272095,
+      "learning_rate": 0.00015934011411560407,
+      "loss": 0.1114,
+      "step": 4296
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.7823874950408936,
+      "learning_rate": 0.00015930290250558174,
+      "loss": 0.1683,
+      "step": 4297
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 3.2042083740234375,
+      "learning_rate": 0.0001592656908955594,
+      "loss": 0.4738,
+      "step": 4298
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.9896401166915894,
+      "learning_rate": 0.00015922847928553707,
+      "loss": 0.0851,
+      "step": 4299
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.8025238513946533,
+      "learning_rate": 0.00015919126767551474,
+      "loss": 0.733,
+      "step": 4300
+    },
+    {
+      "epoch": 1.51,
+      "eval_loss": 0.47069793939590454,
+      "eval_runtime": 51.3136,
+      "eval_samples_per_second": 42.25,
+      "eval_steps_per_second": 10.563,
+      "eval_wer": 0.44680851063829785,
+      "step": 4300
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 2.523890495300293,
+      "learning_rate": 0.00015915405606549242,
+      "loss": 0.2174,
+      "step": 4301
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 7.219611644744873,
+      "learning_rate": 0.0001591168444554701,
+      "loss": 1.3755,
+      "step": 4302
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 2.2744252681732178,
+      "learning_rate": 0.00015907963284544777,
+      "loss": 0.2139,
+      "step": 4303
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 5.458730220794678,
+      "learning_rate": 0.00015904242123542545,
+      "loss": 1.3734,
+      "step": 4304
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 3.0790417194366455,
+      "learning_rate": 0.0001590052096254031,
+      "loss": 0.8824,
+      "step": 4305
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 2.6385726928710938,
+      "learning_rate": 0.00015896799801538077,
+      "loss": 0.629,
+      "step": 4306
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 3.4197399616241455,
+      "learning_rate": 0.00015893078640535847,
+      "loss": 0.6337,
+      "step": 4307
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.8060104846954346,
+      "learning_rate": 0.00015889357479533612,
+      "loss": 0.2284,
+      "step": 4308
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.9204670190811157,
+      "learning_rate": 0.0001588563631853138,
+      "loss": 0.2744,
+      "step": 4309
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.3821399211883545,
+      "learning_rate": 0.0001588191515752915,
+      "loss": 0.2596,
+      "step": 4310
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 2.8792121410369873,
+      "learning_rate": 0.00015878193996526915,
+      "loss": 0.8357,
+      "step": 4311
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.4682904481887817,
+      "learning_rate": 0.00015874472835524683,
+      "loss": 0.3634,
+      "step": 4312
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 2.5377604961395264,
+      "learning_rate": 0.00015870751674522447,
+      "loss": 0.1408,
+      "step": 4313
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 2.061809539794922,
+      "learning_rate": 0.00015867030513520218,
+      "loss": 0.1885,
+      "step": 4314
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 3.3205690383911133,
+      "learning_rate": 0.00015863309352517985,
+      "loss": 0.3503,
+      "step": 4315
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.4646776914596558,
+      "learning_rate": 0.0001585958819151575,
+      "loss": 0.1579,
+      "step": 4316
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 4.019704818725586,
+      "learning_rate": 0.0001585586703051352,
+      "loss": 0.4413,
+      "step": 4317
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 2.0454628467559814,
+      "learning_rate": 0.00015852145869511285,
+      "loss": 0.179,
+      "step": 4318
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 3.3551511764526367,
+      "learning_rate": 0.00015848424708509053,
+      "loss": 0.3165,
+      "step": 4319
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 2.5540342330932617,
+      "learning_rate": 0.0001584470354750682,
+      "loss": 0.1543,
+      "step": 4320
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 3.365748643875122,
+      "learning_rate": 0.00015840982386504588,
+      "loss": 0.3139,
+      "step": 4321
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.9014160633087158,
+      "learning_rate": 0.00015837261225502356,
+      "loss": 0.3596,
+      "step": 4322
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 2.817349910736084,
+      "learning_rate": 0.00015833540064500123,
+      "loss": 0.4239,
+      "step": 4323
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 7.915454864501953,
+      "learning_rate": 0.0001582981890349789,
+      "loss": 0.356,
+      "step": 4324
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 4.057788372039795,
+      "learning_rate": 0.00015826097742495658,
+      "loss": 0.2209,
+      "step": 4325
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 3.6405909061431885,
+      "learning_rate": 0.00015822376581493423,
+      "loss": 0.1062,
+      "step": 4326
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 2.823045015335083,
+      "learning_rate": 0.0001581865542049119,
+      "loss": 0.4549,
+      "step": 4327
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 4.013254642486572,
+      "learning_rate": 0.0001581493425948896,
+      "loss": 0.5598,
+      "step": 4328
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 2.840559244155884,
+      "learning_rate": 0.00015811213098486726,
+      "loss": 0.2415,
+      "step": 4329
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 2.6936604976654053,
+      "learning_rate": 0.00015807491937484494,
+      "loss": 0.7174,
+      "step": 4330
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 2.872133493423462,
+      "learning_rate": 0.00015803770776482258,
+      "loss": 0.8422,
+      "step": 4331
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.364221453666687,
+      "learning_rate": 0.0001580004961548003,
+      "loss": 0.2798,
+      "step": 4332
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.7011353969573975,
+      "learning_rate": 0.00015796328454477796,
+      "loss": 0.3753,
+      "step": 4333
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.4171128273010254,
+      "learning_rate": 0.0001579260729347556,
+      "loss": 0.1875,
+      "step": 4334
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 3.3720600605010986,
+      "learning_rate": 0.00015788886132473331,
+      "loss": 0.8102,
+      "step": 4335
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.8144382238388062,
+      "learning_rate": 0.000157851649714711,
+      "loss": 0.2117,
+      "step": 4336
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 2.3069920539855957,
+      "learning_rate": 0.00015781443810468864,
+      "loss": 0.8746,
+      "step": 4337
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.95841646194458,
+      "learning_rate": 0.00015777722649466634,
+      "loss": 0.3763,
+      "step": 4338
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.462180495262146,
+      "learning_rate": 0.000157740014884644,
+      "loss": 0.1606,
+      "step": 4339
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.9129983186721802,
+      "learning_rate": 0.00015770280327462167,
+      "loss": 0.2706,
+      "step": 4340
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.6017885208129883,
+      "learning_rate": 0.00015766559166459934,
+      "loss": 0.2079,
+      "step": 4341
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 2.809229612350464,
+      "learning_rate": 0.00015762838005457702,
+      "loss": 0.5534,
+      "step": 4342
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.4012563228607178,
+      "learning_rate": 0.0001575911684445547,
+      "loss": 0.1605,
+      "step": 4343
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.8997706174850464,
+      "learning_rate": 0.00015755395683453234,
+      "loss": 0.1523,
+      "step": 4344
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 5.734255313873291,
+      "learning_rate": 0.00015751674522451005,
+      "loss": 0.3817,
+      "step": 4345
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 2.4937775135040283,
+      "learning_rate": 0.00015747953361448772,
+      "loss": 0.3117,
+      "step": 4346
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": NaN,
+      "learning_rate": 0.00015747953361448772,
+      "loss": 0.1006,
+      "step": 4347
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.9749798774719238,
+      "learning_rate": 0.00015744232200446537,
+      "loss": 0.4598,
+      "step": 4348
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.340526819229126,
+      "learning_rate": 0.00015740511039444305,
+      "loss": 0.0752,
+      "step": 4349
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 2.0161750316619873,
+      "learning_rate": 0.00015736789878442072,
+      "loss": 0.2629,
+      "step": 4350
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 2.0471723079681396,
+      "learning_rate": 0.0001573306871743984,
+      "loss": 0.3901,
+      "step": 4351
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 3.22011661529541,
+      "learning_rate": 0.00015729347556437607,
+      "loss": 0.2667,
+      "step": 4352
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 2.5319764614105225,
+      "learning_rate": 0.00015725626395435372,
+      "loss": 0.2311,
+      "step": 4353
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 3.327425718307495,
+      "learning_rate": 0.00015721905234433142,
+      "loss": 0.2222,
+      "step": 4354
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.8986114263534546,
+      "learning_rate": 0.0001571818407343091,
+      "loss": 0.5881,
+      "step": 4355
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.7233716249465942,
+      "learning_rate": 0.00015714462912428675,
+      "loss": 0.6135,
+      "step": 4356
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.6733200550079346,
+      "learning_rate": 0.00015710741751426445,
+      "loss": 0.2479,
+      "step": 4357
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.7937635183334351,
+      "learning_rate": 0.0001570702059042421,
+      "loss": 0.0792,
+      "step": 4358
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 2.3683009147644043,
+      "learning_rate": 0.00015703299429421978,
+      "loss": 0.2897,
+      "step": 4359
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 3.6302194595336914,
+      "learning_rate": 0.00015699578268419748,
+      "loss": 0.5056,
+      "step": 4360
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 2.038682460784912,
+      "learning_rate": 0.00015695857107417513,
+      "loss": 0.339,
+      "step": 4361
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 2.3707022666931152,
+      "learning_rate": 0.0001569213594641528,
+      "loss": 0.5805,
+      "step": 4362
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 11.967528343200684,
+      "learning_rate": 0.00015688414785413045,
+      "loss": 3.6802,
+      "step": 4363
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 3.1059775352478027,
+      "learning_rate": 0.00015684693624410816,
+      "loss": 0.6931,
+      "step": 4364
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.6652494668960571,
+      "learning_rate": 0.00015680972463408583,
+      "loss": 0.3522,
+      "step": 4365
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 2.2658798694610596,
+      "learning_rate": 0.00015677251302406348,
+      "loss": 0.3224,
+      "step": 4366
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 14.272518157958984,
+      "learning_rate": 0.00015673530141404118,
+      "loss": 3.4446,
+      "step": 4367
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.9729360342025757,
+      "learning_rate": 0.00015669808980401886,
+      "loss": 0.5131,
+      "step": 4368
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 2.555955648422241,
+      "learning_rate": 0.0001566608781939965,
+      "loss": 0.4208,
+      "step": 4369
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.9813282489776611,
+      "learning_rate": 0.00015662366658397418,
+      "loss": 0.1237,
+      "step": 4370
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 2.0302422046661377,
+      "learning_rate": 0.00015658645497395186,
+      "loss": 0.3017,
+      "step": 4371
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 2.8068108558654785,
+      "learning_rate": 0.00015654924336392953,
+      "loss": 0.3008,
+      "step": 4372
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 2.404326915740967,
+      "learning_rate": 0.0001565120317539072,
+      "loss": 0.4064,
+      "step": 4373
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.2311917543411255,
+      "learning_rate": 0.00015647482014388486,
+      "loss": 0.1547,
+      "step": 4374
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 2.6930012702941895,
+      "learning_rate": 0.00015643760853386256,
+      "loss": 0.3034,
+      "step": 4375
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.130953073501587,
+      "learning_rate": 0.0001564003969238402,
+      "loss": 0.0701,
+      "step": 4376
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.6812334060668945,
+      "learning_rate": 0.0001563631853138179,
+      "loss": 0.1034,
+      "step": 4377
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 3.887232780456543,
+      "learning_rate": 0.0001563259737037956,
+      "loss": 0.8612,
+      "step": 4378
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 4.974242687225342,
+      "learning_rate": 0.00015628876209377324,
+      "loss": 0.3702,
+      "step": 4379
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 2.9421982765197754,
+      "learning_rate": 0.00015625155048375091,
+      "loss": 0.6489,
+      "step": 4380
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.6786479949951172,
+      "learning_rate": 0.00015621433887372862,
+      "loss": 0.2916,
+      "step": 4381
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 2.749591112136841,
+      "learning_rate": 0.00015617712726370627,
+      "loss": 0.5676,
+      "step": 4382
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.2755649089813232,
+      "learning_rate": 0.00015613991565368394,
+      "loss": 0.1627,
+      "step": 4383
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.326809048652649,
+      "learning_rate": 0.0001561027040436616,
+      "loss": 0.1395,
+      "step": 4384
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.9218014478683472,
+      "learning_rate": 0.0001560654924336393,
+      "loss": 0.1224,
+      "step": 4385
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 2.6354215145111084,
+      "learning_rate": 0.00015602828082361697,
+      "loss": 0.5855,
+      "step": 4386
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 7.072910308837891,
+      "learning_rate": 0.00015599106921359462,
+      "loss": 1.493,
+      "step": 4387
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.5853562355041504,
+      "learning_rate": 0.00015595385760357232,
+      "loss": 0.2689,
+      "step": 4388
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 6.221762657165527,
+      "learning_rate": 0.00015591664599354997,
+      "loss": 0.4643,
+      "step": 4389
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 3.7703330516815186,
+      "learning_rate": 0.00015587943438352765,
+      "loss": 0.4222,
+      "step": 4390
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 2.02024245262146,
+      "learning_rate": 0.00015584222277350532,
+      "loss": 0.5177,
+      "step": 4391
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 2.0577681064605713,
+      "learning_rate": 0.000155805011163483,
+      "loss": 0.2802,
+      "step": 4392
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 2.6712489128112793,
+      "learning_rate": 0.00015576779955346067,
+      "loss": 0.1628,
+      "step": 4393
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 2.0091423988342285,
+      "learning_rate": 0.00015573058794343832,
+      "loss": 0.3688,
+      "step": 4394
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 5.743954181671143,
+      "learning_rate": 0.000155693376333416,
+      "loss": 0.5611,
+      "step": 4395
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 2.4916129112243652,
+      "learning_rate": 0.0001556561647233937,
+      "loss": 0.286,
+      "step": 4396
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 6.647517204284668,
+      "learning_rate": 0.00015561895311337135,
+      "loss": 0.7609,
+      "step": 4397
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.73363196849823,
+      "learning_rate": 0.00015558174150334902,
+      "loss": 0.1074,
+      "step": 4398
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 2.5416290760040283,
+      "learning_rate": 0.00015554452989332673,
+      "loss": 0.2281,
+      "step": 4399
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.1465420722961426,
+      "learning_rate": 0.00015550731828330438,
+      "loss": 0.123,
+      "step": 4400
+    },
+    {
+      "epoch": 1.54,
+      "eval_loss": 0.3980734944343567,
+      "eval_runtime": 51.2139,
+      "eval_samples_per_second": 42.332,
+      "eval_steps_per_second": 10.583,
+      "eval_wer": 0.379173153433662,
+      "step": 4400
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.1583079099655151,
+      "learning_rate": 0.00015547010667328205,
+      "loss": 0.0627,
+      "step": 4401
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.6864182949066162,
+      "learning_rate": 0.0001554328950632597,
+      "loss": 0.1769,
+      "step": 4402
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 2.6987173557281494,
+      "learning_rate": 0.0001553956834532374,
+      "loss": 0.1046,
+      "step": 4403
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 4.390762805938721,
+      "learning_rate": 0.00015535847184321508,
+      "loss": 0.0657,
+      "step": 4404
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 2.5673904418945312,
+      "learning_rate": 0.00015532126023319273,
+      "loss": 0.8568,
+      "step": 4405
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.5236790180206299,
+      "learning_rate": 0.00015528404862317043,
+      "loss": 0.1907,
+      "step": 4406
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.6562654972076416,
+      "learning_rate": 0.00015524683701314808,
+      "loss": 0.1583,
+      "step": 4407
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 2.632766008377075,
