diff --git "a/checkpoint-1500/trainer_state.json" "b/checkpoint-1500/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1500/trainer_state.json"
@@ -0,0 +1,9061 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.6816143497757847,
+  "global_step": 1500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 8.0265,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-07,
+      "loss": 9.6457,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-06,
+      "loss": 9.362,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-06,
+      "loss": 8.7549,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4e-06,
+      "loss": 8.7101,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999999999999997e-06,
+      "loss": 8.7091,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-06,
+      "loss": 8.9487,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2e-06,
+      "loss": 8.7391,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8e-06,
+      "loss": 9.3988,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.399999999999999e-06,
+      "loss": 8.7222,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999999999999999e-06,
+      "loss": 8.649,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.599999999999999e-06,
+      "loss": 8.8004,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.2e-06,
+      "loss": 8.5359,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.799999999999998e-06,
+      "loss": 8.6471,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.4e-06,
+      "loss": 8.6646,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.4e-06,
+      "loss": 8.7913,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.999999999999999e-06,
+      "loss": 8.7291,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.6e-06,
+      "loss": 8.9376,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.02e-05,
+      "loss": 8.6211,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0799999999999998e-05,
+      "loss": 8.4717,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.14e-05,
+      "loss": 8.8119,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1999999999999999e-05,
+      "loss": 8.0399,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.26e-05,
+      "loss": 7.8182,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3199999999999997e-05,
+      "loss": 7.7515,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3799999999999998e-05,
+      "loss": 8.6384,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.44e-05,
+      "loss": 8.0296,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4999999999999999e-05,
+      "loss": 7.7783,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5599999999999996e-05,
+      "loss": 7.3056,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5599999999999996e-05,
+      "loss": 7.2813,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6199999999999997e-05,
+      "loss": 7.2552,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.68e-05,
+      "loss": 6.9639,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.74e-05,
+      "loss": 7.4774,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7999999999999997e-05,
+      "loss": 6.9855,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8599999999999998e-05,
+      "loss": 6.2808,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.92e-05,
+      "loss": 6.3533,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.98e-05,
+      "loss": 5.7131,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.04e-05,
+      "loss": 5.4075,
+      "step": 37
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1e-05,
+      "loss": 5.3897,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1599999999999996e-05,
+      "loss": 4.7228,
+      "step": 39
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2199999999999998e-05,
+      "loss": 4.4425,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.28e-05,
+      "loss": 4.2483,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.34e-05,
+      "loss": 4.0856,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3999999999999997e-05,
+      "loss": 3.7537,
+      "step": 43
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4599999999999998e-05,
+      "loss": 3.5862,
+      "step": 44
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.52e-05,
+      "loss": 3.4923,
+      "step": 45
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5799999999999997e-05,
+      "loss": 3.3951,
+      "step": 46
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6399999999999995e-05,
+      "loss": 3.3228,
+      "step": 47
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6999999999999996e-05,
+      "loss": 3.2852,
+      "step": 48
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.7599999999999997e-05,
+      "loss": 3.2376,
+      "step": 49
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 3.2277,
+      "step": 50
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.88e-05,
+      "loss": 3.1758,
+      "step": 51
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.94e-05,
+      "loss": 3.1334,
+      "step": 52
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 3.1358,
+      "step": 53
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.06e-05,
+      "loss": 3.0882,
+      "step": 54
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.119999999999999e-05,
+      "loss": 3.0612,
+      "step": 55
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.1799999999999994e-05,
+      "loss": 3.0565,
+      "step": 56
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.2399999999999995e-05,
+      "loss": 3.0757,
+      "step": 57
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2999999999999996e-05,
+      "loss": 3.0269,
+      "step": 58
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.36e-05,
+      "loss": 3.0021,
+      "step": 59
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.42e-05,
+      "loss": 2.9934,
+      "step": 60
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.48e-05,
+      "loss": 2.9729,
+      "step": 61
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.539999999999999e-05,
+      "loss": 2.9747,
+      "step": 62
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5999999999999994e-05,
+      "loss": 3.0145,
+      "step": 63
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6599999999999995e-05,
+      "loss": 2.9496,
+      "step": 64
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.7199999999999996e-05,
+      "loss": 2.9465,
+      "step": 65
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.78e-05,
+      "loss": 2.961,
+      "step": 66
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.84e-05,
+      "loss": 2.9338,
+      "step": 67
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9e-05,
+      "loss": 2.9233,
+      "step": 68
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.96e-05,
+      "loss": 3.0073,
+      "step": 69
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.02e-05,
+      "loss": 2.924,
+      "step": 70
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.08e-05,
+      "loss": 2.9184,
+      "step": 71
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.14e-05,
+      "loss": 2.9264,
+      "step": 72
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.2e-05,
+      "loss": 2.9444,
+      "step": 73
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.259999999999999e-05,
+      "loss": 2.9186,
+      "step": 74
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.319999999999999e-05,
+      "loss": 2.9389,
+      "step": 75
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.3799999999999994e-05,
+      "loss": 2.9462,
+      "step": 76
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.4399999999999995e-05,
+      "loss": 2.9256,
+      "step": 77
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 2.8994,
+      "step": 78
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.56e-05,
+      "loss": 2.9328,
+      "step": 79
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.62e-05,
+      "loss": 2.8969,
+      "step": 80
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.68e-05,
+      "loss": 2.9254,
+      "step": 81
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.7399999999999993e-05,
+      "loss": 2.9022,
+      "step": 82
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.7999999999999994e-05,
+      "loss": 2.9015,
+      "step": 83
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8599999999999995e-05,
+      "loss": 2.8977,
+      "step": 84
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9199999999999997e-05,
+      "loss": 2.8985,
+      "step": 85
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.98e-05,
+      "loss": 2.8919,
+      "step": 86
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.04e-05,
+      "loss": 2.8912,
+      "step": 87
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.1e-05,
+      "loss": 2.9541,
+      "step": 88
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.1599999999999994e-05,
+      "loss": 2.8891,
+      "step": 89
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.2199999999999995e-05,
+      "loss": 2.8847,
+      "step": 90
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.279999999999999e-05,
+      "loss": 2.8736,
+      "step": 91
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.339999999999999e-05,
+      "loss": 2.9021,
+      "step": 92
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 2.879,
+      "step": 93
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.459999999999999e-05,
+      "loss": 2.898,
+      "step": 94
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.519999999999999e-05,
+      "loss": 2.8717,
+      "step": 95
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.5799999999999994e-05,
+      "loss": 2.8866,
+      "step": 96
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 2.8688,
+      "step": 97
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 2.8748,
+      "step": 98
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.76e-05,
+      "loss": 2.872,
+      "step": 99
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.82e-05,
+      "loss": 2.9029,
+      "step": 100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.88e-05,
+      "loss": 2.8795,
+      "step": 101
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.94e-05,
+      "loss": 2.8743,
+      "step": 102
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 2.8779,
+      "step": 103
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 6.0599999999999996e-05,
+      "loss": 2.8687,
+      "step": 104
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 6.12e-05,
+      "loss": 2.8558,
+      "step": 105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 6.18e-05,
+      "loss": 2.8933,
+      "step": 106
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 6.239999999999999e-05,
+      "loss": 2.91,
+      "step": 107
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 6.299999999999999e-05,
+      "loss": 2.8661,
+      "step": 108
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 6.359999999999999e-05,
+      "loss": 2.8687,
+      "step": 109
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 6.419999999999999e-05,
+      "loss": 2.8606,
+      "step": 110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 6.479999999999999e-05,
+      "loss": 2.8498,
+      "step": 111
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.539999999999999e-05,
+      "loss": 2.8612,
+      "step": 112
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 2.8685,
+      "step": 113
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.659999999999999e-05,
+      "loss": 2.8563,
+      "step": 114
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.72e-05,
+      "loss": 2.865,
+      "step": 115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.78e-05,
+      "loss": 2.8668,
+      "step": 116
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.84e-05,
+      "loss": 2.8734,
+      "step": 117
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.9e-05,
+      "loss": 2.8983,
+      "step": 118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.96e-05,
+      "loss": 2.8978,
+      "step": 119
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.02e-05,
+      "loss": 2.8513,
+      "step": 120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.079999999999999e-05,
+      "loss": 2.8537,
+      "step": 121
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.139999999999999e-05,
+      "loss": 2.8568,
+      "step": 122
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.199999999999999e-05,
+      "loss": 2.8565,
+      "step": 123
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.259999999999999e-05,
+      "loss": 2.8663,
+      "step": 124
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.319999999999999e-05,
+      "loss": 2.8909,
+      "step": 125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.379999999999999e-05,
+      "loss": 2.8472,
+      "step": 126
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.439999999999999e-05,
+      "loss": 2.8641,
+      "step": 127
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.5e-05,
+      "loss": 2.8596,
+      "step": 128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.56e-05,
+      "loss": 2.8612,
+      "step": 129
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.62e-05,
+      "loss": 2.8476,
+      "step": 130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.68e-05,
+      "loss": 2.8516,
+      "step": 131
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.74e-05,
+      "loss": 2.8784,
+      "step": 132
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.8e-05,
+      "loss": 2.8527,
+      "step": 133
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.86e-05,
+      "loss": 2.8523,
+      "step": 134
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.92e-05,
+      "loss": 2.8557,
+      "step": 135
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.98e-05,
+      "loss": 2.8502,
+      "step": 136
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.04e-05,
+      "loss": 2.8402,
+      "step": 137
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.1e-05,
+      "loss": 2.8783,
+      "step": 138
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.16e-05,
+      "loss": 2.8563,
+      "step": 139
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.22e-05,
+      "loss": 2.8409,
+      "step": 140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.28e-05,
+      "loss": 2.8594,
+      "step": 141
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.34e-05,
+      "loss": 2.8603,
+      "step": 142
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.4e-05,
+      "loss": 2.8627,
+      "step": 143
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.459999999999998e-05,
+      "loss": 2.8864,
+      "step": 144
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.519999999999998e-05,
+      "loss": 2.8493,
+      "step": 145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.579999999999998e-05,
+      "loss": 2.8494,
+      "step": 146
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.639999999999999e-05,
+      "loss": 2.8482,
+      "step": 147
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.699999999999999e-05,
+      "loss": 2.8333,
+      "step": 148
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.759999999999999e-05,
+      "loss": 2.8456,
+      "step": 149
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.819999999999999e-05,
+      "loss": 2.8867,
+      "step": 150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.879999999999999e-05,
+      "loss": 2.8344,
+      "step": 151
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.939999999999999e-05,
+      "loss": 2.8513,
+      "step": 152
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 2.8408,
+      "step": 153
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.059999999999999e-05,
+      "loss": 2.8496,
+      "step": 154
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.12e-05,
+      "loss": 2.8413,
+      "step": 155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.18e-05,
+      "loss": 2.8584,
+      "step": 156
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.24e-05,
+      "loss": 2.8416,
+      "step": 157
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.3e-05,
+      "loss": 2.8487,
+      "step": 158
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.36e-05,
+      "loss": 2.8401,
+      "step": 159
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.419999999999999e-05,
+      "loss": 2.8516,
+      "step": 160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.479999999999999e-05,
+      "loss": 2.8372,
+      "step": 161
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.539999999999999e-05,
+      "loss": 2.8578,
+      "step": 162
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.599999999999999e-05,
+      "loss": 2.8729,
+      "step": 163
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.659999999999999e-05,
+      "loss": 2.8404,
+      "step": 164
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.719999999999999e-05,
+      "loss": 2.8379,
+      "step": 165
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.779999999999999e-05,
+      "loss": 2.8401,
+      "step": 166
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.839999999999999e-05,
+      "loss": 2.8484,
+      "step": 167
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.9e-05,
+      "loss": 2.8378,
+      "step": 168
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.96e-05,
+      "loss": 2.8709,
+      "step": 169
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001002,
+      "loss": 2.8357,
+      "step": 170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001008,
+      "loss": 2.8479,
+      "step": 171
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001014,
+      "loss": 2.8337,
+      "step": 172
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000102,
+      "loss": 2.8393,
+      "step": 173
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001026,
+      "loss": 2.8397,
+      "step": 174
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010319999999999999,
+      "loss": 2.9035,
+      "step": 175
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010379999999999999,
+      "loss": 2.8376,
+      "step": 176
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010439999999999999,
+      "loss": 2.8368,
+      "step": 177
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010499999999999999,
+      "loss": 2.8464,
+      "step": 178
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010559999999999998,
+      "loss": 2.8347,
+      "step": 179
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010619999999999998,
+      "loss": 2.8562,
+      "step": 180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010679999999999998,
+      "loss": 2.849,
+      "step": 181
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010739999999999998,
+      "loss": 2.8344,
+      "step": 182
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010799999999999998,
+      "loss": 2.8496,
+      "step": 183
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010859999999999998,
+      "loss": 2.8328,
+      "step": 184
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010919999999999998,
+      "loss": 2.8375,
+      "step": 185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010979999999999999,
+      "loss": 2.8238,
+      "step": 186
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00011039999999999999,
+      "loss": 2.8507,
+      "step": 187
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00011099999999999999,
+      "loss": 2.831,
+      "step": 188
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00011159999999999999,
+      "loss": 2.8317,
+      "step": 189
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00011219999999999999,
+      "loss": 2.8434,
+      "step": 190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00011279999999999999,
+      "loss": 2.8359,
+      "step": 191
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00011339999999999999,
+      "loss": 2.846,
+      "step": 192
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 2.8551,
+      "step": 193
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001146,
+      "loss": 2.8473,
+      "step": 194
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001152,
+      "loss": 2.8673,
+      "step": 195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001158,
+      "loss": 2.8396,
+      "step": 196
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001164,
+      "loss": 2.8488,
+      "step": 197
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000117,
+      "loss": 2.8357,
+      "step": 198
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001176,
+      "loss": 2.8509,
+      "step": 199
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001182,
+      "loss": 2.8593,
+      "step": 200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001188,
+      "loss": 2.8361,
+      "step": 201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001194,
+      "loss": 2.8432,
+      "step": 202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 2.8481,