+      "learning_rate": 0.00015520962540312576,
+      "loss": 0.5559,
+      "step": 4408
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 2.8173253536224365,
+      "learning_rate": 0.00015517241379310346,
+      "loss": 0.3249,
+      "step": 4409
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 3.325995922088623,
+      "learning_rate": 0.0001551352021830811,
+      "loss": 0.4845,
+      "step": 4410
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.7911338806152344,
+      "learning_rate": 0.00015509799057305878,
+      "loss": 0.2434,
+      "step": 4411
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 2.012998342514038,
+      "learning_rate": 0.00015506077896303646,
+      "loss": 0.2877,
+      "step": 4412
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.2822898626327515,
+      "learning_rate": 0.00015502356735301413,
+      "loss": 0.2108,
+      "step": 4413
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.760624885559082,
+      "learning_rate": 0.0001549863557429918,
+      "loss": 0.1786,
+      "step": 4414
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 2.9734385013580322,
+      "learning_rate": 0.00015494914413296946,
+      "loss": 0.3499,
+      "step": 4415
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.6788557767868042,
+      "learning_rate": 0.00015491193252294713,
+      "loss": 0.2782,
+      "step": 4416
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 2.383653163909912,
+      "learning_rate": 0.00015487472091292484,
+      "loss": 0.4833,
+      "step": 4417
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 4.293880939483643,
+      "learning_rate": 0.00015483750930290249,
+      "loss": 0.5878,
+      "step": 4418
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 2.733428716659546,
+      "learning_rate": 0.00015480029769288016,
+      "loss": 0.2317,
+      "step": 4419
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.5017122030258179,
+      "learning_rate": 0.0001547630860828578,
+      "loss": 0.0682,
+      "step": 4420
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 3.120365619659424,
+      "learning_rate": 0.0001547258744728355,
+      "loss": 0.3306,
+      "step": 4421
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.6654366254806519,
+      "learning_rate": 0.0001546886628628132,
+      "loss": 0.0215,
+      "step": 4422
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 2.2988080978393555,
+      "learning_rate": 0.00015465145125279084,
+      "loss": 0.282,
+      "step": 4423
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 3.34647536277771,
+      "learning_rate": 0.00015461423964276854,
+      "loss": 0.3574,
+      "step": 4424
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.9335713386535645,
+      "learning_rate": 0.00015457702803274622,
+      "loss": 0.1248,
+      "step": 4425
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.3534259796142578,
+      "learning_rate": 0.00015453981642272387,
+      "loss": 0.0529,
+      "step": 4426
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 8.45742130279541,
+      "learning_rate": 0.00015450260481270157,
+      "loss": 1.6093,
+      "step": 4427
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.5706729888916016,
+      "learning_rate": 0.00015446539320267922,
+      "loss": 0.0306,
+      "step": 4428
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 6.064884662628174,
+      "learning_rate": 0.0001544281815926569,
+      "loss": 0.898,
+      "step": 4429
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 2.807861328125,
+      "learning_rate": 0.0001543909699826346,
+      "loss": 1.1501,
+      "step": 4430
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.6200414896011353,
+      "learning_rate": 0.00015435375837261224,
+      "loss": 0.4798,
+      "step": 4431
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 2.61049485206604,
+      "learning_rate": 0.00015431654676258992,
+      "loss": 0.6317,
+      "step": 4432
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.9506374597549438,
+      "learning_rate": 0.00015427933515256757,
+      "loss": 0.2907,
+      "step": 4433
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.570434331893921,
+      "learning_rate": 0.00015424212354254527,
+      "loss": 0.2544,
+      "step": 4434
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 2.3015689849853516,
+      "learning_rate": 0.00015420491193252295,
+      "loss": 0.2989,
+      "step": 4435
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.8835831880569458,
+      "learning_rate": 0.0001541677003225006,
+      "loss": 0.3247,
+      "step": 4436
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 2.0309693813323975,
+      "learning_rate": 0.00015413048871247827,
+      "loss": 0.3081,
+      "step": 4437
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.200584888458252,
+      "learning_rate": 0.00015409327710245595,
+      "loss": 0.1762,
+      "step": 4438
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 3.047585964202881,
+      "learning_rate": 0.00015405606549243362,
+      "loss": 0.3421,
+      "step": 4439
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 2.6859099864959717,
+      "learning_rate": 0.0001540188538824113,
+      "loss": 0.2895,
+      "step": 4440
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 3.2242605686187744,
+      "learning_rate": 0.00015398164227238895,
+      "loss": 0.5598,
+      "step": 4441
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 3.5308659076690674,
+      "learning_rate": 0.00015394443066236665,
+      "loss": 0.581,
+      "step": 4442
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 2.4729762077331543,
+      "learning_rate": 0.00015390721905234433,
+      "loss": 0.371,
+      "step": 4443
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.6830202341079712,
+      "learning_rate": 0.00015387000744232198,
+      "loss": 0.2435,
+      "step": 4444
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 2.3365495204925537,
+      "learning_rate": 0.00015383279583229968,
+      "loss": 0.3798,
+      "step": 4445
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 2.603057384490967,
+      "learning_rate": 0.00015379558422227733,
+      "loss": 0.2935,
+      "step": 4446
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.5318717956542969,
+      "learning_rate": 0.000153758372612255,
+      "loss": 0.1117,
+      "step": 4447
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 3.2214760780334473,
+      "learning_rate": 0.0001537211610022327,
+      "loss": 0.5403,
+      "step": 4448
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.5494588613510132,
+      "learning_rate": 0.00015368394939221035,
+      "loss": 0.1355,
+      "step": 4449
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 2.5164105892181396,
+      "learning_rate": 0.00015364673778218803,
+      "loss": 0.4476,
+      "step": 4450
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 2.3167803287506104,
+      "learning_rate": 0.00015360952617216568,
+      "loss": 0.1465,
+      "step": 4451
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.139276146888733,
+      "learning_rate": 0.00015357231456214338,
+      "loss": 0.0653,
+      "step": 4452
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 9.703216552734375,
+      "learning_rate": 0.00015353510295212106,
+      "loss": 0.5352,
+      "step": 4453
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 2.554642915725708,
+      "learning_rate": 0.0001534978913420987,
+      "loss": 0.0762,
+      "step": 4454
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 2.5939784049987793,
+      "learning_rate": 0.0001534606797320764,
+      "loss": 0.7377,
+      "step": 4455
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 2.136479139328003,
+      "learning_rate": 0.00015342346812205408,
+      "loss": 0.8079,
+      "step": 4456
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.0142302513122559,
+      "learning_rate": 0.00015338625651203173,
+      "loss": 0.2065,
+      "step": 4457
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.6144118309020996,
+      "learning_rate": 0.0001533490449020094,
+      "loss": 0.3742,
+      "step": 4458
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.8569650650024414,
+      "learning_rate": 0.00015331183329198709,
+      "loss": 0.2521,
+      "step": 4459
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.7510970830917358,
+      "learning_rate": 0.00015327462168196476,
+      "loss": 0.2786,
+      "step": 4460
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 2.130952835083008,
+      "learning_rate": 0.00015323741007194244,
+      "loss": 0.1905,
+      "step": 4461
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 2.0437982082366943,
+      "learning_rate": 0.00015320019846192009,
+      "loss": 0.367,
+      "step": 4462
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 2.792440891265869,
+      "learning_rate": 0.0001531629868518978,
+      "loss": 0.5759,
+      "step": 4463
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.3819493055343628,
+      "learning_rate": 0.00015312577524187544,
+      "loss": 0.2252,
+      "step": 4464
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 4.414833068847656,
+      "learning_rate": 0.0001530885636318531,
+      "loss": 0.4762,
+      "step": 4465
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 2.5327303409576416,
+      "learning_rate": 0.00015305135202183082,
+      "loss": 0.4483,
+      "step": 4466
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 2.1207656860351562,
+      "learning_rate": 0.00015301414041180846,
+      "loss": 0.2676,
+      "step": 4467
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 2.1019747257232666,
+      "learning_rate": 0.00015297692880178614,
+      "loss": 0.3797,
+      "step": 4468
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 2.4563004970550537,
+      "learning_rate": 0.00015293971719176384,
+      "loss": 0.2758,
+      "step": 4469
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 3.147643804550171,
+      "learning_rate": 0.0001529025055817415,
+      "loss": 0.9069,
+      "step": 4470
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 4.162635803222656,
+      "learning_rate": 0.00015286529397171917,
+      "loss": 0.6038,
+      "step": 4471
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 4.253012657165527,
+      "learning_rate": 0.00015282808236169682,
+      "loss": 0.8356,
+      "step": 4472
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 2.0794060230255127,
+      "learning_rate": 0.00015279087075167452,
+      "loss": 0.3112,
+      "step": 4473
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 3.6223506927490234,
+      "learning_rate": 0.0001527536591416522,
+      "loss": 0.3925,
+      "step": 4474
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 2.9207382202148438,
+      "learning_rate": 0.00015271644753162984,
+      "loss": 0.3373,
+      "step": 4475
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 2.2720437049865723,
+      "learning_rate": 0.00015267923592160755,
+      "loss": 0.2101,
+      "step": 4476
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.576826572418213,
+      "learning_rate": 0.0001526420243115852,
+      "loss": 0.2305,
+      "step": 4477
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.4385606646537781,
+      "learning_rate": 0.00015260481270156287,
+      "loss": 0.0335,
+      "step": 4478
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 2.5743887424468994,
+      "learning_rate": 0.00015256760109154055,
+      "loss": 0.8183,
+      "step": 4479
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.7291278839111328,
+      "learning_rate": 0.00015253038948151822,
+      "loss": 0.334,
+      "step": 4480
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.7327170372009277,
+      "learning_rate": 0.0001524931778714959,
+      "loss": 0.4015,
+      "step": 4481
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.8238621950149536,
+      "learning_rate": 0.00015245596626147355,
+      "loss": 0.4654,
+      "step": 4482
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.92159104347229,
+      "learning_rate": 0.00015241875465145122,
+      "loss": 0.4713,
+      "step": 4483
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 6.046019077301025,
+      "learning_rate": 0.00015238154304142893,
+      "loss": 0.5623,
+      "step": 4484
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 2.2049782276153564,
+      "learning_rate": 0.00015234433143140657,
+      "loss": 0.3545,
+      "step": 4485
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.759090542793274,
+      "learning_rate": 0.00015230711982138425,
+      "loss": 0.2411,
+      "step": 4486
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 2.1252946853637695,
+      "learning_rate": 0.00015226990821136195,
+      "loss": 0.2737,
+      "step": 4487
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.3396649360656738,
+      "learning_rate": 0.0001522326966013396,
+      "loss": 0.1346,
+      "step": 4488
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.8554527759552002,
+      "learning_rate": 0.00015219548499131728,
+      "loss": 0.3038,
+      "step": 4489
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.3641473054885864,
+      "learning_rate": 0.00015215827338129493,
+      "loss": 0.1102,
+      "step": 4490
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 3.5774457454681396,
+      "learning_rate": 0.00015212106177127263,
+      "loss": 0.4797,
+      "step": 4491
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 4.142420768737793,
+      "learning_rate": 0.0001520838501612503,
+      "loss": 0.3904,
+      "step": 4492
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.1774590015411377,
+      "learning_rate": 0.00015204663855122795,
+      "loss": 0.0748,
+      "step": 4493
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 3.950049638748169,
+      "learning_rate": 0.00015200942694120566,
+      "loss": 0.3689,
+      "step": 4494
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.5134997963905334,
+      "learning_rate": 0.0001519722153311833,
+      "loss": 0.0291,
+      "step": 4495
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 6.9983296394348145,
+      "learning_rate": 0.00015193500372116098,
+      "loss": 2.1653,
+      "step": 4496
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 4.325428485870361,
+      "learning_rate": 0.00015189779211113868,
+      "loss": 0.4339,
+      "step": 4497
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 5.906067371368408,
+      "learning_rate": 0.00015186058050111633,
+      "loss": 0.5815,
+      "step": 4498
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 6.516275882720947,
+      "learning_rate": 0.000151823368891094,
+      "loss": 0.2292,
+      "step": 4499
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 5.414793491363525,
+      "learning_rate": 0.00015178615728107168,
+      "loss": 0.3685,
+      "step": 4500
+    },
+    {
+      "epoch": 1.58,
+      "eval_loss": 0.4531491696834564,
+      "eval_runtime": 51.0994,
+      "eval_samples_per_second": 42.427,
+      "eval_steps_per_second": 10.607,
+      "eval_wer": 0.3689673066943435,
+      "step": 4500
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 5.927810192108154,
+      "learning_rate": 0.00015174894567104936,
+      "loss": 0.4957,
+      "step": 4501
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 3.2968862056732178,
+      "learning_rate": 0.00015171173406102704,
+      "loss": 0.1763,
+      "step": 4502
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 18.59258270263672,
+      "learning_rate": 0.00015167452245100468,
+      "loss": 0.3655,
+      "step": 4503
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 9.49154281616211,
+      "learning_rate": 0.00015163731084098236,
+      "loss": 1.3948,
+      "step": 4504
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.5831422805786133,
+      "learning_rate": 0.00015160009923096006,
+      "loss": 0.5639,
+      "step": 4505
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.5149224996566772,
+      "learning_rate": 0.0001515628876209377,
+      "loss": 0.361,
+      "step": 4506
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.7891191244125366,
+      "learning_rate": 0.0001515256760109154,
+      "loss": 0.4626,
+      "step": 4507
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.3684577941894531,
+      "learning_rate": 0.00015148846440089306,
+      "loss": 0.2943,
+      "step": 4508
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.371821641921997,
+      "learning_rate": 0.00015145125279087074,
+      "loss": 0.1162,
+      "step": 4509
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.8998194932937622,
+      "learning_rate": 0.00015141404118084842,
+      "loss": 0.2725,
+      "step": 4510
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 3.163402557373047,
+      "learning_rate": 0.00015137682957082606,
+      "loss": 0.475,
+      "step": 4511
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 3.110438108444214,
+      "learning_rate": 0.00015133961796080377,
+      "loss": 0.4981,
+      "step": 4512
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 2.5122923851013184,
+      "learning_rate": 0.00015130240635078144,
+      "loss": 0.4744,
+      "step": 4513
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 5.760143280029297,
+      "learning_rate": 0.0001512651947407591,
+      "loss": 1.8897,
+      "step": 4514
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.6696605682373047,
+      "learning_rate": 0.0001512279831307368,
+      "loss": 0.2554,
+      "step": 4515
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 2.809109687805176,
+      "learning_rate": 0.00015119077152071444,
+      "loss": 0.406,
+      "step": 4516
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.2222541570663452,
+      "learning_rate": 0.00015115355991069212,
+      "loss": 0.0803,
+      "step": 4517
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 3.3390469551086426,
+      "learning_rate": 0.00015111634830066982,
+      "loss": 0.4206,
+      "step": 4518
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.309799075126648,
+      "learning_rate": 0.00015107913669064747,
+      "loss": 0.1442,
+      "step": 4519
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 3.854196310043335,
+      "learning_rate": 0.00015104192508062515,
+      "loss": 1.496,
+      "step": 4520
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 2.176786422729492,
+      "learning_rate": 0.0001510047134706028,
+      "loss": 0.2341,
+      "step": 4521
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.9741429090499878,
+      "learning_rate": 0.0001509675018605805,
+      "loss": 0.2108,
+      "step": 4522
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 5.400284290313721,
+      "learning_rate": 0.00015093029025055817,
+      "loss": 2.0911,
+      "step": 4523
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 3.421788215637207,
+      "learning_rate": 0.00015089307864053582,
+      "loss": 0.4519,
+      "step": 4524
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 4.890323162078857,
+      "learning_rate": 0.0001508558670305135,
+      "loss": 0.5325,
+      "step": 4525
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 3.7187812328338623,
+      "learning_rate": 0.00015081865542049117,
+      "loss": 0.392,
+      "step": 4526
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 2.6591031551361084,
+      "learning_rate": 0.00015078144381046885,
+      "loss": 0.3085,
+      "step": 4527
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 3.004119873046875,
+      "learning_rate": 0.00015074423220044653,
+      "loss": 0.3717,
+      "step": 4528
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.017032265663147,
+      "learning_rate": 0.0001507070205904242,
+      "loss": 0.0585,
+      "step": 4529
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 2.0729451179504395,
+      "learning_rate": 0.00015066980898040188,
+      "loss": 0.393,
+      "step": 4530
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.5264289379119873,
+      "learning_rate": 0.00015063259737037955,
+      "loss": 0.3011,
+      "step": 4531
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.535727858543396,
+      "learning_rate": 0.0001505953857603572,
+      "loss": 0.3631,
+      "step": 4532
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 2.5578010082244873,
+      "learning_rate": 0.0001505581741503349,
+      "loss": 0.3524,
+      "step": 4533
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 2.180941104888916,
+      "learning_rate": 0.00015052096254031255,
+      "loss": 0.3525,
+      "step": 4534
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.2360469102859497,
+      "learning_rate": 0.00015048375093029023,
+      "loss": 0.1369,
+      "step": 4535
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.1341615915298462,
+      "learning_rate": 0.00015044653932026793,
+      "loss": 0.1489,
+      "step": 4536
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.9309412240982056,
+      "learning_rate": 0.00015040932771024558,
+      "loss": 0.2772,
+      "step": 4537
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.6193166971206665,
+      "learning_rate": 0.00015037211610022326,
+      "loss": 0.3045,
+      "step": 4538
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.2819677591323853,
+      "learning_rate": 0.0001503349044902009,
+      "loss": 0.2624,
+      "step": 4539
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 2.1933364868164062,
+      "learning_rate": 0.0001502976928801786,
+      "loss": 0.2117,
+      "step": 4540
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 2.030672073364258,
+      "learning_rate": 0.00015026048127015628,
+      "loss": 0.2636,
+      "step": 4541
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 3.077730894088745,
+      "learning_rate": 0.00015022326966013393,
+      "loss": 0.4608,
+      "step": 4542
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.8299646377563477,
+      "learning_rate": 0.00015018605805011163,
+      "loss": 0.1079,
+      "step": 4543
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 2.263606309890747,
+      "learning_rate": 0.0001501488464400893,
+      "loss": 0.2057,
+      "step": 4544
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.6388328075408936,
+      "learning_rate": 0.00015011163483006696,
+      "loss": 0.2149,
+      "step": 4545
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.8084484338760376,
+      "learning_rate": 0.00015007442322004464,
+      "loss": 0.2344,
+      "step": 4546
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.729175329208374,
+      "learning_rate": 0.0001500372116100223,
+      "loss": 0.1099,
+      "step": 4547
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 2.0611491203308105,
+      "learning_rate": 0.00015,
+      "loss": 0.3877,
+      "step": 4548
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 4.244816303253174,
+      "learning_rate": 0.00014996278838997766,
+      "loss": 0.4705,
+      "step": 4549
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 3.561108112335205,
+      "learning_rate": 0.00014992557677995534,
+      "loss": 0.3884,
+      "step": 4550
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 3.055769205093384,
+      "learning_rate": 0.00014988836516993301,
+      "loss": 0.234,
+      "step": 4551
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 4.927545547485352,
+      "learning_rate": 0.0001498511535599107,
+      "loss": 1.9593,
+      "step": 4552
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 5.328706741333008,
+      "learning_rate": 0.00014981394194988834,
+      "loss": 1.9694,
+      "step": 4553
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 2.255901575088501,
+      "learning_rate": 0.00014977673033986601,
+      "loss": 0.1712,
+      "step": 4554
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 2.4930551052093506,
+      "learning_rate": 0.00014973951872984372,
+      "loss": 0.6505,
+      "step": 4555
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 2.2214691638946533,
+      "learning_rate": 0.00014970230711982137,
+      "loss": 0.6706,
+      "step": 4556
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 2.260343313217163,
+      "learning_rate": 0.00014966509550979904,
+      "loss": 0.4399,
+      "step": 4557
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.8949406147003174,
+      "learning_rate": 0.00014962788389977672,
+      "loss": 0.681,
+      "step": 4558
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.0256386995315552,
+      "learning_rate": 0.0001495906722897544,
+      "loss": 0.1752,
+      "step": 4559
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 2.9112942218780518,
+      "learning_rate": 0.00014955346067973207,
+      "loss": 0.2714,
+      "step": 4560
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.3259639739990234,
+      "learning_rate": 0.00014951624906970975,
+      "loss": 0.1771,
+      "step": 4561
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 2.3593876361846924,
+      "learning_rate": 0.0001494790374596874,
+      "loss": 0.51,
+      "step": 4562
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 2.028107166290283,
+      "learning_rate": 0.00014944182584966507,
+      "loss": 0.6044,
+      "step": 4563
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 2.4526591300964355,
+      "learning_rate": 0.00014940461423964277,
+      "loss": 0.3303,
+      "step": 4564
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 3.8882853984832764,
+      "learning_rate": 0.00014936740262962042,
+      "loss": 0.7958,
+      "step": 4565
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.1775007247924805,
+      "learning_rate": 0.0001493301910195981,
+      "loss": 0.128,
+      "step": 4566
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.8732891082763672,
+      "learning_rate": 0.00014929297940957577,
+      "loss": 0.1814,
+      "step": 4567
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 2.4956729412078857,
+      "learning_rate": 0.00014925576779955345,
+      "loss": 0.2284,
+      "step": 4568
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.7072712182998657,
+      "learning_rate": 0.00014921855618953112,
+      "loss": 0.114,
+      "step": 4569
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.2081888914108276,
+      "learning_rate": 0.0001491813445795088,
+      "loss": 0.0593,
+      "step": 4570
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 3.359692335128784,
+      "learning_rate": 0.00014914413296948648,
+      "loss": 0.6738,
+      "step": 4571
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 2.427903652191162,
+      "learning_rate": 0.00014910692135946412,
+      "loss": 0.2915,
+      "step": 4572
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.3507580757141113,
+      "learning_rate": 0.00014906970974944183,
+      "loss": 0.2221,
+      "step": 4573
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.6972339153289795,
+      "learning_rate": 0.00014903249813941948,
+      "loss": 0.1641,
+      "step": 4574
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 4.0066328048706055,
+      "learning_rate": 0.00014899528652939715,
+      "loss": 0.6401,
+      "step": 4575
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 2.1944432258605957,
+      "learning_rate": 0.00014895807491937483,
+      "loss": 0.2104,
+      "step": 4576
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 3.7553157806396484,
+      "learning_rate": 0.0001489208633093525,
+      "loss": 0.3944,
+      "step": 4577
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001489208633093525,
+      "loss": 1.2432,
+      "step": 4578
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.1025826930999756,
+      "learning_rate": 0.00014888365169933018,
+      "loss": 0.0429,
+      "step": 4579
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 3.72552752494812,
+      "learning_rate": 0.00014884644008930786,
+      "loss": 0.7761,
+      "step": 4580
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 2.1023457050323486,
+      "learning_rate": 0.00014880922847928553,
+      "loss": 0.5812,
+      "step": 4581
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 2.557485818862915,
+      "learning_rate": 0.0001487720168692632,
+      "loss": 0.7201,
+      "step": 4582
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.0474027395248413,
+      "learning_rate": 0.00014873480525924088,
+      "loss": 0.174,
+      "step": 4583
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.4817769527435303,
+      "learning_rate": 0.00014869759364921853,
+      "loss": 0.0965,
+      "step": 4584
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 3.8702480792999268,
+      "learning_rate": 0.0001486603820391962,
+      "loss": 0.9629,
+      "step": 4585
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 13.161911010742188,
+      "learning_rate": 0.00014862317042917388,
+      "loss": 2.8826,
+      "step": 4586
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 3.8920679092407227,
+      "learning_rate": 0.00014858595881915156,
+      "loss": 0.6731,
+      "step": 4587
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.4065778255462646,
+      "learning_rate": 0.00014854874720912923,
+      "loss": 0.2983,
+      "step": 4588
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 3.574580430984497,
+      "learning_rate": 0.0001485115355991069,
+      "loss": 0.6195,
+      "step": 4589
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 6.651939392089844,
+      "learning_rate": 0.00014847432398908459,
+      "loss": 0.2192,
+      "step": 4590
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 2.2979915142059326,
+      "learning_rate": 0.00014843711237906226,
+      "loss": 0.51,
+      "step": 4591
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.8823531866073608,
+      "learning_rate": 0.00014839990076903994,
+      "loss": 0.3496,
+      "step": 4592
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 3.272880792617798,
+      "learning_rate": 0.0001483626891590176,
+      "loss": 0.3933,
+      "step": 4593
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 2.0282230377197266,
+      "learning_rate": 0.00014832547754899526,
+      "loss": 0.5362,
+      "step": 4594
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 2.4727609157562256,
+      "learning_rate": 0.00014828826593897294,
+      "loss": 0.3841,
+      "step": 4595
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.0097224712371826,
+      "learning_rate": 0.00014825105432895061,
+      "loss": 0.1183,
+      "step": 4596
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 2.555164337158203,
+      "learning_rate": 0.0001482138427189283,
+      "loss": 0.2854,
+      "step": 4597
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 4.284493446350098,
+      "learning_rate": 0.00014817663110890597,
+      "loss": 0.3708,
+      "step": 4598
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 3.0818679332733154,
+      "learning_rate": 0.00014813941949888364,
+      "loss": 0.2883,
+      "step": 4599
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 5.769961833953857,
+      "learning_rate": 0.00014810220788886132,
+      "loss": 1.0105,
+      "step": 4600
+    },
+    {
+      "epoch": 1.61,
+      "eval_loss": 0.43010884523391724,
+      "eval_runtime": 51.2971,
+      "eval_samples_per_second": 42.264,
+      "eval_steps_per_second": 10.566,
+      "eval_wer": 0.40140114167099117,
+      "step": 4600
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 3.1520931720733643,
+      "learning_rate": 0.000148064996278839,
+      "loss": 0.2645,
+      "step": 4601
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.3779430389404297,
+      "learning_rate": 0.00014802778466881667,
+      "loss": 0.0886,
+      "step": 4602
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 4.223806381225586,
+      "learning_rate": 0.00014799057305879432,
+      "loss": 0.5118,
+      "step": 4603
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 2.133859157562256,
+      "learning_rate": 0.00014795336144877202,
+      "loss": 0.1417,
+      "step": 4604
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 2.7136905193328857,
+      "learning_rate": 0.00014791614983874967,
+      "loss": 0.789,
+      "step": 4605
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.8167808055877686,
+      "learning_rate": 0.00014787893822872734,
+      "loss": 0.3996,
+      "step": 4606
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.4409061670303345,
+      "learning_rate": 0.00014784172661870502,
+      "loss": 0.3649,
+      "step": 4607
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 2.616767406463623,
+      "learning_rate": 0.0001478045150086827,
+      "loss": 0.4231,
+      "step": 4608
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.7175992727279663,
+      "learning_rate": 0.00014776730339866037,
+      "loss": 0.3993,
+      "step": 4609
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.3235002756118774,
+      "learning_rate": 0.00014773009178863805,
+      "loss": 0.1547,
+      "step": 4610
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 2.10302996635437,
+      "learning_rate": 0.00014769288017861572,
+      "loss": 0.3232,
+      "step": 4611
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.746387004852295,
+      "learning_rate": 0.00014765566856859337,
+      "loss": 0.333,
+      "step": 4612
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 2.292447805404663,
+      "learning_rate": 0.00014761845695857108,
+      "loss": 0.2756,
+      "step": 4613
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.5536998510360718,
+      "learning_rate": 0.00014758124534854875,
+      "loss": 0.2888,
+      "step": 4614
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 2.7711429595947266,
+      "learning_rate": 0.0001475440337385264,
+      "loss": 0.5742,
+      "step": 4615
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.2211471796035767,
+      "learning_rate": 0.00014750682212850408,
+      "loss": 0.1224,
+      "step": 4616
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 2.2878341674804688,
+      "learning_rate": 0.00014746961051848175,
+      "loss": 0.3532,
+      "step": 4617
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.6021337509155273,
+      "learning_rate": 0.00014743239890845943,
+      "loss": 0.2547,
+      "step": 4618
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 3.6017909049987793,
+      "learning_rate": 0.0001473951872984371,
+      "loss": 1.4935,
+      "step": 4619
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 2.1262946128845215,
+      "learning_rate": 0.00014735797568841478,
+      "loss": 0.2398,
+      "step": 4620
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 2.549982786178589,
+      "learning_rate": 0.00014732076407839243,
+      "loss": 0.7301,
+      "step": 4621
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.8897354602813721,
+      "learning_rate": 0.00014728355246837013,
+      "loss": 0.0852,
+      "step": 4622
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 2.3464910984039307,
+      "learning_rate": 0.0001472463408583478,
+      "loss": 0.3429,
+      "step": 4623
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 2.2914788722991943,
+      "learning_rate": 0.00014720912924832545,
+      "loss": 0.2171,
+      "step": 4624
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 5.431281566619873,
+      "learning_rate": 0.00014717191763830313,
+      "loss": 0.4327,
+      "step": 4625
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 2.59971022605896,
+      "learning_rate": 0.0001471347060282808,
+      "loss": 0.2839,
+      "step": 4626
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 2.3822665214538574,
+      "learning_rate": 0.00014709749441825848,
+      "loss": 0.1915,
+      "step": 4627
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.2179664373397827,
+      "learning_rate": 0.00014706028280823616,
+      "loss": 0.1606,
+      "step": 4628
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 3.38651180267334,
+      "learning_rate": 0.00014702307119821383,
+      "loss": 1.2565,
+      "step": 4629
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.9841926097869873,
+      "learning_rate": 0.00014698585958819148,
+      "loss": 0.7275,
+      "step": 4630
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 2.126986503601074,
+      "learning_rate": 0.00014694864797816919,
+      "loss": 0.8143,
+      "step": 4631
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.2944574356079102,
+      "learning_rate": 0.00014691143636814686,
+      "loss": 0.2415,
+      "step": 4632
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.2523384094238281,
+      "learning_rate": 0.0001468742247581245,
+      "loss": 0.3349,
+      "step": 4633
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 2.1736972332000732,
+      "learning_rate": 0.00014683701314810219,
+      "loss": 0.2556,
+      "step": 4634
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 2.1065425872802734,
+      "learning_rate": 0.0001467998015380799,