+      "step": 203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012059999999999999,
+      "loss": 2.8504,
+      "step": 204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012119999999999999,
+      "loss": 2.865,
+      "step": 205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012179999999999999,
+      "loss": 2.8561,
+      "step": 206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001224,
+      "loss": 2.873,
+      "step": 207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012299999999999998,
+      "loss": 2.8248,
+      "step": 208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001236,
+      "loss": 2.8255,
+      "step": 209
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00012419999999999998,
+      "loss": 2.8492,
+      "step": 210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00012479999999999997,
+      "loss": 2.8483,
+      "step": 211
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00012539999999999999,
+      "loss": 2.8494,
+      "step": 212
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00012599999999999997,
+      "loss": 2.8752,
+      "step": 213
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001266,
+      "loss": 2.8278,
+      "step": 214
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00012719999999999997,
+      "loss": 2.8374,
+      "step": 215
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001278,
+      "loss": 2.8581,
+      "step": 216
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00012839999999999998,
+      "loss": 2.8287,
+      "step": 217
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000129,
+      "loss": 2.8714,
+      "step": 218
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00012959999999999998,
+      "loss": 2.8519,
+      "step": 219
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001302,
+      "loss": 2.8392,
+      "step": 220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00013079999999999998,
+      "loss": 2.8377,
+      "step": 221
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001314,
+      "loss": 2.831,
+      "step": 222
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00013199999999999998,
+      "loss": 2.838,
+      "step": 223
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001326,
+      "loss": 2.8341,
+      "step": 224
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00013319999999999999,
+      "loss": 2.8417,
+      "step": 225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001338,
+      "loss": 2.83,
+      "step": 226
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001344,
+      "loss": 2.8264,
+      "step": 227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000135,
+      "loss": 2.8398,
+      "step": 228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001356,
+      "loss": 2.8433,
+      "step": 229
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001362,
+      "loss": 2.8408,
+      "step": 230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001368,
+      "loss": 2.8808,
+      "step": 231
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001374,
+      "loss": 2.8628,
+      "step": 232
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000138,
+      "loss": 2.8362,
+      "step": 233
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001386,
+      "loss": 2.8299,
+      "step": 234
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001392,
+      "loss": 2.8406,
+      "step": 235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00013979999999999998,
+      "loss": 2.8327,
+      "step": 236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001404,
+      "loss": 2.8442,
+      "step": 237
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00014099999999999998,
+      "loss": 2.8627,
+      "step": 238
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00014159999999999997,
+      "loss": 2.8465,
+      "step": 239
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001422,
+      "loss": 2.8335,
+      "step": 240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00014279999999999997,
+      "loss": 2.8473,
+      "step": 241
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001434,
+      "loss": 2.8434,
+      "step": 242
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00014399999999999998,
+      "loss": 2.8577,
+      "step": 243
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001446,
+      "loss": 2.8799,
+      "step": 244
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00014519999999999998,
+      "loss": 2.8325,
+      "step": 245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001458,
+      "loss": 2.851,
+      "step": 246
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00014639999999999998,
+      "loss": 2.8372,
+      "step": 247
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000147,
+      "loss": 2.8451,
+      "step": 248
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00014759999999999998,
+      "loss": 2.8528,
+      "step": 249
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001482,
+      "loss": 2.8462,
+      "step": 250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00014879999999999998,
+      "loss": 2.8484,
+      "step": 251
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001494,
+      "loss": 2.8569,
+      "step": 252
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00015,
+      "loss": 2.8296,
+      "step": 253
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00015059999999999997,
+      "loss": 2.8295,
+      "step": 254
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001512,
+      "loss": 2.8319,
+      "step": 255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015179999999999998,
+      "loss": 2.8489,
+      "step": 256
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001524,
+      "loss": 2.856,
+      "step": 257
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015299999999999998,
+      "loss": 2.8317,
+      "step": 258
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001536,
+      "loss": 2.8445,
+      "step": 259
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015419999999999998,
+      "loss": 2.8684,
+      "step": 260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001548,
+      "loss": 2.8355,
+      "step": 261
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015539999999999998,
+      "loss": 2.8629,
+      "step": 262
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000156,
+      "loss": 2.861,
+      "step": 263
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015659999999999998,
+      "loss": 2.8359,
+      "step": 264
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001572,
+      "loss": 2.829,
+      "step": 265
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001578,
+      "loss": 2.8127,
+      "step": 266
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001584,
+      "loss": 2.844,
+      "step": 267
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000159,
+      "loss": 2.826,
+      "step": 268
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001596,
+      "loss": 2.8542,
+      "step": 269
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001602,
+      "loss": 2.8384,
+      "step": 270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001608,
+      "loss": 2.8334,
+      "step": 271
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001614,
+      "loss": 2.8314,
+      "step": 272
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000162,
+      "loss": 2.8504,
+      "step": 273
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001626,
+      "loss": 2.8535,
+      "step": 274
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001632,
+      "loss": 2.8668,
+      "step": 275
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001638,
+      "loss": 2.8491,
+      "step": 276
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001644,
+      "loss": 2.8351,
+      "step": 277
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000165,
+      "loss": 2.836,
+      "step": 278
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001656,
+      "loss": 2.8356,
+      "step": 279
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001662,
+      "loss": 2.8485,
+      "step": 280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001668,
+      "loss": 2.8357,
+      "step": 281
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001674,
+      "loss": 2.8615,
+      "step": 282
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000168,
+      "loss": 2.8306,
+      "step": 283
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001686,
+      "loss": 2.8324,
+      "step": 284
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016919999999999997,
+      "loss": 2.8268,
+      "step": 285
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016979999999999998,
+      "loss": 2.8308,
+      "step": 286
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017039999999999997,
+      "loss": 2.8503,
+      "step": 287
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017099999999999998,
+      "loss": 2.8492,
+      "step": 288
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017159999999999997,
+      "loss": 2.8365,
+      "step": 289
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017219999999999998,
+      "loss": 2.8302,
+      "step": 290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017279999999999997,
+      "loss": 2.8217,
+      "step": 291
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017339999999999996,
+      "loss": 2.827,
+      "step": 292
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017399999999999997,
+      "loss": 2.8504,
+      "step": 293
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017459999999999996,
+      "loss": 2.8404,
+      "step": 294
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017519999999999998,
+      "loss": 2.8345,
+      "step": 295
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017579999999999996,
+      "loss": 2.8334,
+      "step": 296
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017639999999999998,
+      "loss": 2.8301,
+      "step": 297
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017699999999999997,
+      "loss": 2.839,
+      "step": 298
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017759999999999998,
+      "loss": 2.838,
+      "step": 299
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017819999999999997,
+      "loss": 2.8805,
+      "step": 300
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 2.8686437606811523,
+      "eval_runtime": 36.2023,
+      "eval_samples_per_second": 72.979,
+      "eval_steps_per_second": 2.293,
+      "eval_wer": 1.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017879999999999998,
+      "loss": 2.8497,
+      "step": 301
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017939999999999997,
+      "loss": 2.8304,
+      "step": 302
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 2.8437,
+      "step": 303
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018059999999999997,
+      "loss": 2.8487,
+      "step": 304
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018119999999999999,
+      "loss": 2.829,
+      "step": 305
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018179999999999997,
+      "loss": 2.8427,
+      "step": 306
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001824,
+      "loss": 2.8383,
+      "step": 307
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018299999999999998,
+      "loss": 2.8406,
+      "step": 308
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001836,
+      "loss": 2.8353,
+      "step": 309
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018419999999999998,
+      "loss": 2.8616,
+      "step": 310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001848,
+      "loss": 2.8339,
+      "step": 311
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018539999999999998,
+      "loss": 2.8324,
+      "step": 312
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000186,
+      "loss": 2.8454,
+      "step": 313
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018659999999999998,
+      "loss": 2.8174,
+      "step": 314
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001872,
+      "loss": 2.832,
+      "step": 315
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018779999999999998,
+      "loss": 2.8404,
+      "step": 316
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018839999999999997,
+      "loss": 2.8245,
+      "step": 317
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018899999999999999,
+      "loss": 2.8401,
+      "step": 318
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018959999999999997,
+      "loss": 2.8415,
+      "step": 319
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001902,
+      "loss": 2.8,
+      "step": 320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019079999999999998,
+      "loss": 2.8142,
+      "step": 321
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001914,
+      "loss": 2.7965,
+      "step": 322
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019199999999999998,
+      "loss": 2.8312,
+      "step": 323
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001926,
+      "loss": 2.7999,
+      "step": 324
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019319999999999998,
+      "loss": 2.8111,
+      "step": 325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001938,
+      "loss": 2.7723,
+      "step": 326
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019439999999999998,
+      "loss": 2.7224,
+      "step": 327
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000195,
+      "loss": 2.7085,
+      "step": 328
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019559999999999998,
+      "loss": 2.6938,
+      "step": 329
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001962,
+      "loss": 2.6334,
+      "step": 330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019679999999999999,
+      "loss": 2.6267,
+      "step": 331
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001974,
+      "loss": 2.6394,
+      "step": 332
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000198,
+      "loss": 2.5338,
+      "step": 333
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001986,
+      "loss": 2.5299,
+      "step": 334
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001992,
+      "loss": 2.4624,
+      "step": 335
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001998,
+      "loss": 2.4906,
+      "step": 336
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002004,
+      "loss": 2.4839,
+      "step": 337
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000201,
+      "loss": 2.4099,
+      "step": 338
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002016,
+      "loss": 2.3509,
+      "step": 339
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002022,
+      "loss": 2.2866,
+      "step": 340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002028,
+      "loss": 2.2153,
+      "step": 341
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00020339999999999998,
+      "loss": 2.159,
+      "step": 342
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000204,
+      "loss": 2.1476,
+      "step": 343
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00020459999999999999,
+      "loss": 2.1165,
+      "step": 344
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002052,
+      "loss": 1.9656,
+      "step": 345
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002058,
+      "loss": 1.9216,
+      "step": 346
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00020639999999999998,
+      "loss": 1.8228,
+      "step": 347
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00020699999999999996,
+      "loss": 1.7584,
+      "step": 348
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00020759999999999998,
+      "loss": 1.7093,
+      "step": 349
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00020819999999999996,
+      "loss": 1.6913,
+      "step": 350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00020879999999999998,
+      "loss": 1.5864,
+      "step": 351
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00020939999999999997,
+      "loss": 1.4857,
+      "step": 352
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00020999999999999998,
+      "loss": 1.4402,
+      "step": 353
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00021059999999999997,
+      "loss": 1.3677,
+      "step": 354
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00021119999999999996,
+      "loss": 1.3489,
+      "step": 355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00021179999999999997,
+      "loss": 1.3761,
+      "step": 356
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00021239999999999996,
+      "loss": 1.2986,
+      "step": 357
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00021299999999999997,
+      "loss": 1.179,
+      "step": 358
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00021359999999999996,
+      "loss": 1.1818,
+      "step": 359
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00021419999999999998,
+      "loss": 1.171,
+      "step": 360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00021479999999999996,
+      "loss": 1.0919,
+      "step": 361
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00021539999999999998,
+      "loss": 1.1117,
+      "step": 362
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00021599999999999996,
+      "loss": 1.149,
+      "step": 363
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00021659999999999998,
+      "loss": 1.0115,
+      "step": 364
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00021719999999999997,
+      "loss": 0.9594,
+      "step": 365
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00021779999999999998,
+      "loss": 0.9237,
+      "step": 366
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00021839999999999997,
+      "loss": 0.8767,
+      "step": 367
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00021899999999999998,
+      "loss": 0.8822,
+      "step": 368
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00021959999999999997,
+      "loss": 0.9433,
+      "step": 369
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00022019999999999999,
+      "loss": 0.8718,
+      "step": 370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00022079999999999997,
+      "loss": 0.8057,
+      "step": 371
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002214,
+      "loss": 0.8207,
+      "step": 372
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00022199999999999998,
+      "loss": 0.81,
+      "step": 373
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002226,
+      "loss": 0.7698,
+      "step": 374
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00022319999999999998,
+      "loss": 0.8984,
+      "step": 375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002238,
+      "loss": 0.7646,
+      "step": 376
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00022439999999999998,
+      "loss": 0.7558,
+      "step": 377
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000225,
+      "loss": 0.6822,
+      "step": 378
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00022559999999999998,
+      "loss": 0.6844,
+      "step": 379
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00022619999999999997,
+      "loss": 0.6854,
+      "step": 380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00022679999999999998,
+      "loss": 0.7588,
+      "step": 381
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00022739999999999997,
+      "loss": 0.6775,
+      "step": 382
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00022799999999999999,
+      "loss": 0.5956,
+      "step": 383
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00022859999999999997,
+      "loss": 0.6253,
+      "step": 384
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002292,
+      "loss": 0.5935,
+      "step": 385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00022979999999999997,
+      "loss": 0.5634,
+      "step": 386
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002304,
+      "loss": 0.6529,
+      "step": 387
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00023099999999999998,
+      "loss": 0.6543,
+      "step": 388
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002316,
+      "loss": 0.5912,
+      "step": 389
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00023219999999999998,
+      "loss": 0.6042,
+      "step": 390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002328,
+      "loss": 0.5527,
+      "step": 391
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00023339999999999998,
+      "loss": 0.5539,
+      "step": 392
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000234,
+      "loss": 0.5234,
+      "step": 393
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00023459999999999998,
+      "loss": 0.6204,
+      "step": 394
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002352,
+      "loss": 0.4771,
+      "step": 395
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00023579999999999999,
+      "loss": 0.4725,
+      "step": 396
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002364,