+      "loss": 0.2447,
+      "step": 4635
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 3.6076436042785645,
+      "learning_rate": 0.00014676258992805754,
+      "loss": 0.7707,
+      "step": 4636
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 2.045930862426758,
+      "learning_rate": 0.0001467253783180352,
+      "loss": 0.3137,
+      "step": 4637
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.0164872407913208,
+      "learning_rate": 0.0001466881667080129,
+      "loss": 0.1478,
+      "step": 4638
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 2.6273579597473145,
+      "learning_rate": 0.00014665095509799056,
+      "loss": 0.2894,
+      "step": 4639
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 2.061847686767578,
+      "learning_rate": 0.00014661374348796824,
+      "loss": 0.109,
+      "step": 4640
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 2.5759425163269043,
+      "learning_rate": 0.00014657653187794592,
+      "loss": 0.1666,
+      "step": 4641
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 2.0799243450164795,
+      "learning_rate": 0.00014653932026792356,
+      "loss": 0.2198,
+      "step": 4642
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 2.562469959259033,
+      "learning_rate": 0.00014650210865790124,
+      "loss": 0.3698,
+      "step": 4643
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 5.489830493927002,
+      "learning_rate": 0.00014646489704787894,
+      "loss": 0.3143,
+      "step": 4644
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 3.388197422027588,
+      "learning_rate": 0.0001464276854378566,
+      "loss": 0.2593,
+      "step": 4645
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 3.1358988285064697,
+      "learning_rate": 0.00014639047382783427,
+      "loss": 0.2429,
+      "step": 4646
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 2.042487859725952,
+      "learning_rate": 0.00014635326221781194,
+      "loss": 0.2667,
+      "step": 4647
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 3.1954731941223145,
+      "learning_rate": 0.00014631605060778962,
+      "loss": 0.2772,
+      "step": 4648
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 6.20961856842041,
+      "learning_rate": 0.0001462788389977673,
+      "loss": 0.8224,
+      "step": 4649
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 2.716186046600342,
+      "learning_rate": 0.00014624162738774497,
+      "loss": 0.3826,
+      "step": 4650
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 2.333148717880249,
+      "learning_rate": 0.00014620441577772265,
+      "loss": 0.2995,
+      "step": 4651
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 9.35218620300293,
+      "learning_rate": 0.0001461672041677003,
+      "loss": 1.6333,
+      "step": 4652
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 2.3607444763183594,
+      "learning_rate": 0.000146129992557678,
+      "loss": 0.213,
+      "step": 4653
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 3.0793917179107666,
+      "learning_rate": 0.00014609278094765565,
+      "loss": 0.1334,
+      "step": 4654
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 3.1427955627441406,
+      "learning_rate": 0.00014605556933763332,
+      "loss": 0.8882,
+      "step": 4655
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 3.179497718811035,
+      "learning_rate": 0.000146018357727611,
+      "loss": 1.0056,
+      "step": 4656
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 2.290648937225342,
+      "learning_rate": 0.00014598114611758867,
+      "loss": 0.439,
+      "step": 4657
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 2.732677698135376,
+      "learning_rate": 0.00014594393450756635,
+      "loss": 0.7399,
+      "step": 4658
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.9614787101745605,
+      "learning_rate": 0.00014590672289754403,
+      "loss": 0.3696,
+      "step": 4659
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.658251404762268,
+      "learning_rate": 0.0001458695112875217,
+      "loss": 0.3531,
+      "step": 4660
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.365149974822998,
+      "learning_rate": 0.00014583229967749935,
+      "loss": 0.1389,
+      "step": 4661
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.6162738800048828,
+      "learning_rate": 0.00014579508806747705,
+      "loss": 0.2063,
+      "step": 4662
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.5521570444107056,
+      "learning_rate": 0.0001457578764574547,
+      "loss": 0.1486,
+      "step": 4663
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.7068825960159302,
+      "learning_rate": 0.00014572066484743238,
+      "loss": 0.4079,
+      "step": 4664
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 3.820570230484009,
+      "learning_rate": 0.00014568345323741005,
+      "loss": 1.0572,
+      "step": 4665
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.7748942375183105,
+      "learning_rate": 0.00014564624162738773,
+      "loss": 0.2345,
+      "step": 4666
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 2.3004446029663086,
+      "learning_rate": 0.0001456090300173654,
+      "loss": 0.2676,
+      "step": 4667
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.6133116483688354,
+      "learning_rate": 0.00014557181840734308,
+      "loss": 0.1579,
+      "step": 4668
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.5437865257263184,
+      "learning_rate": 0.00014553460679732076,
+      "loss": 0.1428,
+      "step": 4669
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 3.339294910430908,
+      "learning_rate": 0.00014549739518729843,
+      "loss": 0.1919,
+      "step": 4670
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 2.862550973892212,
+      "learning_rate": 0.0001454601835772761,
+      "loss": 0.5847,
+      "step": 4671
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 2.6982133388519287,
+      "learning_rate": 0.00014542297196725378,
+      "loss": 0.0781,
+      "step": 4672
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 4.697220325469971,
+      "learning_rate": 0.00014538576035723143,
+      "loss": 0.4647,
+      "step": 4673
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 2.356121778488159,
+      "learning_rate": 0.0001453485487472091,
+      "loss": 0.253,
+      "step": 4674
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 3.0695745944976807,
+      "learning_rate": 0.00014531133713718678,
+      "loss": 0.1711,
+      "step": 4675
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 2.4559316635131836,
+      "learning_rate": 0.00014527412552716446,
+      "loss": 0.1617,
+      "step": 4676
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 5.379946708679199,
+      "learning_rate": 0.00014523691391714214,
+      "loss": 0.486,
+      "step": 4677
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 4.076511859893799,
+      "learning_rate": 0.0001451997023071198,
+      "loss": 1.2421,
+      "step": 4678
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 3.5741493701934814,
+      "learning_rate": 0.0001451624906970975,
+      "loss": 0.1734,
+      "step": 4679
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 2.880136728286743,
+      "learning_rate": 0.00014512527908707516,
+      "loss": 0.8671,
+      "step": 4680
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 2.4380300045013428,
+      "learning_rate": 0.00014508806747705284,
+      "loss": 0.3913,
+      "step": 4681
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.858528733253479,
+      "learning_rate": 0.0001450508558670305,
+      "loss": 0.4003,
+      "step": 4682
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.5311648845672607,
+      "learning_rate": 0.00014501364425700816,
+      "loss": 0.2132,
+      "step": 4683
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.997759222984314,
+      "learning_rate": 0.00014497643264698584,
+      "loss": 0.4408,
+      "step": 4684
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 2.679245710372925,
+      "learning_rate": 0.00014493922103696352,
+      "loss": 0.4171,
+      "step": 4685
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 3.2098543643951416,
+      "learning_rate": 0.0001449020094269412,
+      "loss": 0.3754,
+      "step": 4686
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 3.1209075450897217,
+      "learning_rate": 0.00014486479781691887,
+      "loss": 0.6859,
+      "step": 4687
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.5616499185562134,
+      "learning_rate": 0.00014482758620689654,
+      "loss": 0.1983,
+      "step": 4688
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 2.346381425857544,
+      "learning_rate": 0.00014479037459687422,
+      "loss": 0.4227,
+      "step": 4689
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 2.0079967975616455,
+      "learning_rate": 0.0001447531629868519,
+      "loss": 0.2501,
+      "step": 4690
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.4115216732025146,
+      "learning_rate": 0.00014471595137682954,
+      "loss": 0.1448,
+      "step": 4691
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.5479143857955933,
+      "learning_rate": 0.00014467873976680725,
+      "loss": 0.0889,
+      "step": 4692
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 2.8004372119903564,
+      "learning_rate": 0.00014464152815678492,
+      "loss": 0.2043,
+      "step": 4693
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 3.007761240005493,
+      "learning_rate": 0.00014460431654676257,
+      "loss": 0.0748,
+      "step": 4694
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.9818124771118164,
+      "learning_rate": 0.00014456710493674025,
+      "loss": 0.2978,
+      "step": 4695
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 2.8312606811523438,
+      "learning_rate": 0.00014452989332671792,
+      "loss": 0.2482,
+      "step": 4696
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 2.239459753036499,
+      "learning_rate": 0.0001444926817166956,
+      "loss": 0.1302,
+      "step": 4697
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 2.9169397354125977,
+      "learning_rate": 0.00014445547010667327,
+      "loss": 0.2396,
+      "step": 4698
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.8439327478408813,
+      "learning_rate": 0.00014441825849665095,
+      "loss": 0.1673,
+      "step": 4699
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 4.448665618896484,
+      "learning_rate": 0.0001443810468866286,
+      "loss": 1.1154,
+      "step": 4700
+    },
+    {
+      "epoch": 1.65,
+      "eval_loss": 0.5081397891044617,
+      "eval_runtime": 51.1352,
+      "eval_samples_per_second": 42.397,
+      "eval_steps_per_second": 10.599,
+      "eval_wer": 0.4475869226777374,
+      "step": 4700
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 2.9089102745056152,
+      "learning_rate": 0.0001443438352766063,
+      "loss": 0.3261,
+      "step": 4701
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 2.1453356742858887,
+      "learning_rate": 0.00014430662366658398,
+      "loss": 0.2774,
+      "step": 4702
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 3.1812984943389893,
+      "learning_rate": 0.00014426941205656163,
+      "loss": 0.1525,
+      "step": 4703
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.623590350151062,
+      "learning_rate": 0.0001442322004465393,
+      "loss": 0.1587,
+      "step": 4704
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 2.106051445007324,
+      "learning_rate": 0.00014419498883651698,
+      "loss": 0.3934,
+      "step": 4705
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 2.303579807281494,
+      "learning_rate": 0.00014415777722649465,
+      "loss": 0.5234,
+      "step": 4706
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 2.132972240447998,
+      "learning_rate": 0.00014412056561647233,
+      "loss": 0.4725,
+      "step": 4707
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.8919379711151123,
+      "learning_rate": 0.00014408335400645,
+      "loss": 0.2028,
+      "step": 4708
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.383896827697754,
+      "learning_rate": 0.00014404614239642765,
+      "loss": 0.1585,
+      "step": 4709
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 2.1082210540771484,
+      "learning_rate": 0.00014400893078640536,
+      "loss": 0.5981,
+      "step": 4710
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.662786841392517,
+      "learning_rate": 0.00014397171917638303,
+      "loss": 0.2439,
+      "step": 4711
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 2.6457130908966064,
+      "learning_rate": 0.00014393450756636068,
+      "loss": 0.43,
+      "step": 4712
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 2.411425828933716,
+      "learning_rate": 0.00014389729595633836,
+      "loss": 0.3565,
+      "step": 4713
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 2.6090455055236816,
+      "learning_rate": 0.00014386008434631606,
+      "loss": 0.1975,
+      "step": 4714
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 2.9632768630981445,
+      "learning_rate": 0.0001438228727362937,
+      "loss": 0.202,
+      "step": 4715
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 2.053697109222412,
+      "learning_rate": 0.00014378566112627138,
+      "loss": 0.3983,
+      "step": 4716
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 2.210340738296509,
+      "learning_rate": 0.00014374844951624906,
+      "loss": 0.3449,
+      "step": 4717
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.4516555070877075,
+      "learning_rate": 0.00014371123790622674,
+      "loss": 0.2329,
+      "step": 4718
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.6123175621032715,
+      "learning_rate": 0.0001436740262962044,
+      "loss": 0.2082,
+      "step": 4719
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.6988117694854736,
+      "learning_rate": 0.0001436368146861821,
+      "loss": 0.1245,
+      "step": 4720
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.9669654369354248,
+      "learning_rate": 0.00014359960307615974,
+      "loss": 0.2072,
+      "step": 4721
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 2.498318672180176,
+      "learning_rate": 0.0001435623914661374,
+      "loss": 0.1746,
+      "step": 4722
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 2.309772253036499,
+      "learning_rate": 0.00014352517985611511,
+      "loss": 0.3295,
+      "step": 4723
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.7596651315689087,
+      "learning_rate": 0.00014348796824609276,
+      "loss": 0.1938,
+      "step": 4724
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 2.1776697635650635,
+      "learning_rate": 0.00014345075663607044,
+      "loss": 0.124,
+      "step": 4725
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 3.659687042236328,
+      "learning_rate": 0.00014341354502604811,
+      "loss": 0.4348,
+      "step": 4726
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 2.932208776473999,
+      "learning_rate": 0.0001433763334160258,
+      "loss": 0.3146,
+      "step": 4727
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 2.094736099243164,
+      "learning_rate": 0.00014333912180600347,
+      "loss": 0.1576,
+      "step": 4728
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.049635887145996,
+      "learning_rate": 0.00014330191019598114,
+      "loss": 0.0841,
+      "step": 4729
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.7156835794448853,
+      "learning_rate": 0.0001432646985859588,
+      "loss": 0.7529,
+      "step": 4730
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.7752554416656494,
+      "learning_rate": 0.00014322748697593647,
+      "loss": 0.1068,
+      "step": 4731
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.4473437070846558,
+      "learning_rate": 0.00014319027536591417,
+      "loss": 0.2724,
+      "step": 4732
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.5380414724349976,
+      "learning_rate": 0.00014315306375589182,
+      "loss": 0.317,
+      "step": 4733
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 3.0182528495788574,
+      "learning_rate": 0.0001431158521458695,
+      "loss": 0.4587,
+      "step": 4734
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.3725048303604126,
+      "learning_rate": 0.00014307864053584717,
+      "loss": 0.2012,
+      "step": 4735
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 5.923343658447266,
+      "learning_rate": 0.00014304142892582485,
+      "loss": 0.4233,
+      "step": 4736
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.7124632596969604,
+      "learning_rate": 0.00014300421731580252,
+      "loss": 0.067,
+      "step": 4737
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 3.3575642108917236,
+      "learning_rate": 0.0001429670057057802,
+      "loss": 0.8093,
+      "step": 4738
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 2.0340492725372314,
+      "learning_rate": 0.00014292979409575787,
+      "loss": 0.3322,
+      "step": 4739
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.9786810874938965,
+      "learning_rate": 0.00014289258248573552,
+      "loss": 0.0871,
+      "step": 4740
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.9138987064361572,
+      "learning_rate": 0.00014285537087571322,
+      "loss": 0.3063,
+      "step": 4741
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.8535488843917847,
+      "learning_rate": 0.00014281815926569087,
+      "loss": 0.3178,
+      "step": 4742
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.4353394508361816,
+      "learning_rate": 0.00014278094765566855,
+      "loss": 0.1811,
+      "step": 4743
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 3.0626792907714844,
+      "learning_rate": 0.00014274373604564622,
+      "loss": 0.4848,
+      "step": 4744
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 2.1779870986938477,
+      "learning_rate": 0.0001427065244356239,
+      "loss": 0.2258,
+      "step": 4745
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 7.890713214874268,
+      "learning_rate": 0.00014266931282560158,
+      "loss": 0.9487,
+      "step": 4746
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 3.38159441947937,
+      "learning_rate": 0.00014263210121557925,
+      "loss": 0.2995,
+      "step": 4747
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 2.917301654815674,
+      "learning_rate": 0.00014259488960555693,
+      "loss": 0.5792,
+      "step": 4748
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 3.2175188064575195,
+      "learning_rate": 0.00014255767799553458,
+      "loss": 0.1773,
+      "step": 4749
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 2.1095893383026123,
+      "learning_rate": 0.00014252046638551228,
+      "loss": 0.2296,
+      "step": 4750
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 3.4463748931884766,
+      "learning_rate": 0.00014248325477548993,
+      "loss": 0.0579,
+      "step": 4751
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 3.1098155975341797,
+      "learning_rate": 0.0001424460431654676,
+      "loss": 0.6259,
+      "step": 4752
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.331878662109375,
+      "learning_rate": 0.00014240883155544528,
+      "loss": 0.0832,
+      "step": 4753
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 2.0971672534942627,
+      "learning_rate": 0.00014237161994542296,
+      "loss": 0.1894,
+      "step": 4754
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 9.917618751525879,
+      "learning_rate": 0.00014233440833540063,
+      "loss": 2.6677,
+      "step": 4755
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.685046672821045,
+      "learning_rate": 0.0001422971967253783,
+      "loss": 0.4792,
+      "step": 4756
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 3.0383598804473877,
+      "learning_rate": 0.00014225998511535598,
+      "loss": 0.6298,
+      "step": 4757
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.575229525566101,
+      "learning_rate": 0.00014222277350533366,
+      "loss": 0.1554,
+      "step": 4758
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.6466107368469238,
+      "learning_rate": 0.00014218556189531133,
+      "loss": 0.2602,
+      "step": 4759
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.8354624509811401,
+      "learning_rate": 0.000142148350285289,
+      "loss": 0.258,
+      "step": 4760
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 2.351383924484253,
+      "learning_rate": 0.00014211113867526666,
+      "loss": 0.3148,
+      "step": 4761
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 4.153576374053955,
+      "learning_rate": 0.00014207392706524433,
+      "loss": 0.4399,
+      "step": 4762
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.340376853942871,
+      "learning_rate": 0.000142036715455222,
+      "loss": 0.1725,
+      "step": 4763
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 2.0633182525634766,
+      "learning_rate": 0.00014199950384519969,
+      "loss": 0.3566,
+      "step": 4764
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.320162057876587,
+      "learning_rate": 0.00014196229223517736,
+      "loss": 0.2584,
+      "step": 4765
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.1114542484283447,
+      "learning_rate": 0.00014192508062515504,
+      "loss": 0.0903,
+      "step": 4766
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 3.655611276626587,
+      "learning_rate": 0.00014188786901513271,
+      "loss": 0.4427,
+      "step": 4767
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 5.079207897186279,
+      "learning_rate": 0.0001418506574051104,
+      "loss": 0.547,
+      "step": 4768
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 4.637149810791016,
+      "learning_rate": 0.00014181344579508807,
+      "loss": 0.2594,
+      "step": 4769
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.4649180173873901,
+      "learning_rate": 0.00014177623418506571,
+      "loss": 0.0639,
+      "step": 4770
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 2.595176935195923,
+      "learning_rate": 0.00014173902257504342,
+      "loss": 0.1618,
+      "step": 4771
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 3.4233720302581787,
+      "learning_rate": 0.00014170181096502107,
+      "loss": 0.3967,
+      "step": 4772
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 5.957653045654297,
+      "learning_rate": 0.00014166459935499874,
+      "loss": 1.8509,
+      "step": 4773
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.967976450920105,
+      "learning_rate": 0.00014162738774497642,
+      "loss": 0.1728,
+      "step": 4774
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 3.4824140071868896,
+      "learning_rate": 0.0001415901761349541,
+      "loss": 0.2681,
+      "step": 4775
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.7179447412490845,
+      "learning_rate": 0.00014155296452493177,
+      "loss": 0.1949,
+      "step": 4776
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 2.076540946960449,
+      "learning_rate": 0.00014151575291490944,
+      "loss": 0.248,
+      "step": 4777
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 2.4143893718719482,
+      "learning_rate": 0.00014147854130488712,
+      "loss": 0.3014,
+      "step": 4778
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 4.271050930023193,
+      "learning_rate": 0.00014144132969486477,
+      "loss": 0.3233,
+      "step": 4779
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 4.913974761962891,
+      "learning_rate": 0.00014140411808484247,
+      "loss": 1.046,
+      "step": 4780
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 3.808880090713501,
+      "learning_rate": 0.00014136690647482015,
+      "loss": 0.5579,
+      "step": 4781
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 3.678056478500366,
+      "learning_rate": 0.0001413296948647978,
+      "loss": 0.5987,
+      "step": 4782
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.150814175605774,
+      "learning_rate": 0.00014129248325477547,
+      "loss": 0.1279,
+      "step": 4783
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 2.413187026977539,
+      "learning_rate": 0.00014125527164475315,
+      "loss": 0.4254,
+      "step": 4784
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.675049066543579,
+      "learning_rate": 0.00014121806003473082,
+      "loss": 0.1614,
+      "step": 4785
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 3.6439006328582764,
+      "learning_rate": 0.0001411808484247085,
+      "loss": 0.4455,
+      "step": 4786
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 3.2664308547973633,
+      "learning_rate": 0.00014114363681468618,
+      "loss": 0.757,
+      "step": 4787
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 2.9309678077697754,
+      "learning_rate": 0.00014110642520466382,
+      "loss": 0.3334,
+      "step": 4788
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 2.2318620681762695,
+      "learning_rate": 0.00014106921359464153,
+      "loss": 0.3742,
+      "step": 4789
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 2.364555597305298,
+      "learning_rate": 0.0001410320019846192,
+      "loss": 0.3565,
+      "step": 4790
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 2.487501859664917,
+      "learning_rate": 0.00014099479037459685,
+      "loss": 0.2205,
+      "step": 4791
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.8816657066345215,
+      "learning_rate": 0.00014095757876457453,
+      "loss": 0.1847,
+      "step": 4792
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 2.2951865196228027,
+      "learning_rate": 0.0001409203671545522,
+      "loss": 0.258,
+      "step": 4793
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.8484877347946167,
+      "learning_rate": 0.00014088315554452988,
+      "loss": 0.1441,
+      "step": 4794
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 3.879554033279419,
+      "learning_rate": 0.00014084594393450755,
+      "loss": 0.5913,
+      "step": 4795
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 8.260385513305664,
+      "learning_rate": 0.00014080873232448523,
+      "loss": 1.9975,
+      "step": 4796
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 2.673550605773926,
+      "learning_rate": 0.0001407715207144629,
+      "loss": 0.6173,
+      "step": 4797
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 2.3849501609802246,
+      "learning_rate": 0.00014073430910444058,
+      "loss": 0.2429,
+      "step": 4798
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 4.342643737792969,
+      "learning_rate": 0.00014069709749441826,
+      "loss": 0.6903,
+      "step": 4799
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 2.8125104904174805,
+      "learning_rate": 0.0001406598858843959,
+      "loss": 0.5278,
+      "step": 4800
+    },
+    {
+      "epoch": 1.68,
+      "eval_loss": 0.4198063313961029,
+      "eval_runtime": 51.395,
+      "eval_samples_per_second": 42.183,
+      "eval_steps_per_second": 10.546,
+      "eval_wer": 0.3914547656114859,
+      "step": 4800
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.2338154315948486,
+      "learning_rate": 0.00014062267427437358,
+      "loss": 0.102,
+      "step": 4801
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.0008563995361328,
+      "learning_rate": 0.00014058546266435129,
+      "loss": 0.0983,
+      "step": 4802
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.7836005687713623,
+      "learning_rate": 0.00014054825105432893,
+      "loss": 0.0784,
+      "step": 4803
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 5.8645453453063965,
+      "learning_rate": 0.0001405110394443066,
+      "loss": 0.5595,
+      "step": 4804
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 6.060259819030762,
+      "learning_rate": 0.00014047382783428429,
+      "loss": 1.0248,
+      "step": 4805
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 2.253270387649536,
+      "learning_rate": 0.00014043661622426196,
+      "loss": 0.417,
+      "step": 4806
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 2.2081809043884277,
+      "learning_rate": 0.00014039940461423964,
+      "loss": 0.4441,
+      "step": 4807
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 2.563767910003662,
+      "learning_rate": 0.0001403621930042173,
+      "loss": 0.5687,
+      "step": 4808
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.858834147453308,
+      "learning_rate": 0.00014032498139419496,
+      "loss": 0.491,
+      "step": 4809
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 2.5288004875183105,
+      "learning_rate": 0.00014028776978417264,
+      "loss": 0.4507,
+      "step": 4810
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.2425075769424438,
+      "learning_rate": 0.00014025055817415034,
+      "loss": 0.2789,
+      "step": 4811
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 3.400965929031372,
+      "learning_rate": 0.000140213346564128,
+      "loss": 0.785,
+      "step": 4812
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 2.7308123111724854,
+      "learning_rate": 0.00014017613495410566,
+      "loss": 0.4628,
+      "step": 4813
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 2.1761913299560547,
+      "learning_rate": 0.00014013892334408334,
+      "loss": 0.4261,
+      "step": 4814
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 3.010317325592041,
+      "learning_rate": 0.00014010171173406102,
+      "loss": 0.2918,
+      "step": 4815
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 2.6612460613250732,
+      "learning_rate": 0.0001400645001240387,
+      "loss": 0.4434,
+      "step": 4816
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.3962461948394775,
+      "learning_rate": 0.00014002728851401637,
+      "loss": 0.182,
+      "step": 4817
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.6107089519500732,
+      "learning_rate": 0.00013999007690399404,
+      "loss": 0.14,
+      "step": 4818
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.7715272903442383,
+      "learning_rate": 0.0001399528652939717,
+      "loss": 0.0546,
+      "step": 4819
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.2983365058898926,
+      "learning_rate": 0.0001399156536839494,
+      "loss": 0.0752,
+      "step": 4820
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.6679033041000366,
+      "learning_rate": 0.00013987844207392704,
+      "loss": 0.1966,
+      "step": 4821
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.6639858484268188,
+      "learning_rate": 0.00013984123046390472,
+      "loss": 0.0978,
+      "step": 4822
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.8459872007369995,
+      "learning_rate": 0.0001398040188538824,
+      "loss": 0.0403,
+      "step": 4823
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 8.972551345825195,
+      "learning_rate": 0.00013976680724386007,
+      "loss": 1.5492,
+      "step": 4824
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 3.672152042388916,
+      "learning_rate": 0.00013972959563383775,
+      "loss": 0.5177,
+      "step": 4825
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 4.917130947113037,
+      "learning_rate": 0.00013969238402381542,
+      "loss": 0.5867,
+      "step": 4826
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 3.8166067600250244,
+      "learning_rate": 0.0001396551724137931,
+      "loss": 0.7695,
+      "step": 4827
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001396551724137931,
+      "loss": 0.0208,
+      "step": 4828
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 5.4278788566589355,
+      "learning_rate": 0.00013961796080377075,
+      "loss": 0.9888,
+      "step": 4829
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 2.555227041244507,
+      "learning_rate": 0.00013958074919374845,
+      "loss": 0.4968,
+      "step": 4830
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.814968228340149,
+      "learning_rate": 0.0001395435375837261,
+      "loss": 0.3631,
+      "step": 4831
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.8207460641860962,
+      "learning_rate": 0.00013950632597370377,
+      "loss": 0.4976,
+      "step": 4832
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 2.339345693588257,
+      "learning_rate": 0.00013946911436368145,
+      "loss": 1.0818,
+      "step": 4833
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 2.1169614791870117,
+      "learning_rate": 0.00013943190275365913,
+      "loss": 0.2953,
+      "step": 4834
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.9661637544631958,
+      "learning_rate": 0.0001393946911436368,
+      "loss": 0.4478,
+      "step": 4835
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.695245623588562,
+      "learning_rate": 0.00013935747953361448,
+      "loss": 0.3099,
+      "step": 4836
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.6964011192321777,
+      "learning_rate": 0.00013932026792359215,
+      "loss": 0.3012,
+      "step": 4837
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.6117279529571533,
+      "learning_rate": 0.00013928305631356983,
+      "loss": 0.0769,
+      "step": 4838
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 4.413363933563232,
+      "learning_rate": 0.0001392458447035475,
+      "loss": 0.5109,
+      "step": 4839
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 5.733808994293213,
+      "learning_rate": 0.00013920863309352518,
+      "loss": 1.3301,
+      "step": 4840
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 2.3918161392211914,
+      "learning_rate": 0.00013917142148350283,
+      "loss": 0.716,
+      "step": 4841
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.3467110395431519,
+      "learning_rate": 0.0001391342098734805,
+      "loss": 0.186,
+      "step": 4842
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.767811894416809,
+      "learning_rate": 0.00013909699826345818,
+      "loss": 0.2146,
+      "step": 4843
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 2.754969835281372,
+      "learning_rate": 0.00013905978665343586,
+      "loss": 0.4884,
+      "step": 4844
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.2362042665481567,
+      "learning_rate": 0.00013902257504341353,
+      "loss": 0.2557,
+      "step": 4845
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.3661056756973267,
+      "learning_rate": 0.0001389853634333912,
+      "loss": 0.1672,
+      "step": 4846
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.7961978912353516,
+      "learning_rate": 0.00013894815182336888,
+      "loss": 0.0539,
+      "step": 4847
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 2.0398333072662354,
+      "learning_rate": 0.00013891094021334656,
+      "loss": 0.2883,
+      "step": 4848
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.8911408185958862,
+      "learning_rate": 0.00013887372860332424,
+      "loss": 0.2,
+      "step": 4849
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 3.2513821125030518,
+      "learning_rate": 0.00013883651699330188,
+      "loss": 0.5697,
+      "step": 4850
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.9013601541519165,
+      "learning_rate": 0.00013879930538327956,
+      "loss": 0.1448,
+      "step": 4851
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 3.6349408626556396,
+      "learning_rate": 0.00013876209377325724,
+      "loss": 0.4995,
+      "step": 4852
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.0852819681167603,
+      "learning_rate": 0.0001387248821632349,
+      "loss": 0.0309,
+      "step": 4853
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001387248821632349,
+      "loss": 0.0687,
+      "step": 4854
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 2.3139288425445557,
+      "learning_rate": 0.0001386876705532126,
+      "loss": 0.572,
+      "step": 4855
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 2.169691324234009,
+      "learning_rate": 0.00013865045894319026,
+      "loss": 0.265,
+      "step": 4856
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 2.2995455265045166,
+      "learning_rate": 0.00013861324733316794,
+      "loss": 0.4024,
+      "step": 4857
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.8771644830703735,
+      "learning_rate": 0.00013857603572314562,
+      "loss": 0.2272,
+      "step": 4858
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.6588895320892334,
+      "learning_rate": 0.0001385388241131233,
+      "loss": 0.2656,
+      "step": 4859
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 2.3973159790039062,
+      "learning_rate": 0.00013850161250310094,
+      "loss": 0.7651,
+      "step": 4860
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.8839689493179321,
+      "learning_rate": 0.00013846440089307864,
+      "loss": 0.4598,
+      "step": 4861
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.9465550184249878,
+      "learning_rate": 0.00013842718928305632,
+      "loss": 0.235,
+      "step": 4862
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 2.5686044692993164,