+      "loss": 0.4875,
+      "step": 397
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000237,
+      "loss": 0.4783,
+      "step": 398
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002376,
+      "loss": 0.5153,
+      "step": 399
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002382,
+      "loss": 0.5878,
+      "step": 400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002388,
+      "loss": 0.4721,
+      "step": 401
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002394,
+      "loss": 0.4378,
+      "step": 402
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 0.4651,
+      "step": 403
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002406,
+      "loss": 0.4455,
+      "step": 404
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024119999999999998,
+      "loss": 0.4101,
+      "step": 405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002418,
+      "loss": 0.5119,
+      "step": 406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00024239999999999998,
+      "loss": 0.4863,
+      "step": 407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000243,
+      "loss": 0.4024,
+      "step": 408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00024359999999999999,
+      "loss": 0.382,
+      "step": 409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00024419999999999997,
+      "loss": 0.3854,
+      "step": 410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002448,
+      "loss": 0.4214,
+      "step": 411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00024539999999999995,
+      "loss": 0.415,
+      "step": 412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00024599999999999996,
+      "loss": 0.4531,
+      "step": 413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002466,
+      "loss": 0.3696,
+      "step": 414
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002472,
+      "loss": 0.369,
+      "step": 415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00024779999999999995,
+      "loss": 0.3885,
+      "step": 416
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00024839999999999997,
+      "loss": 0.3686,
+      "step": 417
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000249,
+      "loss": 0.4116,
+      "step": 418
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00024959999999999994,
+      "loss": 0.4578,
+      "step": 419
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00025019999999999996,
+      "loss": 0.3325,
+      "step": 420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00025079999999999997,
+      "loss": 0.3539,
+      "step": 421
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002514,
+      "loss": 0.3416,
+      "step": 422
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00025199999999999995,
+      "loss": 0.3386,
+      "step": 423
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00025259999999999996,
+      "loss": 0.3693,
+      "step": 424
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002532,
+      "loss": 0.5373,
+      "step": 425
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002538,
+      "loss": 0.3103,
+      "step": 426
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00025439999999999995,
+      "loss": 0.3666,
+      "step": 427
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00025499999999999996,
+      "loss": 0.3187,
+      "step": 428
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002556,
+      "loss": 0.36,
+      "step": 429
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002562,
+      "loss": 0.3618,
+      "step": 430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00025679999999999995,
+      "loss": 0.3716,
+      "step": 431
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00025739999999999997,
+      "loss": 0.3804,
+      "step": 432
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000258,
+      "loss": 0.3472,
+      "step": 433
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002586,
+      "loss": 0.3028,
+      "step": 434
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025919999999999996,
+      "loss": 0.2999,
+      "step": 435
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025979999999999997,
+      "loss": 0.3496,
+      "step": 436
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002604,
+      "loss": 0.2927,
+      "step": 437
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000261,
+      "loss": 0.3586,
+      "step": 438
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00026159999999999996,
+      "loss": 0.2821,
+      "step": 439
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002622,
+      "loss": 0.2791,
+      "step": 440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002628,
+      "loss": 0.3063,
+      "step": 441
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00026339999999999995,
+      "loss": 0.3224,
+      "step": 442
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00026399999999999997,
+      "loss": 0.3651,
+      "step": 443
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002646,
+      "loss": 0.4001,
+      "step": 444
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002652,
+      "loss": 0.2939,
+      "step": 445
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00026579999999999996,
+      "loss": 0.2764,
+      "step": 446
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00026639999999999997,
+      "loss": 0.2818,
+      "step": 447
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000267,
+      "loss": 0.3167,
+      "step": 448
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002676,
+      "loss": 0.2833,
+      "step": 449
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00026819999999999996,
+      "loss": 0.4571,
+      "step": 450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002688,
+      "loss": 0.2955,
+      "step": 451
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002694,
+      "loss": 0.2776,
+      "step": 452
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00027,
+      "loss": 0.2502,
+      "step": 453
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00027059999999999996,
+      "loss": 0.3141,
+      "step": 454
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002712,
+      "loss": 0.2737,
+      "step": 455
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002718,
+      "loss": 0.3388,
+      "step": 456
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002724,
+      "loss": 0.3591,
+      "step": 457
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00027299999999999997,
+      "loss": 0.2728,
+      "step": 458
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002736,
+      "loss": 0.2729,
+      "step": 459
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002742,
+      "loss": 0.2939,
+      "step": 460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002748,
+      "loss": 0.2766,
+      "step": 461
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00027539999999999997,
+      "loss": 0.2568,
+      "step": 462
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000276,
+      "loss": 0.3188,
+      "step": 463
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002766,
+      "loss": 0.2798,
+      "step": 464
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002772,
+      "loss": 0.3083,
+      "step": 465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002778,
+      "loss": 0.2583,
+      "step": 466
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002784,
+      "loss": 0.2578,
+      "step": 467
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000279,
+      "loss": 0.2657,
+      "step": 468
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027959999999999997,
+      "loss": 0.3249,
+      "step": 469
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002802,
+      "loss": 0.2748,
+      "step": 470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002808,
+      "loss": 0.2234,
+      "step": 471
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00028139999999999996,
+      "loss": 0.238,
+      "step": 472
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00028199999999999997,
+      "loss": 0.2584,
+      "step": 473
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002826,
+      "loss": 0.2877,
+      "step": 474
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00028319999999999994,
+      "loss": 0.3936,
+      "step": 475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00028379999999999996,
+      "loss": 0.2264,
+      "step": 476
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002844,
+      "loss": 0.2877,
+      "step": 477
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000285,
+      "loss": 0.2006,
+      "step": 478
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00028559999999999995,
+      "loss": 0.2299,
+      "step": 479
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00028619999999999996,
+      "loss": 0.2588,
+      "step": 480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002868,
+      "loss": 0.2508,
+      "step": 481
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00028739999999999994,
+      "loss": 0.2833,
+      "step": 482
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00028799999999999995,
+      "loss": 0.2436,
+      "step": 483
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00028859999999999997,
+      "loss": 0.2479,
+      "step": 484
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002892,
+      "loss": 0.2512,
+      "step": 485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00028979999999999994,
+      "loss": 0.2199,
+      "step": 486
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029039999999999996,
+      "loss": 0.219,
+      "step": 487
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029099999999999997,
+      "loss": 0.3462,
+      "step": 488
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002916,
+      "loss": 0.2229,
+      "step": 489
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029219999999999995,
+      "loss": 0.251,
+      "step": 490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029279999999999996,
+      "loss": 0.2532,
+      "step": 491
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002934,
+      "loss": 0.2038,
+      "step": 492
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000294,
+      "loss": 0.1987,
+      "step": 493
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029459999999999995,
+      "loss": 0.3284,
+      "step": 494
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029519999999999997,
+      "loss": 0.2204,
+      "step": 495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002958,
+      "loss": 0.2399,
+      "step": 496
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002964,
+      "loss": 0.2073,
+      "step": 497
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029699999999999996,
+      "loss": 0.1897,
+      "step": 498
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029759999999999997,
+      "loss": 0.2006,
+      "step": 499
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002982,
+      "loss": 0.349,
+      "step": 500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002988,
+      "loss": 0.2335,
+      "step": 501
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029939999999999996,
+      "loss": 0.2133,
+      "step": 502
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003,
+      "loss": 0.2455,
+      "step": 503
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029986213235294113,
+      "loss": 0.2455,
+      "step": 504
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002997242647058823,
+      "loss": 0.1986,
+      "step": 505
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002995863970588235,
+      "loss": 0.3048,
+      "step": 506
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002994485294117647,
+      "loss": 0.2729,
+      "step": 507
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029931066176470587,
+      "loss": 0.2483,
+      "step": 508
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000299172794117647,
+      "loss": 0.2313,
+      "step": 509
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029903492647058823,
+      "loss": 0.2179,
+      "step": 510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002988970588235294,
+      "loss": 0.2335,
+      "step": 511
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029875919117647055,
+      "loss": 0.214,
+      "step": 512
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002986213235294117,
+      "loss": 0.2768,
+      "step": 513
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002984834558823529,
+      "loss": 0.2058,
+      "step": 514
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002983455882352941,
+      "loss": 0.2162,
+      "step": 515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002982077205882353,
+      "loss": 0.2289,
+      "step": 516
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029806985294117644,
+      "loss": 0.2009,
+      "step": 517
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029793198529411765,
+      "loss": 0.2307,
+      "step": 518
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002977941176470588,
+      "loss": 0.3397,
+      "step": 519
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029765624999999996,
+      "loss": 0.2383,
+      "step": 520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002975183823529411,
+      "loss": 0.1964,
+      "step": 521
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029738051470588233,
+      "loss": 0.2292,
+      "step": 522
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029724264705882354,
+      "loss": 0.2288,
+      "step": 523
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002971047794117647,
+      "loss": 0.184,
+      "step": 524
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029696691176470585,
+      "loss": 0.2612,
+      "step": 525
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000296829044117647,
+      "loss": 0.1956,
+      "step": 526
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002966911764705882,
+      "loss": 0.2099,
+      "step": 527
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002965533088235294,
+      "loss": 0.175,
+      "step": 528
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029641544117647053,
+      "loss": 0.2164,
+      "step": 529
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029627757352941174,
+      "loss": 0.2202,
+      "step": 530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029613970588235295,
+      "loss": 0.2765,
+      "step": 531
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002960018382352941,
+      "loss": 0.2556,
+      "step": 532
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029586397058823527,
+      "loss": 0.2198,
+      "step": 533
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002957261029411764,
+      "loss": 0.2429,
+      "step": 534
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029558823529411763,
+      "loss": 0.1895,
+      "step": 535
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002954503676470588,
+      "loss": 0.1957,
+      "step": 536
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029531249999999995,
+      "loss": 0.2083,
+      "step": 537
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029517463235294116,
+      "loss": 0.2724,
+      "step": 538
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002950367647058823,
+      "loss": 0.1729,
+      "step": 539
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002948988970588235,
+      "loss": 0.1696,
+      "step": 540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002947610294117647,
+      "loss": 0.1768,
+      "step": 541
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029462316176470584,
+      "loss": 0.2075,
+      "step": 542
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000294485294117647,
+      "loss": 0.2122,
+      "step": 543
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002943474264705882,
+      "loss": 0.2465,
+      "step": 544
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029420955882352936,
+      "loss": 0.1964,
+      "step": 545
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002940716911764706,
+      "loss": 0.197,
+      "step": 546
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029393382352941173,
+      "loss": 0.2031,
+      "step": 547
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029379595588235294,
+      "loss": 0.1767,
+      "step": 548
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002936580882352941,
+      "loss": 0.2136,
+      "step": 549
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029352022058823525,
+      "loss": 0.2455,
+      "step": 550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002933823529411764,
+      "loss": 0.229,
+      "step": 551
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002932444852941176,
+      "loss": 0.1992,
+      "step": 552
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029310661764705883,
+      "loss": 0.1689,
+      "step": 553
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029296875,
+      "loss": 0.1756,
+      "step": 554
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029283088235294115,
+      "loss": 0.21,
+      "step": 555
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002926930147058823,
+      "loss": 0.2413,
+      "step": 556
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002925551470588235,
+      "loss": 0.2557,
+      "step": 557
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029241727941176467,
+      "loss": 0.1795,
+      "step": 558
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002922794117647058,
+      "loss": 0.1704,
+      "step": 559
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029214154411764704,
+      "loss": 0.2196,
+      "step": 560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029200367647058825,
+      "loss": 0.2146,
+      "step": 561
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002918658088235294,
+      "loss": 0.2345,
+      "step": 562
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029172794117647056,
+      "loss": 0.3192,
+      "step": 563
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002915900735294117,
+      "loss": 0.1939,
+      "step": 564
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029145220588235293,
+      "loss": 0.1922,
+      "step": 565
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002913143382352941,
+      "loss": 0.17,
+      "step": 566
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029117647058823524,
+      "loss": 0.1792,
+      "step": 567
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029103860294117645,
+      "loss": 0.1938,
+      "step": 568
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029090073529411766,
+      "loss": 0.2737,
+      "step": 569
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002907628676470588,
+      "loss": 0.1865,
+      "step": 570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000290625,
+      "loss": 0.1764,
+      "step": 571
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029048713235294113,
+      "loss": 0.1853,
+      "step": 572
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002903492647058823,
+      "loss": 0.1883,
+      "step": 573
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002902113970588235,
+      "loss": 0.2234,
+      "step": 574
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029007352941176466,
+      "loss": 0.2207,
+      "step": 575
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028993566176470587,
+      "loss": 0.2421,
+      "step": 576
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000289797794117647,
+      "loss": 0.1486,
+      "step": 577
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028965992647058823,
+      "loss": 0.1737,
+      "step": 578
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002895220588235294,
+      "loss": 0.2059,
+      "step": 579
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028938419117647055,
+      "loss": 0.1899,
+      "step": 580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002892463235294117,
+      "loss": 0.222,
+      "step": 581
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002891084558823529,
+      "loss": 0.1824,
+      "step": 582
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028897058823529407,
+      "loss": 0.1679,
+      "step": 583
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002888327205882353,
+      "loss": 0.1956,
+      "step": 584
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028869485294117644,
+      "loss": 0.158,
+      "step": 585
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028855698529411765,
+      "loss": 0.1591,
+      "step": 586
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002884191176470588,
+      "loss": 0.2049,
+      "step": 587
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028828124999999996,
+      "loss": 0.3607,
+      "step": 588
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002881433823529411,
+      "loss": 0.1694,
+      "step": 589