+      "learning_rate": 0.00013838997767303397,
+      "loss": 0.7046,
+      "step": 4863
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 2.570693254470825,
+      "learning_rate": 0.00013835276606301164,
+      "loss": 0.1989,
+      "step": 4864
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 2.393869638442993,
+      "learning_rate": 0.00013831555445298932,
+      "loss": 0.3419,
+      "step": 4865
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 2.3206074237823486,
+      "learning_rate": 0.000138278342842967,
+      "loss": 0.231,
+      "step": 4866
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.7298521995544434,
+      "learning_rate": 0.00013824113123294467,
+      "loss": 0.2686,
+      "step": 4867
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.511710286140442,
+      "learning_rate": 0.00013820391962292235,
+      "loss": 0.2347,
+      "step": 4868
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 2.156994104385376,
+      "learning_rate": 0.0001381667080129,
+      "loss": 0.2402,
+      "step": 4869
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.5252554416656494,
+      "learning_rate": 0.0001381294964028777,
+      "loss": 0.1506,
+      "step": 4870
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 3.855454683303833,
+      "learning_rate": 0.00013809228479285537,
+      "loss": 0.3318,
+      "step": 4871
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 3.8235645294189453,
+      "learning_rate": 0.00013805507318283302,
+      "loss": 0.6935,
+      "step": 4872
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 3.7631542682647705,
+      "learning_rate": 0.0001380178615728107,
+      "loss": 0.6565,
+      "step": 4873
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 4.347606658935547,
+      "learning_rate": 0.00013798064996278837,
+      "loss": 0.6065,
+      "step": 4874
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.8318398594856262,
+      "learning_rate": 0.00013794343835276605,
+      "loss": 0.0503,
+      "step": 4875
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 3.258185386657715,
+      "learning_rate": 0.00013790622674274373,
+      "loss": 0.6948,
+      "step": 4876
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.6956919431686401,
+      "learning_rate": 0.0001378690151327214,
+      "loss": 0.1668,
+      "step": 4877
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 5.9375224113464355,
+      "learning_rate": 0.00013783180352269908,
+      "loss": 0.6585,
+      "step": 4878
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 4.867272853851318,
+      "learning_rate": 0.00013779459191267675,
+      "loss": 0.2866,
+      "step": 4879
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.3362858295440674,
+      "learning_rate": 0.00013775738030265443,
+      "loss": 0.246,
+      "step": 4880
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 2.0603818893432617,
+      "learning_rate": 0.00013772016869263208,
+      "loss": 0.4112,
+      "step": 4881
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 2.5215094089508057,
+      "learning_rate": 0.00013768295708260975,
+      "loss": 0.47,
+      "step": 4882
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 2.3979105949401855,
+      "learning_rate": 0.00013764574547258746,
+      "loss": 0.5573,
+      "step": 4883
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 4.085318088531494,
+      "learning_rate": 0.0001376085338625651,
+      "loss": 0.4865,
+      "step": 4884
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.9340821504592896,
+      "learning_rate": 0.00013757132225254278,
+      "loss": 0.222,
+      "step": 4885
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 26.611042022705078,
+      "learning_rate": 0.00013753411064252046,
+      "loss": 4.7412,
+      "step": 4886
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 2.0535519123077393,
+      "learning_rate": 0.00013749689903249813,
+      "loss": 0.2508,
+      "step": 4887
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 3.884704828262329,
+      "learning_rate": 0.0001374596874224758,
+      "loss": 0.8941,
+      "step": 4888
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.4614795446395874,
+      "learning_rate": 0.00013742247581245348,
+      "loss": 0.2868,
+      "step": 4889
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.9116908311843872,
+      "learning_rate": 0.00013738526420243113,
+      "loss": 0.2023,
+      "step": 4890
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.9487919807434082,
+      "learning_rate": 0.0001373480525924088,
+      "loss": 0.0826,
+      "step": 4891
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.6531901359558105,
+      "learning_rate": 0.0001373108409823865,
+      "loss": 0.2198,
+      "step": 4892
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.2997630834579468,
+      "learning_rate": 0.00013727362937236416,
+      "loss": 0.1561,
+      "step": 4893
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 4.334744453430176,
+      "learning_rate": 0.00013723641776234184,
+      "loss": 0.5747,
+      "step": 4894
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 6.257345676422119,
+      "learning_rate": 0.0001371992061523195,
+      "loss": 1.8452,
+      "step": 4895
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 6.1517133712768555,
+      "learning_rate": 0.0001371619945422972,
+      "loss": 1.6441,
+      "step": 4896
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 2.8486979007720947,
+      "learning_rate": 0.00013712478293227486,
+      "loss": 0.343,
+      "step": 4897
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 2.724308490753174,
+      "learning_rate": 0.00013708757132225254,
+      "loss": 0.7712,
+      "step": 4898
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.3378300666809082,
+      "learning_rate": 0.00013705035971223021,
+      "loss": 0.0844,
+      "step": 4899
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 4.833279609680176,
+      "learning_rate": 0.00013701314810220786,
+      "loss": 0.1357,
+      "step": 4900
+    },
+    {
+      "epoch": 1.72,
+      "eval_loss": 0.3683335483074188,
+      "eval_runtime": 51.5895,
+      "eval_samples_per_second": 42.024,
+      "eval_steps_per_second": 10.506,
+      "eval_wer": 0.3474312402698495,
+      "step": 4900
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 3.9212887287139893,
+      "learning_rate": 0.00013697593649218557,
+      "loss": 0.5343,
+      "step": 4901
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 3.2143747806549072,
+      "learning_rate": 0.00013693872488216321,
+      "loss": 0.4345,
+      "step": 4902
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 7.424620628356934,
+      "learning_rate": 0.0001369015132721409,
+      "loss": 0.5639,
+      "step": 4903
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 2.1714751720428467,
+      "learning_rate": 0.00013686430166211857,
+      "loss": 0.2382,
+      "step": 4904
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 2.139472007751465,
+      "learning_rate": 0.00013682709005209624,
+      "loss": 0.7981,
+      "step": 4905
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.1302719116210938,
+      "learning_rate": 0.00013678987844207392,
+      "loss": 0.242,
+      "step": 4906
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.639523983001709,
+      "learning_rate": 0.0001367526668320516,
+      "loss": 0.5281,
+      "step": 4907
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.3213592767715454,
+      "learning_rate": 0.00013671545522202927,
+      "loss": 0.2546,
+      "step": 4908
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.2504050731658936,
+      "learning_rate": 0.00013667824361200692,
+      "loss": 0.1545,
+      "step": 4909
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.2312544584274292,
+      "learning_rate": 0.00013664103200198462,
+      "loss": 0.1631,
+      "step": 4910
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.3905742168426514,
+      "learning_rate": 0.00013660382039196227,
+      "loss": 0.1149,
+      "step": 4911
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 2.0212652683258057,
+      "learning_rate": 0.00013656660878193995,
+      "loss": 0.4504,
+      "step": 4912
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.0436809062957764,
+      "learning_rate": 0.00013652939717191762,
+      "loss": 0.1652,
+      "step": 4913
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 3.0124869346618652,
+      "learning_rate": 0.0001364921855618953,
+      "loss": 0.354,
+      "step": 4914
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.9092917442321777,
+      "learning_rate": 0.00013645497395187297,
+      "loss": 0.2767,
+      "step": 4915
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.9692587852478027,
+      "learning_rate": 0.00013641776234185065,
+      "loss": 0.3325,
+      "step": 4916
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 2.2949063777923584,
+      "learning_rate": 0.00013638055073182832,
+      "loss": 0.3535,
+      "step": 4917
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.9984784722328186,
+      "learning_rate": 0.00013634333912180597,
+      "loss": 0.1386,
+      "step": 4918
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 2.6047825813293457,
+      "learning_rate": 0.00013630612751178368,
+      "loss": 0.2291,
+      "step": 4919
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 2.1782517433166504,
+      "learning_rate": 0.00013626891590176135,
+      "loss": 0.261,
+      "step": 4920
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.2911269664764404,
+      "learning_rate": 0.000136231704291739,
+      "loss": 0.0941,
+      "step": 4921
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 2.6118998527526855,
+      "learning_rate": 0.00013619449268171668,
+      "loss": 0.2503,
+      "step": 4922
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.126331090927124,
+      "learning_rate": 0.00013615728107169435,
+      "loss": 0.0635,
+      "step": 4923
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 4.782196044921875,
+      "learning_rate": 0.00013612006946167203,
+      "loss": 0.4418,
+      "step": 4924
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.8012721538543701,
+      "learning_rate": 0.0001360828578516497,
+      "loss": 0.1878,
+      "step": 4925
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 3.5132253170013428,
+      "learning_rate": 0.00013604564624162738,
+      "loss": 0.1484,
+      "step": 4926
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 3.3157382011413574,
+      "learning_rate": 0.00013600843463160506,
+      "loss": 0.2761,
+      "step": 4927
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.4232193231582642,
+      "learning_rate": 0.00013597122302158273,
+      "loss": 0.0699,
+      "step": 4928
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.8785415887832642,
+      "learning_rate": 0.0001359340114115604,
+      "loss": 0.0976,
+      "step": 4929
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 2.8369486331939697,
+      "learning_rate": 0.00013589679980153806,
+      "loss": 0.5884,
+      "step": 4930
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.4012984037399292,
+      "learning_rate": 0.00013585958819151573,
+      "loss": 0.3877,
+      "step": 4931
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.8025954961776733,
+      "learning_rate": 0.0001358223765814934,
+      "loss": 0.1753,
+      "step": 4932
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.4512267112731934,
+      "learning_rate": 0.00013578516497147108,
+      "loss": 0.379,
+      "step": 4933
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 2.033318281173706,
+      "learning_rate": 0.00013574795336144876,
+      "loss": 0.2617,
+      "step": 4934
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 3.8261144161224365,
+      "learning_rate": 0.00013571074175142643,
+      "loss": 0.4695,
+      "step": 4935
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.4579124450683594,
+      "learning_rate": 0.0001356735301414041,
+      "loss": 0.2473,
+      "step": 4936
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.6672676801681519,
+      "learning_rate": 0.00013563631853138179,
+      "loss": 0.2851,
+      "step": 4937
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.9775683283805847,
+      "learning_rate": 0.00013559910692135946,
+      "loss": 0.0493,
+      "step": 4938
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 3.237879514694214,
+      "learning_rate": 0.0001355618953113371,
+      "loss": 0.5164,
+      "step": 4939
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 3.030714273452759,
+      "learning_rate": 0.0001355246837013148,
+      "loss": 0.1808,
+      "step": 4940
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 2.291748285293579,
+      "learning_rate": 0.0001354874720912925,
+      "loss": 0.1526,
+      "step": 4941
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 2.577086925506592,
+      "learning_rate": 0.00013545026048127014,
+      "loss": 0.1565,
+      "step": 4942
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 3.0467169284820557,
+      "learning_rate": 0.00013541304887124781,
+      "loss": 0.3401,
+      "step": 4943
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 2.7685546875,
+      "learning_rate": 0.0001353758372612255,
+      "loss": 0.1665,
+      "step": 4944
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 2.6648311614990234,
+      "learning_rate": 0.00013533862565120317,
+      "loss": 0.1115,
+      "step": 4945
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 4.279283046722412,
+      "learning_rate": 0.00013530141404118084,
+      "loss": 0.59,
+      "step": 4946
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 3.774441719055176,
+      "learning_rate": 0.00013526420243115852,
+      "loss": 1.4274,
+      "step": 4947
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 3.0308432579040527,
+      "learning_rate": 0.00013522699082113617,
+      "loss": 0.2772,
+      "step": 4948
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 2.0633444786071777,
+      "learning_rate": 0.00013518977921111387,
+      "loss": 0.3336,
+      "step": 4949
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 3.2152934074401855,
+      "learning_rate": 0.00013515256760109154,
+      "loss": 0.3653,
+      "step": 4950
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 3.7646780014038086,
+      "learning_rate": 0.0001351153559910692,
+      "loss": 0.3,
+      "step": 4951
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.37087032198905945,
+      "learning_rate": 0.00013507814438104687,
+      "loss": 0.0195,
+      "step": 4952
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.66860830783844,
+      "learning_rate": 0.00013504093277102454,
+      "loss": 0.1063,
+      "step": 4953
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.9260728359222412,
+      "learning_rate": 0.00013500372116100222,
+      "loss": 0.1283,
+      "step": 4954
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.4697273969650269,
+      "learning_rate": 0.0001349665095509799,
+      "loss": 0.6385,
+      "step": 4955
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.3390899896621704,
+      "learning_rate": 0.00013492929794095757,
+      "loss": 0.3823,
+      "step": 4956
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.3227888345718384,
+      "learning_rate": 0.00013489208633093522,
+      "loss": 0.3421,
+      "step": 4957
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 2.2112112045288086,
+      "learning_rate": 0.00013485487472091292,
+      "loss": 0.4267,
+      "step": 4958
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.0806066989898682,
+      "learning_rate": 0.0001348176631108906,
+      "loss": 0.1819,
+      "step": 4959
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.1106966733932495,
+      "learning_rate": 0.00013478045150086825,
+      "loss": 0.0838,
+      "step": 4960
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.180200219154358,
+      "learning_rate": 0.00013474323989084592,
+      "loss": 0.1393,
+      "step": 4961
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.237693190574646,
+      "learning_rate": 0.0001347060282808236,
+      "loss": 0.172,
+      "step": 4962
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 2.4684200286865234,
+      "learning_rate": 0.00013466881667080128,
+      "loss": 0.3471,
+      "step": 4963
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.569759488105774,
+      "learning_rate": 0.00013463160506077895,
+      "loss": 0.246,
+      "step": 4964
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 2.071803092956543,
+      "learning_rate": 0.00013459439345075663,
+      "loss": 0.2071,
+      "step": 4965
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 2.41078519821167,
+      "learning_rate": 0.0001345571818407343,
+      "loss": 0.7541,
+      "step": 4966
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.8611916303634644,
+      "learning_rate": 0.00013451997023071198,
+      "loss": 0.5563,
+      "step": 4967
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.8303391933441162,
+      "learning_rate": 0.00013448275862068965,
+      "loss": 0.2874,
+      "step": 4968
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.8616209030151367,
+      "learning_rate": 0.0001344455470106673,
+      "loss": 0.1496,
+      "step": 4969
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 2.0775692462921143,
+      "learning_rate": 0.00013440833540064498,
+      "loss": 0.3153,
+      "step": 4970
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 2.2036242485046387,
+      "learning_rate": 0.00013437112379062268,
+      "loss": 0.3611,
+      "step": 4971
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 2.958390235900879,
+      "learning_rate": 0.00013433391218060033,
+      "loss": 0.3149,
+      "step": 4972
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.2696552276611328,
+      "learning_rate": 0.000134296700570578,
+      "loss": 0.0836,
+      "step": 4973
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.8660387992858887,
+      "learning_rate": 0.00013425948896055568,
+      "loss": 0.0415,
+      "step": 4974
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 3.5192763805389404,
+      "learning_rate": 0.00013422227735053336,
+      "loss": 0.3977,
+      "step": 4975
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.378849744796753,
+      "learning_rate": 0.00013418506574051103,
+      "loss": 0.0728,
+      "step": 4976
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 3.9449732303619385,
+      "learning_rate": 0.0001341478541304887,
+      "loss": 0.221,
+      "step": 4977
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 5.472590446472168,
+      "learning_rate": 0.00013411064252046636,
+      "loss": 1.7522,
+      "step": 4978
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": NaN,
+      "learning_rate": 0.00013411064252046636,
+      "loss": 0.2462,
+      "step": 4979
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 4.078400611877441,
+      "learning_rate": 0.00013407343091044403,
+      "loss": 0.7826,
+      "step": 4980
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.8537403345108032,
+      "learning_rate": 0.00013403621930042174,
+      "loss": 0.4351,
+      "step": 4981
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.9395902156829834,
+      "learning_rate": 0.00013399900769039939,
+      "loss": 0.201,
+      "step": 4982
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.612997055053711,
+      "learning_rate": 0.00013396179608037706,
+      "loss": 0.2431,
+      "step": 4983
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.5586113929748535,
+      "learning_rate": 0.00013392458447035474,
+      "loss": 0.2738,
+      "step": 4984
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.7553973197937012,
+      "learning_rate": 0.0001338873728603324,
+      "loss": 0.3034,
+      "step": 4985
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.0098979473114014,
+      "learning_rate": 0.0001338501612503101,
+      "loss": 0.0768,
+      "step": 4986
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 3.2401552200317383,
+      "learning_rate": 0.00013381294964028776,
+      "loss": 1.3496,
+      "step": 4987
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.5994211435317993,
+      "learning_rate": 0.00013377573803026544,
+      "loss": 0.1578,
+      "step": 4988
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 2.8812813758850098,
+      "learning_rate": 0.0001337385264202431,
+      "loss": 0.6019,
+      "step": 4989
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.9770814180374146,
+      "learning_rate": 0.0001337013148102208,
+      "loss": 0.2927,
+      "step": 4990
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 3.3492937088012695,
+      "learning_rate": 0.00013366410320019844,
+      "loss": 0.8254,
+      "step": 4991
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.7024646997451782,
+      "learning_rate": 0.00013362689159017612,
+      "loss": 0.0549,
+      "step": 4992
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.3194102048873901,
+      "learning_rate": 0.0001335896799801538,
+      "loss": 0.094,
+      "step": 4993
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.7211250066757202,
+      "learning_rate": 0.00013355246837013147,
+      "loss": 0.1947,
+      "step": 4994
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 2.0496625900268555,
+      "learning_rate": 0.00013351525676010914,
+      "loss": 0.2378,
+      "step": 4995
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.6302729845046997,
+      "learning_rate": 0.00013347804515008682,
+      "loss": 0.2008,
+      "step": 4996
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 4.008142948150635,
+      "learning_rate": 0.0001334408335400645,
+      "loss": 0.5542,
+      "step": 4997
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 2.101016044616699,
+      "learning_rate": 0.00013340362193004214,
+      "loss": 0.1599,
+      "step": 4998
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.9622274041175842,
+      "learning_rate": 0.00013336641032001985,
+      "loss": 0.0923,
+      "step": 4999
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 4.595578670501709,
+      "learning_rate": 0.0001333291987099975,
+      "loss": 0.4642,
+      "step": 5000
+    },
+    {
+      "epoch": 1.75,
+      "eval_loss": 0.3618814945220947,
+      "eval_runtime": 51.7136,
+      "eval_samples_per_second": 41.923,
+      "eval_steps_per_second": 10.481,
+      "eval_wer": 0.3420688462203771,
+      "step": 5000
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 2.2924017906188965,
+      "learning_rate": 0.00013329198709997517,
+      "loss": 0.3131,
+      "step": 5001
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.4063929319381714,
+      "learning_rate": 0.00013325477548995285,
+      "loss": 0.1032,
+      "step": 5002
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.765910267829895,
+      "learning_rate": 0.00013321756387993052,
+      "loss": 0.2847,
+      "step": 5003
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 2.635676383972168,
+      "learning_rate": 0.0001331803522699082,
+      "loss": 0.3925,
+      "step": 5004
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 2.2191758155822754,
+      "learning_rate": 0.00013314314065988587,
+      "loss": 0.5075,
+      "step": 5005
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.3144760131835938,
+      "learning_rate": 0.00013310592904986355,
+      "loss": 0.1606,
+      "step": 5006
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 2.2465732097625732,
+      "learning_rate": 0.00013306871743984123,
+      "loss": 0.3372,
+      "step": 5007
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.2884074449539185,
+      "learning_rate": 0.0001330315058298189,
+      "loss": 0.1578,
+      "step": 5008
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.7158373594284058,
+      "learning_rate": 0.00013299429421979658,
+      "loss": 0.2009,
+      "step": 5009
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.9673197269439697,
+      "learning_rate": 0.00013295708260977423,
+      "loss": 0.2124,
+      "step": 5010
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 2.0683186054229736,
+      "learning_rate": 0.0001329198709997519,
+      "loss": 0.445,
+      "step": 5011
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 2.1322107315063477,
+      "learning_rate": 0.00013288265938972958,
+      "loss": 0.1698,
+      "step": 5012
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.6621609926223755,
+      "learning_rate": 0.00013284544777970725,
+      "loss": 0.1697,
+      "step": 5013
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 2.388695240020752,
+      "learning_rate": 0.00013280823616968493,
+      "loss": 0.291,
+      "step": 5014
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.7287172079086304,
+      "learning_rate": 0.0001327710245596626,
+      "loss": 0.2866,
+      "step": 5015
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 2.0600686073303223,
+      "learning_rate": 0.00013273381294964028,
+      "loss": 0.2617,
+      "step": 5016
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 3.1544766426086426,
+      "learning_rate": 0.00013269660133961796,
+      "loss": 0.3467,
+      "step": 5017
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 3.4372010231018066,
+      "learning_rate": 0.00013265938972959563,
+      "loss": 0.3877,
+      "step": 5018
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 2.706695079803467,
+      "learning_rate": 0.00013262217811957328,
+      "loss": 0.3513,
+      "step": 5019
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 4.3857927322387695,
+      "learning_rate": 0.00013258496650955096,
+      "loss": 1.5684,
+      "step": 5020
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 2.2888591289520264,
+      "learning_rate": 0.00013254775489952863,
+      "loss": 0.2338,
+      "step": 5021
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 2.9611387252807617,
+      "learning_rate": 0.0001325105432895063,
+      "loss": 0.2578,
+      "step": 5022
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 2.1191227436065674,
+      "learning_rate": 0.00013247333167948398,
+      "loss": 0.1561,
+      "step": 5023
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 2.088348865509033,
+      "learning_rate": 0.00013243612006946166,
+      "loss": 0.1646,
+      "step": 5024
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.8998678922653198,
+      "learning_rate": 0.00013239890845943934,
+      "loss": 0.2291,
+      "step": 5025
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 2.2618954181671143,
+      "learning_rate": 0.000132361696849417,
+      "loss": 0.0624,
+      "step": 5026
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 6.207477569580078,
+      "learning_rate": 0.0001323244852393947,
+      "loss": 0.563,
+      "step": 5027
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 21.43695640563965,
+      "learning_rate": 0.00013228727362937234,
+      "loss": 2.0059,
+      "step": 5028
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": NaN,
+      "learning_rate": 0.00013228727362937234,
+      "loss": 1.7438,
+      "step": 5029
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 5383.18994140625,
+      "learning_rate": 0.00013225006201935004,
+      "loss": 16.872,
+      "step": 5030
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 23.968223571777344,
+      "learning_rate": 0.00013221285040932772,
+      "loss": 4.1902,
+      "step": 5031
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 18.13584327697754,
+      "learning_rate": 0.00013217563879930536,
+      "loss": 3.597,
+      "step": 5032
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 27.954679489135742,
+      "learning_rate": 0.00013213842718928304,
+      "loss": 3.8674,
+      "step": 5033
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 12.132146835327148,
+      "learning_rate": 0.00013210121557926072,
+      "loss": 3.337,
+      "step": 5034
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 9.336516380310059,
+      "learning_rate": 0.0001320640039692384,
+      "loss": 3.0708,
+      "step": 5035
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 14.99378776550293,
+      "learning_rate": 0.00013202679235921607,
+      "loss": 3.4457,
+      "step": 5036
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 16.54006004333496,
+      "learning_rate": 0.00013198958074919374,
+      "loss": 3.4698,
+      "step": 5037
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 5.521266460418701,
+      "learning_rate": 0.0001319523691391714,
+      "loss": 3.0527,
+      "step": 5038
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 13.792803764343262,
+      "learning_rate": 0.0001319151575291491,
+      "loss": 3.1825,
+      "step": 5039
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 10.1255464553833,
+      "learning_rate": 0.00013187794591912677,
+      "loss": 3.3301,
+      "step": 5040
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 6.728694915771484,
+      "learning_rate": 0.00013184073430910442,
+      "loss": 3.4529,
+      "step": 5041
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 13.570353507995605,
+      "learning_rate": 0.0001318035226990821,
+      "loss": 3.1857,
+      "step": 5042
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 9.944263458251953,
+      "learning_rate": 0.00013176631108905977,
+      "loss": 3.0247,
+      "step": 5043
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 7.628637790679932,
+      "learning_rate": 0.00013172909947903745,
+      "loss": 3.0557,
+      "step": 5044
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 8.50871753692627,
+      "learning_rate": 0.00013169188786901512,
+      "loss": 2.9197,
+      "step": 5045
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 3.350416421890259,
+      "learning_rate": 0.0001316546762589928,
+      "loss": 2.7846,
+      "step": 5046
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 8.267583847045898,
+      "learning_rate": 0.00013161746464897047,
+      "loss": 2.9316,
+      "step": 5047
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 10.566850662231445,
+      "learning_rate": 0.00013158025303894815,
+      "loss": 3.5342,
+      "step": 5048
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 7.871520042419434,
+      "learning_rate": 0.00013154304142892583,
+      "loss": 3.2533,
+      "step": 5049
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 14.895259857177734,
+      "learning_rate": 0.00013150582981890347,
+      "loss": 3.3095,
+      "step": 5050
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 6.724153995513916,
+      "learning_rate": 0.00013146861820888115,
+      "loss": 2.804,
+      "step": 5051
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 10.182531356811523,
+      "learning_rate": 0.00013143140659885885,
+      "loss": 2.9485,
+      "step": 5052
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 12.471945762634277,
+      "learning_rate": 0.0001313941949888365,
+      "loss": 3.0598,
+      "step": 5053
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 4.7943267822265625,
+      "learning_rate": 0.00013135698337881418,
+      "loss": 2.6582,
+      "step": 5054
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 3.8134336471557617,
+      "learning_rate": 0.00013131977176879185,
+      "loss": 3.2571,
+      "step": 5055
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 2.2540090084075928,
+      "learning_rate": 0.00013128256015876953,
+      "loss": 3.1262,
+      "step": 5056
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 3.4563486576080322,
+      "learning_rate": 0.0001312453485487472,
+      "loss": 2.8854,
+      "step": 5057
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 4.2413249015808105,
+      "learning_rate": 0.00013120813693872488,
+      "loss": 2.9584,
+      "step": 5058
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 12.829059600830078,
+      "learning_rate": 0.00013117092532870253,
+      "loss": 3.1434,
+      "step": 5059
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 2.502635955810547,
+      "learning_rate": 0.0001311337137186802,
+      "loss": 3.0336,
+      "step": 5060
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 2.319552421569824,
+      "learning_rate": 0.0001310965021086579,
+      "loss": 2.9348,
+      "step": 5061
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 6.4483866691589355,
+      "learning_rate": 0.00013105929049863556,
+      "loss": 2.7653,
+      "step": 5062
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 2.6617424488067627,
+      "learning_rate": 0.00013102207888861323,
+      "loss": 2.9234,
+      "step": 5063
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 4.615228652954102,
+      "learning_rate": 0.0001309848672785909,
+      "loss": 2.7653,
+      "step": 5064
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 6.482029438018799,
+      "learning_rate": 0.00013094765566856858,
+      "loss": 2.7954,
+      "step": 5065
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 7.622623920440674,
+      "learning_rate": 0.00013091044405854626,
+      "loss": 3.0746,
+      "step": 5066
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 8.83546257019043,
+      "learning_rate": 0.00013087323244852394,
+      "loss": 2.9216,
+      "step": 5067
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 9.417671203613281,
+      "learning_rate": 0.0001308360208385016,
+      "loss": 2.9592,
+      "step": 5068
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 3.7668020725250244,
+      "learning_rate": 0.00013079880922847926,
+      "loss": 2.7894,
+      "step": 5069
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 8.39608097076416,
+      "learning_rate": 0.00013076159761845696,
+      "loss": 2.7767,
+      "step": 5070
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 3.686868190765381,
+      "learning_rate": 0.0001307243860084346,
+      "loss": 3.019,
+      "step": 5071
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 10.237363815307617,
+      "learning_rate": 0.0001306871743984123,
+      "loss": 3.0566,
+      "step": 5072
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 4.589514255523682,
+      "learning_rate": 0.00013064996278838996,
+      "loss": 3.2709,
+      "step": 5073
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 4.9313225746154785,
+      "learning_rate": 0.00013061275117836764,
+      "loss": 2.6611,
+      "step": 5074
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 4.68777322769165,
+      "learning_rate": 0.00013057553956834531,
+      "loss": 2.9001,
+      "step": 5075
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 3.2555599212646484,
+      "learning_rate": 0.000130538327958323,
+      "loss": 2.8576,
+      "step": 5076
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 3.0508415699005127,
+      "learning_rate": 0.00013050111634830067,
+      "loss": 2.7034,
+      "step": 5077
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 3.908602237701416,
+      "learning_rate": 0.00013046390473827832,
+      "loss": 2.8313,
+      "step": 5078
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 2.733027696609497,
+      "learning_rate": 0.00013042669312825602,
+      "loss": 2.3422,
+      "step": 5079
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 4.607684135437012,
+      "learning_rate": 0.00013038948151823367,
+      "loss": 3.3491,
+      "step": 5080
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 3.3591644763946533,
+      "learning_rate": 0.00013035226990821134,
+      "loss": 2.9754,
+      "step": 5081
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 3.9644618034362793,
+      "learning_rate": 0.00013031505829818902,
+      "loss": 3.053,
+      "step": 5082
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 8.168998718261719,
+      "learning_rate": 0.0001302778466881667,
+      "loss": 2.9517,
+      "step": 5083
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 3.216783046722412,
+      "learning_rate": 0.00013024063507814437,
+      "loss": 3.0665,
+      "step": 5084
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 2.6006112098693848,
+      "learning_rate": 0.00013020342346812205,
+      "loss": 2.8623,
+      "step": 5085
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 2.3029751777648926,
+      "learning_rate": 0.00013016621185809972,
+      "loss": 2.8781,
+      "step": 5086
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 3.3542487621307373,
+      "learning_rate": 0.00013012900024807737,
+      "loss": 2.7165,
+      "step": 5087
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 3.487226724624634,
+      "learning_rate": 0.00013009178863805507,
+      "loss": 2.8395,
+      "step": 5088
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 10.391279220581055,
+      "learning_rate": 0.00013005457702803275,
+      "loss": 3.1138,
+      "step": 5089
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 9.242776870727539,