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028800551470588233,
+      "loss": 0.195,
+      "step": 590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002878676470588235,
+      "loss": 0.1648,
+      "step": 591
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002877297794117647,
+      "loss": 0.1933,
+      "step": 592
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028759191176470585,
+      "loss": 0.1617,
+      "step": 593
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000287454044117647,
+      "loss": 0.2915,
+      "step": 594
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002873161764705882,
+      "loss": 0.1973,
+      "step": 595
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002871783088235294,
+      "loss": 0.1943,
+      "step": 596
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028704044117647054,
+      "loss": 0.183,
+      "step": 597
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028690257352941175,
+      "loss": 0.1919,
+      "step": 598
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028676470588235296,
+      "loss": 0.2166,
+      "step": 599
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002866268382352941,
+      "loss": 0.2459,
+      "step": 600
+    },
+    {
+      "epoch": 0.67,
+      "eval_loss": 0.18583357334136963,
+      "eval_runtime": 35.9923,
+      "eval_samples_per_second": 73.405,
+      "eval_steps_per_second": 2.306,
+      "eval_wer": 0.15543435144783815,
+      "step": 600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028648897058823527,
+      "loss": 0.1762,
+      "step": 601
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002863511029411764,
+      "loss": 0.1583,
+      "step": 602
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028621323529411764,
+      "loss": 0.1616,
+      "step": 603
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002860753676470588,
+      "loss": 0.1766,
+      "step": 604
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028593749999999995,
+      "loss": 0.1782,
+      "step": 605
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028579963235294116,
+      "loss": 0.1988,
+      "step": 606
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002856617647058823,
+      "loss": 0.2673,
+      "step": 607
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028552389705882353,
+      "loss": 0.1573,
+      "step": 608
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002853860294117647,
+      "loss": 0.1808,
+      "step": 609
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028524816176470584,
+      "loss": 0.2238,
+      "step": 610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000285110294117647,
+      "loss": 0.196,
+      "step": 611
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002849724264705882,
+      "loss": 0.1743,
+      "step": 612
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028483455882352937,
+      "loss": 0.2697,
+      "step": 613
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002846966911764706,
+      "loss": 0.15,
+      "step": 614
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028455882352941173,
+      "loss": 0.1736,
+      "step": 615
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028442095588235294,
+      "loss": 0.1847,
+      "step": 616
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002842830882352941,
+      "loss": 0.153,
+      "step": 617
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028414522058823526,
+      "loss": 0.1525,
+      "step": 618
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002840073529411764,
+      "loss": 0.2648,
+      "step": 619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002838694852941176,
+      "loss": 0.1795,
+      "step": 620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002837316176470588,
+      "loss": 0.1398,
+      "step": 621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028359375,
+      "loss": 0.1629,
+      "step": 622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028345588235294115,
+      "loss": 0.1709,
+      "step": 623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002833180147058823,
+      "loss": 0.1586,
+      "step": 624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002831801470588235,
+      "loss": 0.2224,
+      "step": 625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028304227941176467,
+      "loss": 0.1939,
+      "step": 626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028290441176470583,
+      "loss": 0.1702,
+      "step": 627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028276654411764704,
+      "loss": 0.1956,
+      "step": 628
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002826286764705882,
+      "loss": 0.1787,
+      "step": 629
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002824908088235294,
+      "loss": 0.1459,
+      "step": 630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028235294117647056,
+      "loss": 0.1886,
+      "step": 631
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002822150735294117,
+      "loss": 0.1658,
+      "step": 632
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028207720588235293,
+      "loss": 0.1539,
+      "step": 633
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002819393382352941,
+      "loss": 0.1701,
+      "step": 634
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028180147058823524,
+      "loss": 0.1592,
+      "step": 635
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028166360294117645,
+      "loss": 0.1738,
+      "step": 636
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002815257352941176,
+      "loss": 0.2302,
+      "step": 637
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002813878676470588,
+      "loss": 0.2331,
+      "step": 638
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028125,
+      "loss": 0.1587,
+      "step": 639
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028111213235294114,
+      "loss": 0.1322,
+      "step": 640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002809742647058823,
+      "loss": 0.1559,
+      "step": 641
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002808363970588235,
+      "loss": 0.1665,
+      "step": 642
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028069852941176466,
+      "loss": 0.163,
+      "step": 643
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028056066176470587,
+      "loss": 0.2965,
+      "step": 644
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000280422794117647,
+      "loss": 0.1377,
+      "step": 645
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028028492647058824,
+      "loss": 0.1458,
+      "step": 646
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002801470588235294,
+      "loss": 0.149,
+      "step": 647
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028000919117647055,
+      "loss": 0.1406,
+      "step": 648
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002798713235294117,
+      "loss": 0.1939,
+      "step": 649
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002797334558823529,
+      "loss": 0.1819,
+      "step": 650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002795955882352941,
+      "loss": 0.1508,
+      "step": 651
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002794577205882353,
+      "loss": 0.1726,
+      "step": 652
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00027931985294117644,
+      "loss": 0.1846,
+      "step": 653
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00027918198529411765,
+      "loss": 0.1135,
+      "step": 654
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002790441176470588,
+      "loss": 0.1706,
+      "step": 655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00027890624999999997,
+      "loss": 0.1844,
+      "step": 656
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002787683823529411,
+      "loss": 0.233,
+      "step": 657
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00027863051470588233,
+      "loss": 0.1419,
+      "step": 658
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002784926470588235,
+      "loss": 0.1739,
+      "step": 659
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002783547794117647,
+      "loss": 0.1825,
+      "step": 660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00027821691176470586,
+      "loss": 0.1892,
+      "step": 661
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000278079044117647,
+      "loss": 0.2514,
+      "step": 662
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002779411764705882,
+      "loss": 0.2101,
+      "step": 663
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002778033088235294,
+      "loss": 0.1327,
+      "step": 664
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00027766544117647054,
+      "loss": 0.156,
+      "step": 665
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00027752757352941175,
+      "loss": 0.1522,
+      "step": 666
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002773897058823529,
+      "loss": 0.1527,
+      "step": 667
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002772518382352941,
+      "loss": 0.195,
+      "step": 668
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00027711397058823527,
+      "loss": 0.2168,
+      "step": 669
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00027697610294117643,
+      "loss": 0.1728,
+      "step": 670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00027683823529411764,
+      "loss": 0.1518,
+      "step": 671
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002767003676470588,
+      "loss": 0.1735,
+      "step": 672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00027656249999999995,
+      "loss": 0.1766,
+      "step": 673
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00027642463235294116,
+      "loss": 0.1268,
+      "step": 674
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002762867647058823,
+      "loss": 0.2667,
+      "step": 675
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00027614889705882353,
+      "loss": 0.1677,
+      "step": 676
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002760110294117647,
+      "loss": 0.1459,
+      "step": 677
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00027587316176470584,
+      "loss": 0.1599,
+      "step": 678
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000275735294117647,
+      "loss": 0.2014,
+      "step": 679
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002755974264705882,
+      "loss": 0.1594,
+      "step": 680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00027545955882352937,
+      "loss": 0.1888,
+      "step": 681
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002753216911764706,
+      "loss": 0.2057,
+      "step": 682
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00027518382352941174,
+      "loss": 0.1609,
+      "step": 683
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00027504595588235295,
+      "loss": 0.1457,
+      "step": 684
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002749080882352941,
+      "loss": 0.1862,
+      "step": 685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00027477022058823526,
+      "loss": 0.1366,
+      "step": 686
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002746323529411764,
+      "loss": 0.1416,
+      "step": 687
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002744944852941176,
+      "loss": 0.2726,
+      "step": 688
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002743566176470588,
+      "loss": 0.1536,
+      "step": 689
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00027421875,
+      "loss": 0.1957,
+      "step": 690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00027408088235294115,
+      "loss": 0.1367,
+      "step": 691
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002739430147058823,
+      "loss": 0.1445,
+      "step": 692
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002738051470588235,
+      "loss": 0.162,
+      "step": 693
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002736672794117647,
+      "loss": 0.1934,
+      "step": 694
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00027352941176470583,
+      "loss": 0.1506,
+      "step": 695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00027339154411764704,
+      "loss": 0.1364,
+      "step": 696
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002732536764705882,
+      "loss": 0.1283,
+      "step": 697
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002731158088235294,
+      "loss": 0.151,
+      "step": 698
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00027297794117647057,
+      "loss": 0.178,
+      "step": 699
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002728400735294117,
+      "loss": 0.2993,
+      "step": 700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00027270220588235293,
+      "loss": 0.1452,
+      "step": 701
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002725643382352941,
+      "loss": 0.1516,
+      "step": 702
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00027242647058823525,
+      "loss": 0.139,
+      "step": 703
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00027228860294117646,
+      "loss": 0.1428,
+      "step": 704
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002721507352941176,
+      "loss": 0.1592,
+      "step": 705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002720128676470588,
+      "loss": 0.2154,
+      "step": 706
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000271875,
+      "loss": 0.1581,
+      "step": 707
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00027173713235294114,
+      "loss": 0.146,
+      "step": 708
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002715992647058823,
+      "loss": 0.1299,
+      "step": 709
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002714613970588235,
+      "loss": 0.124,
+      "step": 710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00027132352941176466,
+      "loss": 0.1376,
+      "step": 711
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00027118566176470587,
+      "loss": 0.1519,
+      "step": 712
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00027104779411764703,
+      "loss": 0.2034,
+      "step": 713
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00027090992647058824,
+      "loss": 0.161,
+      "step": 714
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002707720588235294,
+      "loss": 0.1658,
+      "step": 715
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00027063419117647055,
+      "loss": 0.1687,
+      "step": 716
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002704963235294117,
+      "loss": 0.1541,
+      "step": 717
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002703584558823529,
+      "loss": 0.1802,
+      "step": 718
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002702205882352941,
+      "loss": 0.2373,
+      "step": 719
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002700827205882353,
+      "loss": 0.1298,
+      "step": 720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00026994485294117644,
+      "loss": 0.1676,
+      "step": 721
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00026980698529411766,
+      "loss": 0.1527,
+      "step": 722
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002696691176470588,
+      "loss": 0.1488,
+      "step": 723
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00026953124999999997,
+      "loss": 0.158,
+      "step": 724
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002693933823529411,
+      "loss": 0.2039,
+      "step": 725
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00026925551470588234,
+      "loss": 0.1317,
+      "step": 726
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002691176470588235,
+      "loss": 0.1403,
+      "step": 727
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002689797794117647,
+      "loss": 0.1216,
+      "step": 728
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00026884191176470586,
+      "loss": 0.1662,
+      "step": 729
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000268704044117647,
+      "loss": 0.1588,
+      "step": 730
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002685661764705882,
+      "loss": 0.1779,
+      "step": 731
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002684283088235294,
+      "loss": 0.1637,
+      "step": 732
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00026829044117647054,
+      "loss": 0.1457,
+      "step": 733
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00026815257352941175,
+      "loss": 0.1491,
+      "step": 734
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002680147058823529,
+      "loss": 0.1595,
+      "step": 735
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002678768382352941,
+      "loss": 0.1442,
+      "step": 736
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002677389705882353,
+      "loss": 0.1954,
+      "step": 737
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00026760110294117643,
+      "loss": 0.2041,
+      "step": 738
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00026746323529411764,
+      "loss": 0.1552,
+      "step": 739
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002673253676470588,
+      "loss": 0.1593,
+      "step": 740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00026718749999999996,
+      "loss": 0.1676,
+      "step": 741
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00026704963235294117,
+      "loss": 0.14,
+      "step": 742
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002669117647058823,
+      "loss": 0.1798,
+      "step": 743
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00026677389705882353,
+      "loss": 0.2534,
+      "step": 744
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002666360294117647,
+      "loss": 0.142,
+      "step": 745
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00026649816176470585,
+      "loss": 0.1586,
+      "step": 746
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000266360294117647,
+      "loss": 0.1541,
+      "step": 747
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002662224264705882,
+      "loss": 0.1569,
+      "step": 748
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00026608455882352937,
+      "loss": 0.135,
+      "step": 749
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002659466911764706,
+      "loss": 0.1994,
+      "step": 750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00026580882352941174,
+      "loss": 0.142,
+      "step": 751
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00026567095588235295,
+      "loss": 0.1368,
+      "step": 752
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002655330882352941,
+      "loss": 0.1216,
+      "step": 753
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00026539522058823526,
+      "loss": 0.1408,
+      "step": 754
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002652573529411764,
+      "loss": 0.1469,
+      "step": 755
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00026511948529411763,
+      "loss": 0.1254,
+      "step": 756
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002649816176470588,
+      "loss": 0.1363,
+      "step": 757
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00026484375,
+      "loss": 0.1748,
+      "step": 758
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00026470588235294115,
+      "loss": 0.1231,
+      "step": 759
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002645680147058823,
+      "loss": 0.1225,
+      "step": 760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002644301470588235,
+      "loss": 0.1865,
+      "step": 761
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002642922794117647,
+      "loss": 0.1555,
+      "step": 762
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00026415441176470583,
+      "loss": 0.2018,
+      "step": 763
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00026401654411764704,
+      "loss": 0.1417,
+      "step": 764
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002638786764705882,
+      "loss": 0.1309,
+      "step": 765
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002637408088235294,
+      "loss": 0.152,
+      "step": 766
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00026360294117647057,
+      "loss": 0.1476,
+      "step": 767
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002634650735294117,
+      "loss": 0.1876,
+      "step": 768
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00026332720588235294,
+      "loss": 0.2062,
+      "step": 769
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002631893382352941,
+      "loss": 0.1248,
+      "step": 770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00026305147058823525,
+      "loss": 0.1462,
+      "step": 771
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00026291360294117646,
+      "loss": 0.1194,
+      "step": 772
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002627757352941176,
+      "loss": 0.1114,
+      "step": 773