+      "learning_rate": 0.0001300173654180104,
+      "loss": 3.0096,
+      "step": 5090
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 4.211190223693848,
+      "learning_rate": 0.00012998015380798807,
+      "loss": 2.8559,
+      "step": 5091
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 2.6914520263671875,
+      "learning_rate": 0.00012994294219796575,
+      "loss": 2.9511,
+      "step": 5092
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 4.44136381149292,
+      "learning_rate": 0.00012990573058794342,
+      "loss": 2.7139,
+      "step": 5093
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 3.7667293548583984,
+      "learning_rate": 0.0001298685189779211,
+      "loss": 2.7297,
+      "step": 5094
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 3.589275360107422,
+      "learning_rate": 0.00012983130736789878,
+      "loss": 2.8284,
+      "step": 5095
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 4.216124534606934,
+      "learning_rate": 0.00012979409575787645,
+      "loss": 2.9066,
+      "step": 5096
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 10.489272117614746,
+      "learning_rate": 0.00012975688414785413,
+      "loss": 2.9559,
+      "step": 5097
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 11.246708869934082,
+      "learning_rate": 0.0001297196725378318,
+      "loss": 2.9586,
+      "step": 5098
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 10.041492462158203,
+      "learning_rate": 0.00012968246092780945,
+      "loss": 2.736,
+      "step": 5099
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 5.19471549987793,
+      "learning_rate": 0.00012964524931778713,
+      "loss": 2.646,
+      "step": 5100
+    },
+    {
+      "epoch": 1.79,
+      "eval_loss": 2.873821258544922,
+      "eval_runtime": 51.453,
+      "eval_samples_per_second": 42.136,
+      "eval_steps_per_second": 10.534,
+      "eval_wer": 1.0,
+      "step": 5100
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 9.86611557006836,
+      "learning_rate": 0.0001296080377077648,
+      "loss": 2.7976,
+      "step": 5101
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 10.880091667175293,
+      "learning_rate": 0.00012957082609774248,
+      "loss": 3.1771,
+      "step": 5102
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 6.943602085113525,
+      "learning_rate": 0.00012953361448772016,
+      "loss": 2.7736,
+      "step": 5103
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 7.127685070037842,
+      "learning_rate": 0.00012949640287769783,
+      "loss": 3.0676,
+      "step": 5104
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 18.559324264526367,
+      "learning_rate": 0.0001294591912676755,
+      "loss": 3.2199,
+      "step": 5105
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 2.158505916595459,
+      "learning_rate": 0.00012942197965765318,
+      "loss": 2.756,
+      "step": 5106
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 5.563910484313965,
+      "learning_rate": 0.00012938476804763086,
+      "loss": 2.8269,
+      "step": 5107
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 7.628482818603516,
+      "learning_rate": 0.0001293475564376085,
+      "loss": 2.9859,
+      "step": 5108
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 2.8963091373443604,
+      "learning_rate": 0.00012931034482758618,
+      "loss": 2.8255,
+      "step": 5109
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 3.270557165145874,
+      "learning_rate": 0.00012927313321756389,
+      "loss": 2.7845,
+      "step": 5110
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 6.280425548553467,
+      "learning_rate": 0.00012923592160754154,
+      "loss": 2.6916,
+      "step": 5111
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 6.498332977294922,
+      "learning_rate": 0.0001291987099975192,
+      "loss": 2.8811,
+      "step": 5112
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 5.853482723236084,
+      "learning_rate": 0.0001291614983874969,
+      "loss": 2.9358,
+      "step": 5113
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 6.113433361053467,
+      "learning_rate": 0.00012912428677747456,
+      "loss": 3.4643,
+      "step": 5114
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 4.190027236938477,
+      "learning_rate": 0.00012908707516745224,
+      "loss": 2.7947,
+      "step": 5115
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 7.508610248565674,
+      "learning_rate": 0.00012904986355742991,
+      "loss": 2.9305,
+      "step": 5116
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 7.606897830963135,
+      "learning_rate": 0.00012901265194740756,
+      "loss": 2.543,
+      "step": 5117
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 3.0015764236450195,
+      "learning_rate": 0.00012897544033738527,
+      "loss": 2.9905,
+      "step": 5118
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 3.0580461025238037,
+      "learning_rate": 0.00012893822872736294,
+      "loss": 3.1613,
+      "step": 5119
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 3.363950729370117,
+      "learning_rate": 0.0001289010171173406,
+      "loss": 2.8196,
+      "step": 5120
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 7.74603796005249,
+      "learning_rate": 0.00012886380550731827,
+      "loss": 2.9443,
+      "step": 5121
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 5.362524032592773,
+      "learning_rate": 0.00012882659389729594,
+      "loss": 2.7295,
+      "step": 5122
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 12.855659484863281,
+      "learning_rate": 0.00012878938228727362,
+      "loss": 3.0975,
+      "step": 5123
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 8.280109405517578,
+      "learning_rate": 0.0001287521706772513,
+      "loss": 3.0533,
+      "step": 5124
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 4.31693172454834,
+      "learning_rate": 0.00012871495906722897,
+      "loss": 2.5704,
+      "step": 5125
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 3.7444570064544678,
+      "learning_rate": 0.00012867774745720664,
+      "loss": 2.609,
+      "step": 5126
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 7.781627178192139,
+      "learning_rate": 0.00012864053584718432,
+      "loss": 2.6856,
+      "step": 5127
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 5.853876113891602,
+      "learning_rate": 0.000128603324237162,
+      "loss": 2.6211,
+      "step": 5128
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 4.829381942749023,
+      "learning_rate": 0.00012856611262713965,
+      "loss": 2.6558,
+      "step": 5129
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 7.683784484863281,
+      "learning_rate": 0.00012852890101711732,
+      "loss": 3.1132,
+      "step": 5130
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 7.247925758361816,
+      "learning_rate": 0.000128491689407095,
+      "loss": 2.9654,
+      "step": 5131
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 7.606195449829102,
+      "learning_rate": 0.00012845447779707267,
+      "loss": 3.026,
+      "step": 5132
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 7.540022373199463,
+      "learning_rate": 0.00012841726618705035,
+      "loss": 2.9083,
+      "step": 5133
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 8.643896102905273,
+      "learning_rate": 0.00012838005457702802,
+      "loss": 3.2663,
+      "step": 5134
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 5.016622543334961,
+      "learning_rate": 0.0001283428429670057,
+      "loss": 3.1068,
+      "step": 5135
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 3.617004871368408,
+      "learning_rate": 0.00012830563135698338,
+      "loss": 2.6929,
+      "step": 5136
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 2.4919373989105225,
+      "learning_rate": 0.00012826841974696105,
+      "loss": 2.7526,
+      "step": 5137
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 2.7996411323547363,
+      "learning_rate": 0.0001282312081369387,
+      "loss": 3.0863,
+      "step": 5138
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 14.32989501953125,
+      "learning_rate": 0.00012819399652691638,
+      "loss": 3.3443,
+      "step": 5139
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 13.708571434020996,
+      "learning_rate": 0.00012815678491689408,
+      "loss": 3.1483,
+      "step": 5140
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 10.349092483520508,
+      "learning_rate": 0.00012811957330687173,
+      "loss": 3.1138,
+      "step": 5141
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 2.638105630874634,
+      "learning_rate": 0.0001280823616968494,
+      "loss": 2.9172,
+      "step": 5142
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 3.8446462154388428,
+      "learning_rate": 0.00012804515008682708,
+      "loss": 3.0105,
+      "step": 5143
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 2.970163106918335,
+      "learning_rate": 0.00012800793847680475,
+      "loss": 2.6563,
+      "step": 5144
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 3.778756618499756,
+      "learning_rate": 0.00012797072686678243,
+      "loss": 2.5726,
+      "step": 5145
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 4.075989723205566,
+      "learning_rate": 0.0001279335152567601,
+      "loss": 2.6662,
+      "step": 5146
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 5.896014213562012,
+      "learning_rate": 0.00012789630364673778,
+      "loss": 2.9875,
+      "step": 5147
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 4.679042816162109,
+      "learning_rate": 0.00012785909203671543,
+      "loss": 2.9295,
+      "step": 5148
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 4.0041422843933105,
+      "learning_rate": 0.00012782188042669313,
+      "loss": 2.8009,
+      "step": 5149
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 5.066998481750488,
+      "learning_rate": 0.00012778466881667078,
+      "loss": 2.7084,
+      "step": 5150
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 4.032702445983887,
+      "learning_rate": 0.00012774745720664846,
+      "loss": 2.4773,
+      "step": 5151
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 5.38712215423584,
+      "learning_rate": 0.00012771024559662613,
+      "loss": 2.7513,
+      "step": 5152
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 9.350571632385254,
+      "learning_rate": 0.0001276730339866038,
+      "loss": 2.756,
+      "step": 5153
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 3.612746238708496,
+      "learning_rate": 0.00012763582237658149,
+      "loss": 3.0026,
+      "step": 5154
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 5.044943809509277,
+      "learning_rate": 0.00012759861076655916,
+      "loss": 3.0165,
+      "step": 5155
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 5.973738193511963,
+      "learning_rate": 0.00012756139915653684,
+      "loss": 3.0884,
+      "step": 5156
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 5.725467681884766,
+      "learning_rate": 0.00012752418754651449,
+      "loss": 2.9013,
+      "step": 5157
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 2.828871488571167,
+      "learning_rate": 0.0001274869759364922,
+      "loss": 2.7721,
+      "step": 5158
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 4.192399978637695,
+      "learning_rate": 0.00012744976432646984,
+      "loss": 2.9885,
+      "step": 5159
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 2.731215476989746,
+      "learning_rate": 0.0001274125527164475,
+      "loss": 2.9879,
+      "step": 5160
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 6.002933025360107,
+      "learning_rate": 0.0001273753411064252,
+      "loss": 2.9442,
+      "step": 5161
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 2.4598872661590576,
+      "learning_rate": 0.00012733812949640286,
+      "loss": 2.7012,
+      "step": 5162
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 6.382826328277588,
+      "learning_rate": 0.00012730091788638054,
+      "loss": 2.9316,
+      "step": 5163
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 3.4656591415405273,
+      "learning_rate": 0.00012726370627635822,
+      "loss": 2.6503,
+      "step": 5164
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 9.048042297363281,
+      "learning_rate": 0.0001272264946663359,
+      "loss": 2.9199,
+      "step": 5165
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 4.229365825653076,
+      "learning_rate": 0.00012718928305631354,
+      "loss": 2.8474,
+      "step": 5166
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 4.915844440460205,
+      "learning_rate": 0.00012715207144629124,
+      "loss": 2.8055,
+      "step": 5167
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 4.766310214996338,
+      "learning_rate": 0.00012711485983626892,
+      "loss": 2.7214,
+      "step": 5168
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 3.777912139892578,
+      "learning_rate": 0.00012707764822624657,
+      "loss": 2.8381,
+      "step": 5169
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 3.8435094356536865,
+      "learning_rate": 0.00012704043661622424,
+      "loss": 2.7613,
+      "step": 5170
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 3.1860551834106445,
+      "learning_rate": 0.00012700322500620192,
+      "loss": 2.8166,
+      "step": 5171
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 3.1261537075042725,
+      "learning_rate": 0.0001269660133961796,
+      "loss": 2.7175,
+      "step": 5172
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 13.234203338623047,
+      "learning_rate": 0.00012692880178615727,
+      "loss": 2.6832,
+      "step": 5173
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 4.1125264167785645,
+      "learning_rate": 0.00012689159017613495,
+      "loss": 3.0747,
+      "step": 5174
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 6.321409225463867,
+      "learning_rate": 0.0001268543785661126,
+      "loss": 2.6672,
+      "step": 5175
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 5.741054058074951,
+      "learning_rate": 0.0001268171669560903,
+      "loss": 2.5869,
+      "step": 5176
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 7.433377265930176,
+      "learning_rate": 0.00012677995534606797,
+      "loss": 2.5097,
+      "step": 5177
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 10.795042991638184,
+      "learning_rate": 0.00012674274373604562,
+      "loss": 2.767,
+      "step": 5178
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 3.698216199874878,
+      "learning_rate": 0.0001267055321260233,
+      "loss": 2.578,
+      "step": 5179
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 5.1897149085998535,
+      "learning_rate": 0.00012666832051600098,
+      "loss": 3.1661,
+      "step": 5180
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 13.520951271057129,
+      "learning_rate": 0.00012663110890597865,
+      "loss": 3.0335,
+      "step": 5181
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 3.7349276542663574,
+      "learning_rate": 0.00012659389729595633,
+      "loss": 2.9884,
+      "step": 5182
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 5.528043746948242,
+      "learning_rate": 0.000126556685685934,
+      "loss": 2.857,
+      "step": 5183
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 12.178468704223633,
+      "learning_rate": 0.00012651947407591168,
+      "loss": 2.8438,
+      "step": 5184
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 12.532594680786133,
+      "learning_rate": 0.00012648226246588935,
+      "loss": 2.8363,
+      "step": 5185
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 12.38633918762207,
+      "learning_rate": 0.00012644505085586703,
+      "loss": 3.0286,
+      "step": 5186
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 6.513081073760986,
+      "learning_rate": 0.00012640783924584468,
+      "loss": 2.9366,
+      "step": 5187
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 6.644528865814209,
+      "learning_rate": 0.00012637062763582235,
+      "loss": 2.8487,
+      "step": 5188
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 3.157069683074951,
+      "learning_rate": 0.00012633341602580006,
+      "loss": 2.9607,
+      "step": 5189
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 4.591786861419678,
+      "learning_rate": 0.0001262962044157777,
+      "loss": 2.6186,
+      "step": 5190
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 3.7421250343322754,
+      "learning_rate": 0.00012625899280575538,
+      "loss": 2.8948,
+      "step": 5191
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 6.926886558532715,
+      "learning_rate": 0.00012622178119573306,
+      "loss": 2.989,
+      "step": 5192
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 18.810298919677734,
+      "learning_rate": 0.00012618456958571073,
+      "loss": 3.0692,
+      "step": 5193
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 6.420024871826172,
+      "learning_rate": 0.0001261473579756884,
+      "loss": 2.7154,
+      "step": 5194
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 17.48590087890625,
+      "learning_rate": 0.00012611014636566608,
+      "loss": 2.7142,
+      "step": 5195
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 5.477769374847412,
+      "learning_rate": 0.00012607293475564373,
+      "loss": 2.8943,
+      "step": 5196
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 3.2958991527557373,
+      "learning_rate": 0.00012603572314562144,
+      "loss": 3.0727,
+      "step": 5197
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 2.588256359100342,
+      "learning_rate": 0.0001259985115355991,
+      "loss": 2.6309,
+      "step": 5198
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 4.680807113647461,
+      "learning_rate": 0.00012596129992557676,
+      "loss": 2.4823,
+      "step": 5199
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 8.458367347717285,
+      "learning_rate": 0.00012592408831555444,
+      "loss": 2.6377,
+      "step": 5200
+    },
+    {
+      "epoch": 1.82,
+      "eval_loss": 2.794273853302002,
+      "eval_runtime": 51.4561,
+      "eval_samples_per_second": 42.133,
+      "eval_steps_per_second": 10.533,
+      "eval_wer": 1.0,
+      "step": 5200
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 8562,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 400,
+  "total_flos": 2.1187209341970678e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}