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00026263786764705883,
+      "loss": 0.1781,
+      "step": 774
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002625,
+      "loss": 0.3114,
+      "step": 775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00026236213235294114,
+      "loss": 0.1437,
+      "step": 776
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002622242647058823,
+      "loss": 0.122,
+      "step": 777
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002620863970588235,
+      "loss": 0.1261,
+      "step": 778
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00026194852941176466,
+      "loss": 0.156,
+      "step": 779
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002618106617647059,
+      "loss": 0.1239,
+      "step": 780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00026167279411764703,
+      "loss": 0.2116,
+      "step": 781
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00026153492647058824,
+      "loss": 0.1504,
+      "step": 782
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002613970588235294,
+      "loss": 0.1207,
+      "step": 783
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00026125919117647056,
+      "loss": 0.1394,
+      "step": 784
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002611213235294117,
+      "loss": 0.1452,
+      "step": 785
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002609834558823529,
+      "loss": 0.1263,
+      "step": 786
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002608455882352941,
+      "loss": 0.1701,
+      "step": 787
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002607077205882353,
+      "loss": 0.2147,
+      "step": 788
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00026056985294117645,
+      "loss": 0.1588,
+      "step": 789
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00026043198529411766,
+      "loss": 0.1468,
+      "step": 790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002602941176470588,
+      "loss": 0.1124,
+      "step": 791
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00026015624999999997,
+      "loss": 0.1174,
+      "step": 792
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00026001838235294113,
+      "loss": 0.1305,
+      "step": 793
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00025988051470588234,
+      "loss": 0.1958,
+      "step": 794
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002597426470588235,
+      "loss": 0.1266,
+      "step": 795
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002596047794117647,
+      "loss": 0.1246,
+      "step": 796
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00025946691176470586,
+      "loss": 0.1394,
+      "step": 797
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000259329044117647,
+      "loss": 0.1418,
+      "step": 798
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00025919117647058823,
+      "loss": 0.151,
+      "step": 799
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002590533088235294,
+      "loss": 0.249,
+      "step": 800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00025891544117647054,
+      "loss": 0.1075,
+      "step": 801
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00025877757352941175,
+      "loss": 0.1353,
+      "step": 802
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002586397058823529,
+      "loss": 0.174,
+      "step": 803
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002585018382352941,
+      "loss": 0.1401,
+      "step": 804
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002583639705882353,
+      "loss": 0.0974,
+      "step": 805
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00025822610294117643,
+      "loss": 0.1541,
+      "step": 806
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00025808823529411764,
+      "loss": 0.1884,
+      "step": 807
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002579503676470588,
+      "loss": 0.1473,
+      "step": 808
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00025781249999999996,
+      "loss": 0.1211,
+      "step": 809
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00025767463235294117,
+      "loss": 0.1548,
+      "step": 810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002575367647058823,
+      "loss": 0.1449,
+      "step": 811
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00025739889705882354,
+      "loss": 0.2047,
+      "step": 812
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002572610294117647,
+      "loss": 0.1809,
+      "step": 813
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00025712316176470585,
+      "loss": 0.127,
+      "step": 814
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000256985294117647,
+      "loss": 0.1283,
+      "step": 815
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002568474264705882,
+      "loss": 0.152,
+      "step": 816
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002567095588235294,
+      "loss": 0.1239,
+      "step": 817
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002565716911764706,
+      "loss": 0.1234,
+      "step": 818
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00025643382352941174,
+      "loss": 0.224,
+      "step": 819
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00025629595588235295,
+      "loss": 0.125,
+      "step": 820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002561580882352941,
+      "loss": 0.1129,
+      "step": 821
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00025602022058823526,
+      "loss": 0.1195,
+      "step": 822
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002558823529411764,
+      "loss": 0.0991,
+      "step": 823
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00025574448529411763,
+      "loss": 0.157,
+      "step": 824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002556066176470588,
+      "loss": 0.1458,
+      "step": 825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00025546875,
+      "loss": 0.1522,
+      "step": 826
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00025533088235294116,
+      "loss": 0.1311,
+      "step": 827
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002551930147058823,
+      "loss": 0.1162,
+      "step": 828
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002550551470588235,
+      "loss": 0.1573,
+      "step": 829
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002549172794117647,
+      "loss": 0.1246,
+      "step": 830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00025477941176470584,
+      "loss": 0.1802,
+      "step": 831
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00025464154411764705,
+      "loss": 0.1993,
+      "step": 832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002545036764705882,
+      "loss": 0.0957,
+      "step": 833
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002543658088235294,
+      "loss": 0.0997,
+      "step": 834
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00025422794117647057,
+      "loss": 0.1466,
+      "step": 835
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00025409007352941173,
+      "loss": 0.1217,
+      "step": 836
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00025395220588235294,
+      "loss": 0.1679,
+      "step": 837
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002538143382352941,
+      "loss": 0.2743,
+      "step": 838
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00025367647058823525,
+      "loss": 0.1116,
+      "step": 839
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00025353860294117646,
+      "loss": 0.1334,
+      "step": 840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002534007352941176,
+      "loss": 0.1112,
+      "step": 841
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00025326286764705883,
+      "loss": 0.1343,
+      "step": 842
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000253125,
+      "loss": 0.1349,
+      "step": 843
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00025298713235294114,
+      "loss": 0.1779,
+      "step": 844
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002528492647058823,
+      "loss": 0.1207,
+      "step": 845
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002527113970588235,
+      "loss": 0.1082,
+      "step": 846
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00025257352941176467,
+      "loss": 0.1359,
+      "step": 847
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002524356617647059,
+      "loss": 0.1007,
+      "step": 848
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00025229779411764703,
+      "loss": 0.1108,
+      "step": 849
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00025215992647058824,
+      "loss": 0.1481,
+      "step": 850
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002520220588235294,
+      "loss": 0.1172,
+      "step": 851
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00025188419117647056,
+      "loss": 0.1166,
+      "step": 852
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002517463235294117,
+      "loss": 0.1251,
+      "step": 853
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002516084558823529,
+      "loss": 0.1442,
+      "step": 854
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002514705882352941,
+      "loss": 0.1689,
+      "step": 855
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002513327205882353,
+      "loss": 0.1149,
+      "step": 856
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00025119485294117645,
+      "loss": 0.126,
+      "step": 857
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00025105698529411766,
+      "loss": 0.0965,
+      "step": 858
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002509191176470588,
+      "loss": 0.1364,
+      "step": 859
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00025078125,
+      "loss": 0.1111,
+      "step": 860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00025064338235294113,
+      "loss": 0.1205,
+      "step": 861
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00025050551470588234,
+      "loss": 0.16,
+      "step": 862
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002503676470588235,
+      "loss": 0.2185,
+      "step": 863
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002502297794117647,
+      "loss": 0.1302,
+      "step": 864
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00025009191176470586,
+      "loss": 0.1134,
+      "step": 865
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000249954044117647,
+      "loss": 0.1052,
+      "step": 866
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00024981617647058823,
+      "loss": 0.1228,
+      "step": 867
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002496783088235294,
+      "loss": 0.1408,
+      "step": 868
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00024954044117647055,
+      "loss": 0.2025,
+      "step": 869
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00024940257352941176,
+      "loss": 0.1254,
+      "step": 870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002492647058823529,
+      "loss": 0.1102,
+      "step": 871
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002491268382352941,
+      "loss": 0.1203,
+      "step": 872
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002489889705882353,
+      "loss": 0.1333,
+      "step": 873
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00024885110294117644,
+      "loss": 0.1404,
+      "step": 874
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00024871323529411765,
+      "loss": 0.2056,
+      "step": 875
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002485753676470588,
+      "loss": 0.1262,
+      "step": 876
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00024843749999999996,
+      "loss": 0.1243,
+      "step": 877
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00024829963235294117,
+      "loss": 0.1454,
+      "step": 878
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00024816176470588233,
+      "loss": 0.1232,
+      "step": 879
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00024802389705882354,
+      "loss": 0.16,
+      "step": 880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002478860294117647,
+      "loss": 0.1781,
+      "step": 881
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00024774816176470585,
+      "loss": 0.1742,
+      "step": 882
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000247610294117647,
+      "loss": 0.1204,
+      "step": 883
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002474724264705882,
+      "loss": 0.1319,
+      "step": 884
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002473345588235294,
+      "loss": 0.143,
+      "step": 885
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002471966911764706,
+      "loss": 0.1646,
+      "step": 886
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00024705882352941174,
+      "loss": 0.1212,
+      "step": 887
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00024692095588235295,
+      "loss": 0.1479,
+      "step": 888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002467830882352941,
+      "loss": 0.135,
+      "step": 889
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00024664522058823527,
+      "loss": 0.1116,
+      "step": 890
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002465073529411764,
+      "loss": 0.1272,
+      "step": 891
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00024636948529411763,
+      "loss": 0.1766,
+      "step": 892
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002462316176470588,
+      "loss": 0.1227,
+      "step": 893
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00024609375,
+      "loss": 0.0832,
+      "step": 894
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00024595588235294116,
+      "loss": 0.0935,
+      "step": 895
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002458180147058823,
+      "loss": 0.1131,
+      "step": 896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002456801470588235,
+      "loss": 0.0916,
+      "step": 897
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002455422794117647,
+      "loss": 0.1908,
+      "step": 898
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00024540441176470584,
+      "loss": 0.1284,
+      "step": 899
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00024526654411764705,
+      "loss": 0.1114,
+      "step": 900
+    },
+    {
+      "epoch": 1.01,
+      "eval_loss": 0.1378510743379593,
+      "eval_runtime": 35.8335,
+      "eval_samples_per_second": 73.73,
+      "eval_steps_per_second": 2.316,
+      "eval_wer": 0.11911939706465688,
+      "step": 900
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002451286764705882,
+      "loss": 0.1082,
+      "step": 901
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002449908088235294,
+      "loss": 0.1348,
+      "step": 902
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002448529411764706,
+      "loss": 0.0989,
+      "step": 903
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00024471507352941173,
+      "loss": 0.1015,
+      "step": 904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00024457720588235294,
+      "loss": 0.2002,
+      "step": 905
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002444393382352941,
+      "loss": 0.1134,
+      "step": 906
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00024430147058823525,
+      "loss": 0.1157,
+      "step": 907
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00024416360294117646,
+      "loss": 0.1098,
+      "step": 908
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00024402573529411765,
+      "loss": 0.1084,
+      "step": 909
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002438878676470588,
+      "loss": 0.1169,
+      "step": 910
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00024375,
+      "loss": 0.1278,
+      "step": 911
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00024361213235294115,
+      "loss": 0.1143,
+      "step": 912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00024347426470588233,
+      "loss": 0.1167,
+      "step": 913
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002433363970588235,
+      "loss": 0.0906,
+      "step": 914
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002431985294117647,
+      "loss": 0.0832,
+      "step": 915
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00024306066176470585,
+      "loss": 0.0884,
+      "step": 916
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00024292279411764704,
+      "loss": 0.1295,
+      "step": 917
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00024278492647058822,
+      "loss": 0.1027,
+      "step": 918
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002426470588235294,
+      "loss": 0.0931,
+      "step": 919
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00024250919117647056,
+      "loss": 0.1146,
+      "step": 920
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00024237132352941174,
+      "loss": 0.1155,
+      "step": 921
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00024223345588235293,
+      "loss": 0.1021,
+      "step": 922
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002420955882352941,
+      "loss": 0.0917,
+      "step": 923
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00024195772058823527,
+      "loss": 0.1122,
+      "step": 924
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00024181985294117645,
+      "loss": 0.0895,
+      "step": 925
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00024168198529411764,
+      "loss": 0.111,
+      "step": 926
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00024154411764705882,
+      "loss": 0.0976,
+      "step": 927
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00024140624999999998,
+      "loss": 0.1051,
+      "step": 928
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00024126838235294116,
+      "loss": 0.1296,
+      "step": 929
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00024113051470588232,
+      "loss": 0.1638,
+      "step": 930
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00024099264705882353,
+      "loss": 0.118,
+      "step": 931
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00024085477941176468,
+      "loss": 0.1027,
+      "step": 932
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00024071691176470587,
+      "loss": 0.1165,
+      "step": 933
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00024057904411764702,
+      "loss": 0.1089,
+      "step": 934
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00024044117647058823,
+      "loss": 0.0935,
+      "step": 935
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002403033088235294,
+      "loss": 0.0949,
+      "step": 936
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00024016544117647057,
+      "loss": 0.0883,
+      "step": 937
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00024002757352941173,
+      "loss": 0.1008,
+      "step": 938
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00023988970588235294,
+      "loss": 0.0786,
+      "step": 939
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002397518382352941,
+      "loss": 0.0853,
+      "step": 940
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00023961397058823528,
+      "loss": 0.1418,
+      "step": 941
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00023947610294117644,
+      "loss": 0.1479,
+      "step": 942
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00023933823529411765,
+      "loss": 0.1123,
+      "step": 943
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002392003676470588,
+      "loss": 0.107,
+      "step": 944
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002390625,
+      "loss": 0.0995,
+      "step": 945
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00023892463235294115,
+      "loss": 0.1218,
+      "step": 946
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002387867647058823,
+      "loss": 0.0966,
+      "step": 947
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00023864889705882351,
+      "loss": 0.1372,
+      "step": 948
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002385110294117647,
+      "loss": 0.1278,
+      "step": 949
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00023837316176470585,
+      "loss": 0.1115,
+      "step": 950
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000238235294117647,
+      "loss": 0.1068,
+      "step": 951
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00023809742647058822,
+      "loss": 0.0939,
+      "step": 952
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002379595588235294,
+      "loss": 0.0712,
+      "step": 953
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00023782169117647056,
+      "loss": 0.11,
+      "step": 954
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00023768382352941175,
+      "loss": 0.1444,
+      "step": 955
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00023754595588235293,
+      "loss": 0.0918,
+      "step": 956
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002374080882352941,
+      "loss": 0.0971,
+      "step": 957
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00023727022058823527,
+      "loss": 0.1065,
+      "step": 958
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00023713235294117645,
+      "loss": 0.1246,
+      "step": 959
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00023699448529411764,
+      "loss": 0.0925,
+      "step": 960
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00023685661764705882,
+      "loss": 0.1155,
+      "step": 961
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00023671874999999998,
+      "loss": 0.0983,
+      "step": 962
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00023658088235294116,
+      "loss": 0.0902,
+      "step": 963
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00023644301470588232,
+      "loss": 0.1066,
+      "step": 964
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00023630514705882353,
+      "loss": 0.1248,
+      "step": 965
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00023616727941176468,
+      "loss": 0.1166,
+      "step": 966
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00023602941176470587,
+      "loss": 0.2354,
+      "step": 967
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00023589154411764703,
+      "loss": 0.1055,
+      "step": 968
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00023575367647058824,
+      "loss": 0.1094,
+      "step": 969
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002356158088235294,
+      "loss": 0.0922,
+      "step": 970
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00023547794117647058,
+      "loss": 0.1012,
+      "step": 971
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00023534007352941173,
+      "loss": 0.1134,
+      "step": 972
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00023520220588235294,
+      "loss": 0.1341,
+      "step": 973
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002350643382352941,
+      "loss": 0.0887,
+      "step": 974
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00023492647058823528,
+      "loss": 0.1274,
+      "step": 975
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00023478860294117644,
+      "loss": 0.1232,
+      "step": 976
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00023465073529411765,
+      "loss": 0.1295,
+      "step": 977
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002345128676470588,
+      "loss": 0.1086,
+      "step": 978
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000234375,
+      "loss": 0.0941,
+      "step": 979
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00023423713235294115,
+      "loss": 0.22,
+      "step": 980
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002340992647058823,
+      "loss": 0.084,
+      "step": 981
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00023396139705882352,
+      "loss": 0.1108,
+      "step": 982
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002338235294117647,
+      "loss": 0.1082,
+      "step": 983
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00023368566176470586,
+      "loss": 0.1203,
+      "step": 984
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000233547794117647,
+      "loss": 0.1048,
+      "step": 985
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00023340992647058822,
+      "loss": 0.1852,
+      "step": 986
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002332720588235294,
+      "loss": 0.1063,
+      "step": 987
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00023313419117647056,
+      "loss": 0.0914,
+      "step": 988
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00023299632352941172,
+      "loss": 0.0685,
+      "step": 989
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00023285845588235293,
+      "loss": 0.0906,
+      "step": 990
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00023272058823529411,
+      "loss": 0.1358,
+      "step": 991
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00023258272058823527,
+      "loss": 0.1864,
+      "step": 992
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00023244485294117643,
+      "loss": 0.1015,
+      "step": 993
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00023230698529411764,
+      "loss": 0.1151,
+      "step": 994
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00023216911764705882,
+      "loss": 0.1043,
+      "step": 995
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00023203124999999998,
+      "loss": 0.1037,
+      "step": 996
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00023189338235294113,
+      "loss": 0.0874,
+      "step": 997
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00023175551470588232,
+      "loss": 0.1432,
+      "step": 998
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00023161764705882353,
+      "loss": 0.1474,
+      "step": 999
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00023147977941176469,
+      "loss": 0.077,
+      "step": 1000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00023134191176470587,
+      "loss": 0.1062,
+      "step": 1001
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00023120404411764703,
+      "loss": 0.0884,
+      "step": 1002
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00023106617647058824,
+      "loss": 0.0919,
+      "step": 1003
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002309283088235294,
+      "loss": 0.1269,
+      "step": 1004
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00023079044117647058,
+      "loss": 0.1463,
+      "step": 1005
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00023065257352941173,
+      "loss": 0.0897,
+      "step": 1006
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00023051470588235294,
+      "loss": 0.0918,
+      "step": 1007
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002303768382352941,
+      "loss": 0.0831,
+      "step": 1008
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00023023897058823528,
+      "loss": 0.1185,
+      "step": 1009
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00023010110294117644,
+      "loss": 0.1393,
+      "step": 1010
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00022996323529411765,
+      "loss": 0.2278,
+      "step": 1011
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002298253676470588,
+      "loss": 0.0945,
+      "step": 1012
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002296875,
+      "loss": 0.0988,
+      "step": 1013
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00022954963235294115,
+      "loss": 0.1153,
+      "step": 1014
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002294117647058823,
+      "loss": 0.0794,
+      "step": 1015
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00022927389705882352,
+      "loss": 0.0983,
+      "step": 1016
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002291360294117647,
+      "loss": 0.2507,
+      "step": 1017
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00022899816176470586,
+      "loss": 0.096,
+      "step": 1018
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.000228860294117647,
+      "loss": 0.0992,
+      "step": 1019
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00022872242647058822,
+      "loss": 0.0745,
+      "step": 1020
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002285845588235294,
+      "loss": 0.0948,
+      "step": 1021
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00022844669117647056,
+      "loss": 0.1257,
+      "step": 1022
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00022830882352941172,
+      "loss": 0.151,
+      "step": 1023
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00022817095588235293,
+      "loss": 0.1252,
+      "step": 1024
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00022803308823529412,
+      "loss": 0.0961,
+      "step": 1025
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00022789522058823527,
+      "loss": 0.1024,
+      "step": 1026
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00022775735294117643,
+      "loss": 0.1111,
+      "step": 1027
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00022761948529411764,
+      "loss": 0.0914,
+      "step": 1028
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00022748161764705882,
+      "loss": 0.133,
+      "step": 1029
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00022734374999999998,
+      "loss": 0.1961,
+      "step": 1030
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00022720588235294114,
+      "loss": 0.1127,
+      "step": 1031
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00022706801470588232,
+      "loss": 0.1074,
+      "step": 1032
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00022693014705882353,
+      "loss": 0.1179,
+      "step": 1033
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002267922794117647,
+      "loss": 0.09,
+      "step": 1034
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00022665441176470584,
+      "loss": 0.0933,
+      "step": 1035
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00022651654411764703,
+      "loss": 0.1083,
+      "step": 1036
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00022637867647058824,
+      "loss": 0.1027,
+      "step": 1037
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002262408088235294,
+      "loss": 0.0893,
+      "step": 1038
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00022610294117647055,
+      "loss": 0.1181,
+      "step": 1039
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00022596507352941174,
+      "loss": 0.1165,
+      "step": 1040
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00022582720588235295,
+      "loss": 0.0991,
+      "step": 1041
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002256893382352941,
+      "loss": 0.1176,
+      "step": 1042
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00022555147058823526,
+      "loss": 0.1193,
+      "step": 1043
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00022541360294117644,
+      "loss": 0.0903,
+      "step": 1044
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00022527573529411765,
+      "loss": 0.0957,
+      "step": 1045
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002251378676470588,
+      "loss": 0.0781,
+      "step": 1046
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.000225,
+      "loss": 0.1108,
+      "step": 1047
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00022486213235294115,
+      "loss": 0.1336,
+      "step": 1048
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002247242647058823,
+      "loss": 0.0913,
+      "step": 1049
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00022458639705882352,
+      "loss": 0.085,
+      "step": 1050
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002244485294117647,
+      "loss": 0.1179,
+      "step": 1051
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00022431066176470586,
+      "loss": 0.0762,
+      "step": 1052
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00022417279411764701,
+      "loss": 0.1025,
+      "step": 1053
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00022403492647058823,
+      "loss": 0.1162,
+      "step": 1054
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002238970588235294,
+      "loss": 0.1214,
+      "step": 1055
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00022375919117647057,
+      "loss": 0.0748,
+      "step": 1056
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00022362132352941172,
+      "loss": 0.0749,
+      "step": 1057
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00022348345588235293,
+      "loss": 0.0796,
+      "step": 1058
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00022334558823529412,
+      "loss": 0.1033,
+      "step": 1059
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00022320772058823527,
+      "loss": 0.0745,
+      "step": 1060
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00022306985294117643,
+      "loss": 0.1633,
+      "step": 1061
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00022293198529411764,
+      "loss": 0.0865,
+      "step": 1062
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00022279411764705882,
+      "loss": 0.0914,
+      "step": 1063
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00022265624999999998,
+      "loss": 0.0956,
+      "step": 1064
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00022251838235294114,
+      "loss": 0.0993,
+      "step": 1065
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00022238051470588232,
+      "loss": 0.1004,
+      "step": 1066
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00022224264705882353,
+      "loss": 0.1739,
+      "step": 1067
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002221047794117647,
+      "loss": 0.0918,
+      "step": 1068
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00022196691176470585,
+      "loss": 0.1212,
+      "step": 1069
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00022182904411764703,
+      "loss": 0.082,
+      "step": 1070
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00022169117647058824,
+      "loss": 0.099,
+      "step": 1071
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002215533088235294,
+      "loss": 0.092,
+      "step": 1072
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00022141544117647055,
+      "loss": 0.1161,
+      "step": 1073
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00022127757352941174,
+      "loss": 0.1395,
+      "step": 1074
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00022113970588235295,
+      "loss": 0.0763,
+      "step": 1075
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002210018382352941,
+      "loss": 0.1008,
+      "step": 1076
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00022086397058823526,
+      "loss": 0.0943,
+      "step": 1077
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00022072610294117644,
+      "loss": 0.1207,
+      "step": 1078
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00022058823529411765,
+      "loss": 0.0948,
+      "step": 1079
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002204503676470588,
+      "loss": 0.1978,
+      "step": 1080
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00022031249999999997,
+      "loss": 0.0978,
+      "step": 1081
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00022017463235294115,
+      "loss": 0.1183,
+      "step": 1082
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002200367647058823,
+      "loss": 0.0769,
+      "step": 1083
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00021989889705882352,
+      "loss": 0.1169,
+      "step": 1084
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00021976102941176468,
+      "loss": 0.1047,
+      "step": 1085
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00021962316176470586,
+      "loss": 0.2041,
+      "step": 1086
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00021948529411764702,
+      "loss": 0.0795,
+      "step": 1087
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00021934742647058823,
+      "loss": 0.0811,
+      "step": 1088
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00021920955882352938,
+      "loss": 0.0741,
+      "step": 1089
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00021907169117647057,
+      "loss": 0.106,
+      "step": 1090
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00021893382352941172,
+      "loss": 0.0947,
+      "step": 1091
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00021879595588235293,
+      "loss": 0.1198,
+      "step": 1092
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002186580882352941,
+      "loss": 0.1007,
+      "step": 1093
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00021852022058823527,
+      "loss": 0.0817,
+      "step": 1094
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00021838235294117643,
+      "loss": 0.0797,
+      "step": 1095
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00021824448529411764,
+      "loss": 0.0898,
+      "step": 1096
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00021810661764705883,
+      "loss": 0.1379,
+      "step": 1097
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00021796874999999998,
+      "loss": 0.1244,
+      "step": 1098
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00021783088235294114,
+      "loss": 0.1087,
+      "step": 1099
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00021769301470588232,
+      "loss": 0.079,
+      "step": 1100
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00021755514705882353,
+      "loss": 0.0929,
+      "step": 1101
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002174172794117647,
+      "loss": 0.0994,
+      "step": 1102
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00021727941176470585,
+      "loss": 0.0923,
+      "step": 1103
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00021714154411764703,
+      "loss": 0.1235,
+      "step": 1104
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00021700367647058824,
+      "loss": 0.1637,
+      "step": 1105
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002168658088235294,
+      "loss": 0.0704,
+      "step": 1106
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00021672794117647055,
+      "loss": 0.0761,
+      "step": 1107
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00021659007352941174,
+      "loss": 0.1008,
+      "step": 1108
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00021645220588235295,
+      "loss": 0.0805,
+      "step": 1109
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002163143382352941,
+      "loss": 0.0882,
+      "step": 1110
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00021617647058823526,
+      "loss": 0.1483,
+      "step": 1111
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00021603860294117645,
+      "loss": 0.1002,
+      "step": 1112
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00021590073529411766,
+      "loss": 0.0702,
+      "step": 1113
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002157628676470588,
+      "loss": 0.0712,
+      "step": 1114
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00021562499999999997,
+      "loss": 0.1032,
+      "step": 1115
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00021548713235294115,
+      "loss": 0.0716,
+      "step": 1116
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002153492647058823,
+      "loss": 0.1724,
+      "step": 1117
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00021521139705882352,
+      "loss": 0.0879,
+      "step": 1118
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00021507352941176468,
+      "loss": 0.0915,
+      "step": 1119
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00021493566176470586,
+      "loss": 0.0756,
+      "step": 1120
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00021479779411764702,
+      "loss": 0.089,
+      "step": 1121
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00021465992647058823,
+      "loss": 0.0728,
+      "step": 1122
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00021452205882352938,
+      "loss": 0.1121,
+      "step": 1123
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00021438419117647057,
+      "loss": 0.1074,
+      "step": 1124
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00021424632352941172,
+      "loss": 0.0784,
+      "step": 1125
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00021410845588235294,
+      "loss": 0.0896,
+      "step": 1126
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002139705882352941,
+      "loss": 0.1044,
+      "step": 1127
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00021383272058823528,
+      "loss": 0.0907,
+      "step": 1128
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00021369485294117643,
+      "loss": 0.0711,
+      "step": 1129
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00021355698529411764,
+      "loss": 0.128,
+      "step": 1130
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002134191176470588,
+      "loss": 0.0928,
+      "step": 1131
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00021328124999999998,
+      "loss": 0.0976,
+      "step": 1132
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00021314338235294114,
+      "loss": 0.0866,
+      "step": 1133
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00021300551470588232,
+      "loss": 0.1107,
+      "step": 1134
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002128676470588235,
+      "loss": 0.0892,
+      "step": 1135
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002127297794117647,
+      "loss": 0.1354,
+      "step": 1136
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00021259191176470585,
+      "loss": 0.0736,
+      "step": 1137
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00021245404411764703,
+      "loss": 0.0989,
+      "step": 1138
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00021231617647058822,
+      "loss": 0.087,
+      "step": 1139
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002121783088235294,
+      "loss": 0.0968,
+      "step": 1140
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00021204044117647056,
+      "loss": 0.0948,
+      "step": 1141
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00021190257352941174,
+      "loss": 0.1476,
+      "step": 1142
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00021176470588235295,
+      "loss": 0.1112,
+      "step": 1143
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002116268382352941,
+      "loss": 0.1063,
+      "step": 1144
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00021148897058823526,
+      "loss": 0.0895,
+      "step": 1145
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00021135110294117645,
+      "loss": 0.074,
+      "step": 1146
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00021121323529411766,
+      "loss": 0.0805,
+      "step": 1147
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00021107536764705881,
+      "loss": 0.1153,
+      "step": 1148
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00021093749999999997,
+      "loss": 0.1255,
+      "step": 1149
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00021079963235294115,
+      "loss": 0.1005,
+      "step": 1150
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002106617647058823,
+      "loss": 0.0806,
+      "step": 1151
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00021052389705882352,
+      "loss": 0.088,
+      "step": 1152
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00021038602941176468,
+      "loss": 0.102,
+      "step": 1153
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00021024816176470586,
+      "loss": 0.0905,
+      "step": 1154
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00021011029411764702,
+      "loss": 0.1508,
+      "step": 1155
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00020997242647058823,
+      "loss": 0.0999,
+      "step": 1156
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00020983455882352939,
+      "loss": 0.0656,
+      "step": 1157
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00020969669117647057,
+      "loss": 0.0944,
+      "step": 1158
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00020955882352941173,
+      "loss": 0.0888,
+      "step": 1159
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00020942095588235294,
+      "loss": 0.0961,
+      "step": 1160
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002092830882352941,
+      "loss": 0.1175,
+      "step": 1161
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00020914522058823528,
+      "loss": 0.0701,
+      "step": 1162
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00020900735294117643,
+      "loss": 0.0922,
+      "step": 1163
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00020886948529411764,
+      "loss": 0.081,
+      "step": 1164
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002087316176470588,
+      "loss": 0.0818,
+      "step": 1165
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00020859374999999998,
+      "loss": 0.0935,
+      "step": 1166
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00020845588235294114,
+      "loss": 0.148,
+      "step": 1167
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00020831801470588232,
+      "loss": 0.0739,
+      "step": 1168
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002081801470588235,
+      "loss": 0.0985,
+      "step": 1169
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002080422794117647,
+      "loss": 0.0913,
+      "step": 1170
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00020790441176470585,
+      "loss": 0.0799,
+      "step": 1171
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00020776654411764703,
+      "loss": 0.1094,
+      "step": 1172
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00020762867647058822,
+      "loss": 0.1447,
+      "step": 1173
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002074908088235294,
+      "loss": 0.107,
+      "step": 1174
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00020735294117647056,
+      "loss": 0.0759,
+      "step": 1175
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00020721507352941174,
+      "loss": 0.0827,
+      "step": 1176
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00020707720588235292,
+      "loss": 0.0882,
+      "step": 1177
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002069393382352941,
+      "loss": 0.097,
+      "step": 1178
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00020680147058823526,
+      "loss": 0.1135,
+      "step": 1179
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00020666360294117645,
+      "loss": 0.1296,
+      "step": 1180
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00020652573529411763,
+      "loss": 0.0777,
+      "step": 1181
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00020638786764705882,
+      "loss": 0.0936,
+      "step": 1182
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00020624999999999997,
+      "loss": 0.0877,
+      "step": 1183
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00020611213235294116,
+      "loss": 0.0758,
+      "step": 1184
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002059742647058823,
+      "loss": 0.1035,
+      "step": 1185
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00020583639705882352,
+      "loss": 0.1266,
+      "step": 1186
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00020569852941176468,
+      "loss": 0.0991,
+      "step": 1187
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00020556066176470586,
+      "loss": 0.0697,
+      "step": 1188
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00020542279411764702,
+      "loss": 0.0876,
+      "step": 1189
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00020528492647058823,
+      "loss": 0.0925,
+      "step": 1190
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002051470588235294,
+      "loss": 0.1026,
+      "step": 1191
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00020500919117647057,
+      "loss": 0.1339,
+      "step": 1192
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00020487132352941173,
+      "loss": 0.0805,
+      "step": 1193
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00020473345588235294,
+      "loss": 0.0844,
+      "step": 1194
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002045955882352941,
+      "loss": 0.0955,
+      "step": 1195
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00020445772058823528,
+      "loss": 0.0826,
+      "step": 1196
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00020431985294117643,
+      "loss": 0.1218,
+      "step": 1197
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00020418198529411765,
+      "loss": 0.1047,
+      "step": 1198
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002040441176470588,
+      "loss": 0.0982,
+      "step": 1199
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00020390624999999999,
+      "loss": 0.0867,
+      "step": 1200
+    },
+    {
+      "epoch": 1.35,
+      "eval_loss": 0.11300347000360489,
+      "eval_runtime": 36.3345,
+      "eval_samples_per_second": 72.713,
+      "eval_steps_per_second": 2.284,
+      "eval_wer": 0.09613248710829037,
+      "step": 1200
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00020376838235294114,
+      "loss": 0.0802,
+      "step": 1201
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00020363051470588233,
+      "loss": 0.1153,
+      "step": 1202
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002034926470588235,
+      "loss": 0.0712,
+      "step": 1203
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002033547794117647,
+      "loss": 0.1112,
+      "step": 1204
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00020321691176470585,
+      "loss": 0.1461,
+      "step": 1205
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00020307904411764703,
+      "loss": 0.0704,
+      "step": 1206
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00020294117647058822,
+      "loss": 0.0874,
+      "step": 1207
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002028033088235294,
+      "loss": 0.0586,
+      "step": 1208
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00020266544117647056,
+      "loss": 0.0819,
+      "step": 1209
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00020252757352941174,
+      "loss": 0.1222,
+      "step": 1210
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00020238970588235293,
+      "loss": 0.1838,
+      "step": 1211
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002022518382352941,
+      "loss": 0.1049,
+      "step": 1212
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00020211397058823527,
+      "loss": 0.0968,
+      "step": 1213
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00020197610294117645,
+      "loss": 0.0698,
+      "step": 1214
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00020183823529411763,
+      "loss": 0.0945,
+      "step": 1215
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00020170036764705882,
+      "loss": 0.0888,
+      "step": 1216
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00020156249999999997,
+      "loss": 0.1131,
+      "step": 1217
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00020142463235294116,
+      "loss": 0.0798,
+      "step": 1218
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002012867647058823,
+      "loss": 0.0868,
+      "step": 1219
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00020114889705882352,
+      "loss": 0.0811,
+      "step": 1220
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00020101102941176468,
+      "loss": 0.0582,
+      "step": 1221
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00020087316176470586,
+      "loss": 0.0959,
+      "step": 1222
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00020073529411764702,
+      "loss": 0.1148,
+      "step": 1223
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00020059742647058823,
+      "loss": 0.128,
+      "step": 1224
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002004595588235294,
+      "loss": 0.08,
+      "step": 1225
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00020032169117647057,
+      "loss": 0.0649,
+      "step": 1226
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00020018382352941173,
+      "loss": 0.0878,
+      "step": 1227
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00020004595588235294,
+      "loss": 0.0901,
+      "step": 1228
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001999080882352941,
+      "loss": 0.0974,
+      "step": 1229
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00019977022058823528,
+      "loss": 0.1762,
+      "step": 1230
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00019963235294117644,
+      "loss": 0.0764,
+      "step": 1231
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00019949448529411765,
+      "loss": 0.0744,
+      "step": 1232
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001993566176470588,
+      "loss": 0.0859,
+      "step": 1233
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00019921875,
+      "loss": 0.0721,
+      "step": 1234
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00019908088235294114,
+      "loss": 0.0684,
+      "step": 1235
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00019894301470588233,
+      "loss": 0.1021,
+      "step": 1236
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001988051470588235,
+      "loss": 0.0604,
+      "step": 1237
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001986672794117647,
+      "loss": 0.0796,
+      "step": 1238
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00019852941176470585,
+      "loss": 0.0999,
+      "step": 1239
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00019839154411764704,
+      "loss": 0.108,
+      "step": 1240
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00019825367647058822,
+      "loss": 0.0941,
+      "step": 1241
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001981158088235294,
+      "loss": 0.1051,
+      "step": 1242
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00019797794117647056,
+      "loss": 0.1,
+      "step": 1243
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00019784007352941174,
+      "loss": 0.0731,
+      "step": 1244
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00019770220588235293,
+      "loss": 0.0786,
+      "step": 1245
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001975643382352941,
+      "loss": 0.087,
+      "step": 1246
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00019742647058823527,
+      "loss": 0.0917,
+      "step": 1247
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00019728860294117645,
+      "loss": 0.1095,
+      "step": 1248
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00019715073529411763,
+      "loss": 0.1395,
+      "step": 1249
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00019701286764705882,
+      "loss": 0.0846,
+      "step": 1250
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00019687499999999997,
+      "loss": 0.0963,
+      "step": 1251
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00019673713235294116,
+      "loss": 0.0936,
+      "step": 1252
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00019659926470588231,
+      "loss": 0.0926,
+      "step": 1253
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00019646139705882353,
+      "loss": 0.1177,
+      "step": 1254
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00019632352941176468,
+      "loss": 0.161,
+      "step": 1255
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00019618566176470587,
+      "loss": 0.0984,
+      "step": 1256
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00019604779411764702,
+      "loss": 0.1075,
+      "step": 1257
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00019590992647058823,
+      "loss": 0.0675,
+      "step": 1258
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001957720588235294,
+      "loss": 0.0835,
+      "step": 1259
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00019563419117647057,
+      "loss": 0.1029,
+      "step": 1260
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00019549632352941173,
+      "loss": 0.1434,
+      "step": 1261
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00019535845588235294,
+      "loss": 0.0649,
+      "step": 1262
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001952205882352941,
+      "loss": 0.0769,
+      "step": 1263
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00019508272058823528,
+      "loss": 0.1055,
+      "step": 1264
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00019494485294117644,
+      "loss": 0.086,
+      "step": 1265
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00019480698529411765,
+      "loss": 0.1006,
+      "step": 1266
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001946691176470588,
+      "loss": 0.2192,
+      "step": 1267
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00019453125,
+      "loss": 0.0961,
+      "step": 1268
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00019439338235294115,
+      "loss": 0.0949,
+      "step": 1269
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00019425551470588233,
+      "loss": 0.0816,
+      "step": 1270
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001941176470588235,
+      "loss": 0.0682,
+      "step": 1271
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001939797794117647,
+      "loss": 0.0796,
+      "step": 1272
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019384191176470585,
+      "loss": 0.1104,
+      "step": 1273
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019370404411764704,
+      "loss": 0.1439,
+      "step": 1274
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019356617647058822,
+      "loss": 0.0715,
+      "step": 1275
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001934283088235294,
+      "loss": 0.0796,
+      "step": 1276
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019329044117647056,
+      "loss": 0.0871,
+      "step": 1277
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019315257352941174,
+      "loss": 0.0726,
+      "step": 1278
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019301470588235293,
+      "loss": 0.1029,
+      "step": 1279
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001928768382352941,
+      "loss": 0.1369,
+      "step": 1280
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019273897058823527,
+      "loss": 0.0774,
+      "step": 1281
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019260110294117645,
+      "loss": 0.0853,
+      "step": 1282
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019246323529411764,
+      "loss": 0.0634,
+      "step": 1283
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019232536764705882,
+      "loss": 0.0789,
+      "step": 1284
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019218749999999998,
+      "loss": 0.0906,
+      "step": 1285
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019204963235294116,
+      "loss": 0.1213,
+      "step": 1286
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019191176470588232,
+      "loss": 0.0733,
+      "step": 1287
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019177389705882353,
+      "loss": 0.0785,
+      "step": 1288
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00019163602941176468,
+      "loss": 0.1042,
+      "step": 1289
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00019149816176470587,
+      "loss": 0.0805,
+      "step": 1290
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00019136029411764702,
+      "loss": 0.1104,
+      "step": 1291
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00019122242647058823,
+      "loss": 0.1634,
+      "step": 1292
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001910845588235294,
+      "loss": 0.1201,
+      "step": 1293
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00019094669117647057,
+      "loss": 0.1055,
+      "step": 1294
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00019080882352941173,
+      "loss": 0.0928,
+      "step": 1295
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00019067095588235294,
+      "loss": 0.0919,
+      "step": 1296
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001905330882352941,
+      "loss": 0.0881,
+      "step": 1297
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00019039522058823528,
+      "loss": 0.1116,
+      "step": 1298
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00019025735294117644,
+      "loss": 0.111,
+      "step": 1299
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00019011948529411765,
+      "loss": 0.0962,
+      "step": 1300
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001899816176470588,
+      "loss": 0.0832,
+      "step": 1301
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00018984375,
+      "loss": 0.0874,
+      "step": 1302
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00018970588235294115,
+      "loss": 0.1052,
+      "step": 1303
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00018956801470588233,
+      "loss": 0.112,
+      "step": 1304
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00018943014705882351,
+      "loss": 0.1191,
+      "step": 1305
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001892922794117647,
+      "loss": 0.0906,
+      "step": 1306
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00018915441176470585,
+      "loss": 0.0824,
+      "step": 1307
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00018901654411764704,
+      "loss": 0.0797,
+      "step": 1308
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00018887867647058822,
+      "loss": 0.0846,
+      "step": 1309
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001887408088235294,
+      "loss": 0.15,
+      "step": 1310
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00018860294117647056,
+      "loss": 0.1649,
+      "step": 1311
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00018846507352941175,
+      "loss": 0.0888,
+      "step": 1312
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00018832720588235293,
+      "loss": 0.0765,
+      "step": 1313
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001881893382352941,
+      "loss": 0.0851,
+      "step": 1314
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00018805147058823527,
+      "loss": 0.0774,
+      "step": 1315
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00018791360294117645,
+      "loss": 0.0855,
+      "step": 1316
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00018777573529411764,
+      "loss": 0.1352,
+      "step": 1317
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00018763786764705882,
+      "loss": 0.0939,
+      "step": 1318
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00018749999999999998,
+      "loss": 0.1195,
+      "step": 1319
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00018736213235294116,
+      "loss": 0.1295,
+      "step": 1320
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00018722426470588232,
+      "loss": 0.0648,
+      "step": 1321
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00018708639705882353,
+      "loss": 0.099,
+      "step": 1322
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00018694852941176468,
+      "loss": 0.1165,
+      "step": 1323
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00018681066176470587,
+      "loss": 0.1642,
+      "step": 1324
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00018667279411764702,
+      "loss": 0.0741,
+      "step": 1325
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00018653492647058824,
+      "loss": 0.0939,
+      "step": 1326
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001863970588235294,
+      "loss": 0.0718,
+      "step": 1327
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00018625919117647058,
+      "loss": 0.0887,
+      "step": 1328
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00018612132352941173,
+      "loss": 0.0676,
+      "step": 1329
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00018598345588235294,
+      "loss": 0.1411,
+      "step": 1330
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001858455882352941,
+      "loss": 0.0752,
+      "step": 1331
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00018570772058823528,
+      "loss": 0.0921,
+      "step": 1332
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00018556985294117644,
+      "loss": 0.0918,
+      "step": 1333
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00018543198529411765,
+      "loss": 0.0991,
+      "step": 1334
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001852941176470588,
+      "loss": 0.0882,
+      "step": 1335
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00018515625,
+      "loss": 0.1498,
+      "step": 1336
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00018501838235294115,
+      "loss": 0.0712,
+      "step": 1337
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00018488051470588233,
+      "loss": 0.0693,
+      "step": 1338
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00018474264705882351,
+      "loss": 0.1007,
+      "step": 1339
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001846047794117647,
+      "loss": 0.104,
+      "step": 1340
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00018446691176470586,
+      "loss": 0.0971,
+      "step": 1341
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00018432904411764704,
+      "loss": 0.2248,
+      "step": 1342
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00018419117647058822,
+      "loss": 0.0877,
+      "step": 1343
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001840533088235294,
+      "loss": 0.0796,
+      "step": 1344
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00018391544117647056,
+      "loss": 0.0848,
+      "step": 1345
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00018377757352941175,
+      "loss": 0.0724,
+      "step": 1346
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00018363970588235293,
+      "loss": 0.106,
+      "step": 1347
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00018350183823529411,
+      "loss": 0.1327,
+      "step": 1348
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00018336397058823527,
+      "loss": 0.0847,
+      "step": 1349
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00018322610294117645,
+      "loss": 0.0752,
+      "step": 1350
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00018308823529411764,
+      "loss": 0.0931,
+      "step": 1351
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00018295036764705882,
+      "loss": 0.0753,
+      "step": 1352
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00018281249999999998,
+      "loss": 0.1184,
+      "step": 1353
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00018267463235294116,
+      "loss": 0.0923,
+      "step": 1354
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00018253676470588232,
+      "loss": 0.1308,
+      "step": 1355
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00018239889705882353,
+      "loss": 0.0789,
+      "step": 1356
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00018226102941176469,
+      "loss": 0.0848,
+      "step": 1357
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00018212316176470587,
+      "loss": 0.1014,
+      "step": 1358
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00018198529411764703,
+      "loss": 0.076,
+      "step": 1359
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00018184742647058824,
+      "loss": 0.1127,
+      "step": 1360
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001817095588235294,
+      "loss": 0.1151,
+      "step": 1361
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00018157169117647058,
+      "loss": 0.0968,
+      "step": 1362
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00018143382352941173,
+      "loss": 0.071,
+      "step": 1363
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00018129595588235294,
+      "loss": 0.0749,
+      "step": 1364
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001811580882352941,
+      "loss": 0.0763,
+      "step": 1365
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00018102022058823528,
+      "loss": 0.0818,
+      "step": 1366
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00018088235294117644,
+      "loss": 0.1002,
+      "step": 1367
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00018074448529411765,
+      "loss": 0.085,
+      "step": 1368
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001806066176470588,
+      "loss": 0.0839,
+      "step": 1369
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00018046875,
+      "loss": 0.0995,
+      "step": 1370
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00018033088235294115,
+      "loss": 0.069,
+      "step": 1371
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00018019301470588233,
+      "loss": 0.0763,
+      "step": 1372
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00018005514705882352,
+      "loss": 0.1358,
+      "step": 1373
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001799172794117647,
+      "loss": 0.0669,
+      "step": 1374
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00017977941176470586,
+      "loss": 0.0811,
+      "step": 1375
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00017964154411764704,
+      "loss": 0.0933,
+      "step": 1376
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00017950367647058822,
+      "loss": 0.0976,
+      "step": 1377
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001793658088235294,
+      "loss": 0.0831,
+      "step": 1378
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00017922794117647056,
+      "loss": 0.1142,
+      "step": 1379
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00017909007352941175,
+      "loss": 0.1523,
+      "step": 1380
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00017895220588235293,
+      "loss": 0.0827,
+      "step": 1381
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00017881433823529412,
+      "loss": 0.0915,
+      "step": 1382
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00017867647058823527,
+      "loss": 0.0766,
+      "step": 1383
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00017853860294117646,
+      "loss": 0.0932,
+      "step": 1384
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00017840073529411764,
+      "loss": 0.082,
+      "step": 1385
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00017826286764705882,
+      "loss": 0.088,
+      "step": 1386
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00017812499999999998,
+      "loss": 0.0838,
+      "step": 1387
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00017798713235294116,
+      "loss": 0.0864,
+      "step": 1388
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00017784926470588232,
+      "loss": 0.0982,
+      "step": 1389
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00017771139705882353,
+      "loss": 0.0915,
+      "step": 1390
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001775735294117647,
+      "loss": 0.1082,
+      "step": 1391
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00017743566176470587,
+      "loss": 0.116,
+      "step": 1392
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00017729779411764703,
+      "loss": 0.0754,
+      "step": 1393
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00017715992647058824,
+      "loss": 0.0792,
+      "step": 1394
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001770220588235294,
+      "loss": 0.0606,
+      "step": 1395
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00017688419117647058,
+      "loss": 0.0833,
+      "step": 1396
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00017674632352941173,
+      "loss": 0.0568,
+      "step": 1397
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00017660845588235295,
+      "loss": 0.1158,
+      "step": 1398
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001764705882352941,
+      "loss": 0.0838,
+      "step": 1399
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00017633272058823529,
+      "loss": 0.0582,
+      "step": 1400
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00017619485294117644,
+      "loss": 0.0726,
+      "step": 1401
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00017605698529411765,
+      "loss": 0.0822,
+      "step": 1402
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001759191176470588,
+      "loss": 0.0842,
+      "step": 1403
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00017578125,
+      "loss": 0.0834,
+      "step": 1404
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00017564338235294115,
+      "loss": 0.0791,
+      "step": 1405
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00017550551470588233,
+      "loss": 0.0998,
+      "step": 1406
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00017536764705882352,
+      "loss": 0.1046,
+      "step": 1407
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001752297794117647,
+      "loss": 0.073,
+      "step": 1408
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00017509191176470586,
+      "loss": 0.0554,
+      "step": 1409
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00017495404411764704,
+      "loss": 0.0698,
+      "step": 1410
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00017481617647058823,
+      "loss": 0.1404,
+      "step": 1411
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001746783088235294,
+      "loss": 0.0909,
+      "step": 1412
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00017454044117647057,
+      "loss": 0.0857,
+      "step": 1413
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00017440257352941175,
+      "loss": 0.092,
+      "step": 1414
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00017426470588235293,
+      "loss": 0.0538,
+      "step": 1415
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00017412683823529412,
+      "loss": 0.0833,
+      "step": 1416
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00017398897058823527,
+      "loss": 0.1078,
+      "step": 1417
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00017385110294117646,
+      "loss": 0.1007,
+      "step": 1418
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00017371323529411764,
+      "loss": 0.085,
+      "step": 1419
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00017357536764705882,
+      "loss": 0.0928,
+      "step": 1420
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00017343749999999998,
+      "loss": 0.0683,
+      "step": 1421
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00017329963235294116,
+      "loss": 0.0799,
+      "step": 1422
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00017316176470588232,
+      "loss": 0.0581,
+      "step": 1423
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00017302389705882353,
+      "loss": 0.0973,
+      "step": 1424
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001728860294117647,
+      "loss": 0.0898,
+      "step": 1425
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00017274816176470587,
+      "loss": 0.0712,
+      "step": 1426
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00017261029411764703,
+      "loss": 0.0925,
+      "step": 1427
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00017247242647058824,
+      "loss": 0.0722,
+      "step": 1428
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001723345588235294,
+      "loss": 0.0674,
+      "step": 1429
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00017219669117647058,
+      "loss": 0.1442,
+      "step": 1430
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00017205882352941174,
+      "loss": 0.091,
+      "step": 1431
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00017192095588235295,
+      "loss": 0.092,
+      "step": 1432
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001717830882352941,
+      "loss": 0.0772,
+      "step": 1433
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001716452205882353,
+      "loss": 0.0814,
+      "step": 1434
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00017150735294117644,
+      "loss": 0.0863,
+      "step": 1435
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00017136948529411765,
+      "loss": 0.0993,
+      "step": 1436
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001712316176470588,
+      "loss": 0.071,
+      "step": 1437
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00017109375,
+      "loss": 0.0854,
+      "step": 1438
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00017095588235294115,
+      "loss": 0.0794,
+      "step": 1439
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00017081801470588234,
+      "loss": 0.0704,
+      "step": 1440
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00017068014705882352,
+      "loss": 0.0877,
+      "step": 1441
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001705422794117647,
+      "loss": 0.1029,
+      "step": 1442
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00017040441176470586,
+      "loss": 0.1082,
+      "step": 1443
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00017026654411764704,
+      "loss": 0.0706,
+      "step": 1444
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00017012867647058823,
+      "loss": 0.0947,
+      "step": 1445
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001699908088235294,
+      "loss": 0.084,
+      "step": 1446
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00016985294117647057,
+      "loss": 0.0813,
+      "step": 1447
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00016971507352941175,
+      "loss": 0.0968,
+      "step": 1448
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00016957720588235293,
+      "loss": 0.0849,
+      "step": 1449
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00016943933823529412,
+      "loss": 0.0742,
+      "step": 1450
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00016930147058823527,
+      "loss": 0.0538,
+      "step": 1451
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00016916360294117646,
+      "loss": 0.1062,
+      "step": 1452
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00016902573529411764,
+      "loss": 0.079,
+      "step": 1453
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00016888786764705883,
+      "loss": 0.0944,
+      "step": 1454
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00016874999999999998,
+      "loss": 0.1213,
+      "step": 1455
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00016861213235294117,
+      "loss": 0.0892,
+      "step": 1456
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00016847426470588232,
+      "loss": 0.0929,
+      "step": 1457
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00016833639705882353,
+      "loss": 0.1051,
+      "step": 1458
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001681985294117647,
+      "loss": 0.0705,
+      "step": 1459
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00016806066176470587,
+      "loss": 0.0892,
+      "step": 1460
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00016792279411764703,
+      "loss": 0.1394,
+      "step": 1461
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00016778492647058824,
+      "loss": 0.0761,
+      "step": 1462
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001676470588235294,
+      "loss": 0.0705,
+      "step": 1463
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00016750919117647058,
+      "loss": 0.0748,
+      "step": 1464
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00016737132352941174,
+      "loss": 0.079,
+      "step": 1465
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00016723345588235295,
+      "loss": 0.0876,
+      "step": 1466
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001670955882352941,
+      "loss": 0.0971,
+      "step": 1467
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001669577205882353,
+      "loss": 0.0749,
+      "step": 1468
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00016681985294117644,
+      "loss": 0.0554,
+      "step": 1469
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00016668198529411766,
+      "loss": 0.0712,
+      "step": 1470
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001665441176470588,
+      "loss": 0.0752,
+      "step": 1471
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00016640625,
+      "loss": 0.0783,
+      "step": 1472
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00016626838235294115,
+      "loss": 0.0919,
+      "step": 1473
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001661305147058823,
+      "loss": 0.1119,
+      "step": 1474
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00016599264705882352,
+      "loss": 0.0642,
+      "step": 1475
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001658547794117647,
+      "loss": 0.075,
+      "step": 1476
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016571691176470586,
+      "loss": 0.0732,
+      "step": 1477
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016557904411764704,
+      "loss": 0.0892,
+      "step": 1478
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016544117647058823,
+      "loss": 0.1434,
+      "step": 1479
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001653033088235294,
+      "loss": 0.1399,
+      "step": 1480
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016516544117647057,
+      "loss": 0.0765,
+      "step": 1481
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016502757352941175,
+      "loss": 0.0908,
+      "step": 1482
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016488970588235294,
+      "loss": 0.0643,
+      "step": 1483
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016475183823529412,
+      "loss": 0.0896,
+      "step": 1484
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016461397058823528,
+      "loss": 0.1142,
+      "step": 1485
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016447610294117646,
+      "loss": 0.1327,
+      "step": 1486
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016433823529411764,
+      "loss": 0.0756,
+      "step": 1487
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016420036764705883,
+      "loss": 0.0641,
+      "step": 1488
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016406249999999998,
+      "loss": 0.0833,
+      "step": 1489
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016392463235294117,
+      "loss": 0.0741,
+      "step": 1490
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016378676470588232,
+      "loss": 0.089,
+      "step": 1491
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016364889705882353,
+      "loss": 0.1794,
+      "step": 1492
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001635110294117647,
+      "loss": 0.0576,
+      "step": 1493
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016337316176470587,
+      "loss": 0.0914,
+      "step": 1494
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016323529411764703,
+      "loss": 0.0783,
+      "step": 1495
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016309742647058824,
+      "loss": 0.0835,
+      "step": 1496
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001629595588235294,
+      "loss": 0.0798,
+      "step": 1497
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016282169117647058,
+      "loss": 0.1071,
+      "step": 1498
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016268382352941174,
+      "loss": 0.1542,
+      "step": 1499
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016254595588235295,
+      "loss": 0.0698,
+      "step": 1500
+    },
+    {
+      "epoch": 1.68,
+      "eval_loss": 0.10318898409605026,
+      "eval_runtime": 36.1757,
+      "eval_samples_per_second": 73.032,
+      "eval_steps_per_second": 2.294,
+      "eval_wer": 0.08772312574375248,
+      "step": 1500
+    }
+  ],
+  "max_steps": 2676,
+  "num_train_epochs": 3,
+  "total_flos": 6.129184990879023e+18,
+  "trial_name": null,
+  "trial_params": null
+}